Citation
Linking Changes in Dynamic Cotton Canopy to Passive Microwave Remote Sensing

Material Information

Title:
Linking Changes in Dynamic Cotton Canopy to Passive Microwave Remote Sensing
Creator:
TIEN, KAI-JEN CALVIN ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Calibration ( jstor )
Cotton ( jstor )
Microwaves ( jstor )
Modeling ( jstor )
Radiometers ( jstor )
Remote sensing ( jstor )
Soil moisture ( jstor )
Soil temperature regimes ( jstor )
Soils ( jstor )
Vegetation canopies ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Kai-Jen Calvin Tien. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
12/31/2007
Resource Identifier:
659562924 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

LINKING CHANGES IN DYNAMIC CO TTON CANOPY TO PASSIVE MICROWAVE REMOTE SENSING By KAI-JEN CALVIN TIEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Kai-Jen Calvin Tien

PAGE 3

This document is dedicated to my wife, TzuYi, Hsu and my parents, Jyue-Min Tien and Chi-Shih, Wu.

PAGE 4

iv ACKNOWLEDGMENTS First and foremost, I would like to thank my parents, Jyue-Min Tien and Chi-Shih, Wu, for their encouragement and dedication th roughout the years of my graduate study in the United States. They have done everything imaginable to ensure my success. Acknowledgement also goes to my wife, TzuYi Hsu, for her infin ite love and support during the victories as well as defeats. Credit is due Dr. Jasmeet Judge, my me ntor, for her encouragement, advice, guidance, and endless patience. I would like to extend my gratitude to Dr. Wendy Graham and Clint Slatton of the University of Florida, Dr. Roger De Roo of the University of Michigan, and Dr. Wade Cr ow of the United States Department of Agriculture, for the ideas they provided as members of my supervisory committee. I would like to thank Mr. Larry Miller and Or lando Lanni for their technical support. I would also like to thank Tzu-yun Lin, Joaquin Casanova, and Mi-Young Jang, for their companionship during my field experiment. Thanks also go to the PSREU Research Coordinator, Mr. James Boyer, and his team for providing excellent management of the study fields. Last but not least, I w ould like to express my a ppreciation to the National Aeronautics and Space Administration Earth System Science Fellowship program, which provides the financial suppor t for my dissertation.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.............................................................................................................x ABSTRACT.....................................................................................................................xvi CHAPTERS 1 INTRODUCTION........................................................................................................1 Historical Development of Microwave Remote Sensing for Soil Moisture.................4 Dissertation Objectives.................................................................................................6 Dissertation Format......................................................................................................7 2 MICROWAVE WATER AND ENERGY BALANCE EXPERIMENTS...................9 The First Microwave Water and Energy Balance Experiment (MicroWEX-1)...........9 Sensors........................................................................................................................12 University of Florida C-band Micr owave Radiometer (UFCMR) System.........12 Micrometeorological Subsystems.......................................................................18 Eddy covariance system...............................................................................18 Net radiometer..............................................................................................20 Thermal infrared sensor...............................................................................22 Soil moisture and temperature probes..........................................................22 Soil heat flux plates......................................................................................25 Tipping buck raingauges..............................................................................26 Florida automated weat her network (FAWN)..............................................26 Soil Sampling..............................................................................................................29 Gravimetric Soil Moisture...................................................................................29 Soil Temperature.................................................................................................29 Vegetation Sampling..................................................................................................32 Canopy Height.....................................................................................................32 Leaf Area Index (LAI).........................................................................................32 Green and Dry Biomass.......................................................................................33 The Second Microwave Water and Energy Balance Experiment (MicroWEX-2).....34 The Third Microwave Water and Energy Balance Experiment (MicroWEX-3).......35

PAGE 6

vi 3 RADIOMETRIC CALIBRATION FOR C-BAND MICROWAVE RADIOMETERS........................................................................................................45 Introduction.................................................................................................................45 C-band Radiometers...................................................................................................47 Field Experiments.......................................................................................................48 Microwave Water and Energy Bala nce Experiments (MicroWEXs)..................48 The Tenth Radiobrightness and Energy Balance Experiment (REBEX-10).......48 Calibration Methodology............................................................................................49 External Calibration.............................................................................................49 Tipping-Curve Calibration..................................................................................50 Internal Calibration..............................................................................................51 Lake Emission Model (LEM)..............................................................................52 Results and Discussion...............................................................................................52 Conclusion..................................................................................................................59 4 MICROWAVE BRIGHTNESS MODEL FOR COTTON........................................60 Microwave Brightness Mode l for Cotton (MB-Cotton).............................................60 Bare Soil Component (TB,S,p)..............................................................................61 Vegetation Component (TB,C,p)............................................................................65 Summary.....................................................................................................................67 5 MODEL EVALUATION...........................................................................................69 Input Variables............................................................................................................69 Input Variables for Soil.......................................................................................70 MicroWEX-3 field observations..................................................................70 Land Surface Process (LSP) model..............................................................70 Input Variables for Vegetation............................................................................73 Dielectric Properties for Soil and Vegetation......................................................75 Evaluation Methodology............................................................................................75 Results and Discussion...............................................................................................76 The Early Season.................................................................................................76 Comparison using VSM and temperature from LSP...................................76 Comparison using observed VSM and temperature at 2 cm........................78 The Mid Season...................................................................................................79 Comparison using VSM and temperature from LSP...................................81 Comparison using observed VSM and temperature at 2 cm........................83 The Late Season..................................................................................................84 Comparison using VSM and temperature from LSP...................................85 Comparison using observed VSM and temperature at 2 cm........................87 The Complete Growing Season...........................................................................89 Conclusion..................................................................................................................90 6 SENSITIVITY ANALYSIS OF BRIGHTNESS TEMPERATURE TO CHANGES IN SOIL MOISTURE.............................................................................92

PAGE 7

vii MicroWEX-3 Season..................................................................................................92 Simulated Sensitivity Using MB-Cotton Model.................................................92 Observed Sensitivity............................................................................................99 MicroWEX-1 Season................................................................................................102 Comparison Between MicroWEX-1 and 3 Seasons.................................................104 Conclusion................................................................................................................105 7 CONCLUSIONS......................................................................................................107 Summary...................................................................................................................107 Contributions............................................................................................................112 Recommendations for Future Research....................................................................113 Improvements in Data Collection during Field Experiments............................113 Improvements in Brightness Modeli ng using the MB-Cotton Model...............113 LIST OF REFERENCES.................................................................................................116 BIOGRAPHICAL SKETCH...........................................................................................127

PAGE 8

viii LIST OF TABLES Table page 2-1. Radiometer specifications for UFCMR......................................................................13 2-2. Micrometeorological parameters measured by the FAWN station............................27 2-3 Growth stages of the cott on observed during MicroWEX-3.......................................38 3-1. Mean and standard deviation of the slop es (S) and Intercepts (I) for the H-and Vpol calibration curve during MicroWEX -1 (MW-1) and REBEX-10 (RB-10).......53 3-2. RMSE estimation for the operation and calibration for UFCMR and TMRS-3C......56 5-1. The physical and hydraulic properties of soil used in the LSP model (Yan, 2006)...72 5-2. The canopy parameters used in the LSP model..........................................................72 5-3. Regression coefficients and Ymax values for the canopy cover (Cc) and gravimetric water content of canopy (Mg) empirical models...................................74 5-4. The start and end times of the dry down period in the early season during MicroWEX-3............................................................................................................76 5-5. Mean absolute differences (MAD) and root mean square differences (RMSD) using MicroWEX-3 measurements at 2 cm (2 cm) and LSP estimations between 0 and 2 cm (0-2 cm) during the dry down in the early season during MicroWEX3.............................................................................................................................. ..78 5-6. The start and end times of the dry down period in the mid season during MicroWEX-3............................................................................................................80 5-7. Mean absolute differences (MAD) and root mean square differences (RMSD) using MicroWEX-3 measurements at 2 cm (2 cm) and LSP estimations between 0 and 2 cm (0-2 cm) during the dry down in the mid season during MicroWEX3.............................................................................................................................. ..83 5-8. The start and end times of the dr y down period in the late season during MicroWEX-3............................................................................................................85 5-9. Mean absolute differences (MAD) and root mean square differences (RMSD) using MicroWEX-3 measurements at 2 cm (2 cm) and LSP estimations between

PAGE 9

ix 0 and 2 cm (0-2cm) during the dry down in the mid season during MicroWEX3.............................................................................................................................. ..87 6-1. R-square (R2) values for the fits for the relationships between Mg, Hc, and PWC versus Cc...................................................................................................................94 6-2. The simulated and observed sensitivity values during MicroWEX-1 and 3............105

PAGE 10

x LIST OF FIGURES Figure page 1-1. Linkage between SVAT-MB models and da ta assimilation using remote sensing observations................................................................................................................4 2-1. Location of PSREU/IFAS..........................................................................................10 2-2. Location of the field site for Micr oWEX-1 at the UF/IFAS PSREU where the MicroWEXs were conducted...................................................................................10 2-3. Layout of the sensors during MicroWEX-1...............................................................11 2-4. University of Florida C-band Microwav e Radiometer (UFCMR): (a) field setup, (b) front view showing the antenna, a nd (c) side view showing the rotary system.......................................................................................................................13 2-5. Observed brightness temperatures at vertically (V-) and horizontally (H-pol) polarizations during MicroWEX-1...........................................................................14 2-6. Block diagram of the University of Florida C-band Radiometer (De Roo, 2003).....17 2-8. Latent (LE) and sensible heat fluxe s (H) observed by the eddy covariance system during MicroWEX-1................................................................................................19 2-9. Horizontal wind speed observed by the eddy covariance system during MicroWEX-1............................................................................................................19 2-10. Horizontal wind direction observed by the eddy covariance system during MicroWEX-1............................................................................................................19 2-11. Kipp and Zonen CNR-1 net radiometer...................................................................20 2-12. Downand up-welling solar radiati on observed by CNR-1 during MicroWEX-1..20 2-13. Downand up-welling far infrared radiation observed by CNR-1 during MicroWEX-1............................................................................................................21 2-14. Net total radiation observe d by CNR-1 during MicroWEX-1.................................21 2-15. Solar albedo observed by CNR-1 during MicroWEX-1..........................................21

PAGE 11

xi 2-16. Surface temperature observed by TIR sensor during MicroWEX-1........................22 2-17. Volumetric soil moisture conten t (VSM) observed by Hydra probes during MicroWEX-1............................................................................................................23 2-18. Volumetric soil moisture content (VSM) observed by TDR during MicroWEX1.............................................................................................................................. ..24 2-19. Volumetric soil moisture content (VSM) at 4cm observed by the Hydra and TDR probes during MicroWEX-1............................................................................24 2-20. Soil temperatures observed by th e Hydra probes during MicroWEX-1..................24 2-21. Soil temperatures at 4 cm observe d by Hydra probes during MicroWEX-1............25 2-23. Rainfall/irrigation at the east edge of the study site observed by the tipping bucket raingauge during MicroWEX-1....................................................................26 2-24. Rainfall/irrigation at the west edge of the study site observed by the tipping bucket raingauge during MicroWEX-1....................................................................26 2-25. Air temperature observed by FAWN during MicroWEX-1.....................................27 2-26. Soil temperature at 10 cm obser ved by FAWN during MicroWEX-1.....................27 2-27. Net radiation observed by FAWN during MicroWEX-1.........................................28 2-28. Relative humidity observed by FAWN during MicroWEX-1..................................28 2-29. Rainfall observed by FAWN during MicroWEX-1.................................................28 2-30. Observed volumetric soil moisture co ntent (VSM) derived by the gravimetric soil samples during MicroWEX-1............................................................................30 2-31. Observed volumetric soil moisture co ntent (VSM) derived by the gravimetric soil samples near the UFCMR footprint during MicroWEX-1................................30 2-33. Observed soil temperatures near th e UFCMR footprint during MicroWEX-1........31 2-34. Canopy height inside th e radiometer footprint observed during MicroWEX-1.......33 2-35. LAI inside the UFCMR footpr int observed during MicroWEX-1...........................33 2-36. Green biomass and plant water cont ent (PWC) observed during MicroWEX-1.....34 2-37. Location of the field site for Micr oWEX-2 and 3 at the UF/IFAS PSREU.............35 2-38. Layout of the sensors during MicroWEX-2 and 3...................................................36

PAGE 12

xii 2-39. Observed brightness temperatures at vertically (V-) and horizontally (H-pol) polarizations during MicroWEX-3...........................................................................38 2-40. Surface and soil temperatures at 2 cm observed by thermistor during MicroWEX-3............................................................................................................39 2-41. Volumetric soil moisture content (VSM) at 2 cm observed by the TDR probe during MicroWEX-3................................................................................................39 2-42. Canopy height observed during MicroWEX-3.........................................................39 2-43. Canopy width observe d during MicroWEX-3..........................................................40 2-44. LAI observed during MicroWEX-3.........................................................................40 2-45. Green biomass observe d during MicroWEX-3........................................................40 2-46. Plant water content (PWC) observed during MicroWEX-3.....................................41 2-47. Green biomass observed from NW area during MicroWEX-3................................41 2-48. Green biomass observed from NE area during MicroWEX-3.................................41 2-49. Green biomass observed from SW area during MicroWEX-3.................................42 2-50. Green biomass observed from SE area during MicroWEX-3..................................42 2-51. Plant water content (PWC) observe d from NW area during MicroWEX-3.............42 2-52. Plant water content (PWC) observe d from NE area during MicroWEX-3..............43 2-53. Plant water content (PWC) observe d from SW area during MicroWEX-3.............43 2-54. Plant water content (PWC) observe d from SE area during MicroWEX-3...............43 2-55. (a) Square, (b) bloom, (c) bo ll, and (d) open boll of cotton.....................................44 3-1. Slopes for the calibration curve at Hpol during MicroWEX-1. RMSE of EC = 1.20, TC = 1.84, and IC = 1.10 K/volt.....................................................................53 3-2. Slopes for the calibration curve at Hpol during MicroWEX-2. RMSE of EC = 1.12, TC = 1.43, and IC = 1.02 K/volt.....................................................................54 3-3. Slopes for the calibration curve at Hpol during MicroWEX-3. RMSE of EC = 1.12, TC = 1.43, and IC = 1.02 K/volt.....................................................................54 3-4. Slopes for the calibration curve at Hpol during REBEX-10. RMSE of EC = 1.20, TC = 1.84, and IC = 1.10 K/volt..............................................................................55

PAGE 13

xiii 3-5. RMSE due to antenna sidel obe during the lake observations....................................56 3-6. Comparison of the brightne ss temperature differences ( TB) by observations and LEM at H-pol on DOY 173. The total RMSE in observations and simulation are 1.46, 2.02, and 1.39 for EC, TC, and IC, respectively.............................................57 3-7. Comparison of the brightne ss temperature differences ( TB) by observations and LEM at H-pol on DOY 174. The total RMSE in observations and simulation are 2.58, 2.73, and 2.54 for EC, TC, and IC, respectively.............................................57 3-8. Dielectric constant of pure water simulated by Ulaby et al. (1986) and Meissner and Wentz (2004).....................................................................................................58 4-1. Schematic of the microwave brightness model for cotton.........................................60 4-2. Schematic of Teff approximation................................................................................62 4-3. Penetration depth at Cband as a function of volumetric soil moisture content using Equation 4.7....................................................................................................63 5-1. Soil profiles estimated by the LSP model..................................................................70 5-2. Surface and soil temperature profiles in the top 2 cm simulated by the Land Surface Process (LSP) model during MicroWEX-3................................................72 5-3. Soil moisture profiles in the top 2 cm simulated by the Land Surface Process (LSP) model during MicroWEX-3...........................................................................73 5-4. Input variables for vege tation: (a) Canopy height (Hc), (b) canopy cover (Cc), and (c) gravimetric water content of cotton (Mg) observed during MicroWEX-3. Error bars show one standard de viation around the mean values............................74 5-6. Comparison of the simulated and observed TB at H-pol at 6.7 GHz using (a) 0-2 cm LSP simulated soil temperature and moisture as inputs to the MB-Cotton model, (b) the 2 cm observed and 0 to 2 cm LSP simulated VSM, (c) the observed and LSP simulated surface temp eratures, and (d) the 2 cm observed and 0 to 2 cm LSP simulated soil temp eratures during th e early season of MicroWEX-3............................................................................................................77 5-7. Comparison of the simulated and observed TB at H-pol at 6.7 GHz during the early season of MicroWEX-3 using 2 cm field observed soil temperature and moisture as inputs to the MB-Cotton model............................................................79 5-8. Cotton canopy in the early season. Phot o on the left and right were taken on DOY 237 (August 24) and DOY 278 (October 4), respectively.......................................80 5-9. Comparison of the simulated and observed TB at H-pol at 6.7 GHz using (a) 0-2 cm LSP simulated soil temperature and moisture as inputs to the MB-Cotton

PAGE 14

xiv model, (b) the 2 cm observed and 0 to 2 cm LSP simulated VSM, (c) the observed and LSP simulated surface temp eratures, and (d) the 2 cm observed and 0 to 2 cm LSP simulated soil te mperatures during the mid season of MicroWEX-3............................................................................................................82 5-10. Comparison of the simulated and observed TB at H-pol at 6.7 GHz during the mid season of MicroWEX-3 using 2 cm field observed soil temperature and moisture as inputs to the MB-Cotton model............................................................84 5-11. Cotton canopy in the late season. The photo was taken on DOY 306 (November 1)............................................................................................................................. ..85 5-12. Comparison of the simulated and observed TB at H-pol at 6.7 GHz using (a) 0-2 cm LSP simulated soil temperature and moisture as inputs to the MB-Cotton model, (b) the 2 cm observed and 0 to 2 cm LSP simulated VSM, (c) the observed and LSP simulated surface temp eratures, and (d) the 2 cm observed and 0 to 2 cm LSP simulated soil temp eratures during the late season of MicroWEX-3............................................................................................................86 5-13. Comparison of the simulated and observed TB at H-pol at 6.7 GHz during the late season of MicroWEX-3 using 2 cm field observed soil temperature and moisture as input to the MB-Cotton model..............................................................88 5-14. Comparison of the observed and simulated TB at H-pol using 2 cm measured temperature and VSM and 0 to 2 cm temperature and moisture profiles estimated by the LSP model before DOY 196 during the whole season of MicroWEX-3............................................................................................................89 6-1. Relationships between the vegetation properties derived us ing data collected during MicroWEX-3................................................................................................94 6-2. Sensitivity of the model brightness te mperatures to changes in volumetric soil moisture (VSM) at different plant wa ter content (PWC) at H-pol. For this simulation, Tcanopy = Teff = 300 K.............................................................................96 6-3. Sensitivity of the model brightness te mperatures to changes in volumetric soil moisture (VSM) at different plant wate r content (PWC) at V-pol. . For this simulation, Tcanopy = Teff = 300 K.............................................................................96 6-4. Sensitivity of brightness temperatures to effective physical temperature of soil (Teff) for bare soil condition, i.e. PWC = 0 kg/m2....................................................98 6-5. Sensitivity of brightness temperatures to effective physical temperature of soil (Teff) for PWC = 0.5 kg/m2.......................................................................................98 6-6. Sensitivity of brightness temperatures to effective physical temperature of soil (Teff) for PWC = 1.2 kg/m2.......................................................................................98

PAGE 15

xv 6-7. Response of brightness temperatures at Vand H-pol to the soil moisture changes at 2 cm during the early season..............................................................................100 6-8. Response of brightness temperatures at Vand H-pol to the soil moisture changes at 2 cm during the mid season................................................................................100 6-9. Response of brightness temperatures at Vand H-pol to the soil moisture changes at 2 cm during the late season................................................................................101 6-10. Response of brightness temperatures at Vand H-pol to the soil moisture changes at 4 cm during the early season................................................................103 6-11. Response of brightness temperatures of Vand H-pol to the soil moisture changes at 4 cm during the late season..................................................................104

PAGE 16

xvi Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy LINKING CHANGES IN DYNAMIC CO TTON CANOPY TO PASSIVE MICROWAVE REMOTE SENSING By Kai-Jen Calvin Tien December 2006 Chair: Jasmeet Judge Major Department: Agricultur al and Biological Engineering Soil moisture is one of the most importan t variables in land-atmosphere processes. It determines how precipitation partitions into infiltration, surface runoff, and groundwater recharge. Additionally, soil mois ture is important in partitioning the available energy into the latent and sensible heat fluxes at the land surface. The control of soil moisture is the key mechanism for the feedback mechanisms between land and atmospheric fluxes. Accurate estimates of these land surface fluxes are essential for understanding and quantifying the global, regional, and local hydrological cycles. Even though the biophysics of moisture and energy transport is captured in most current Soil-VegetationAtmosphere-Transfer (SVAT) models that provide estimates of soil moisture, the computational errors accumulate over time and the model estimates diverge from reality. One promising way to significantly improve model estimates of soil moisture is by

PAGE 17

xvii assimilating remotely sensed data that are sensitive to soil moisture, for example, microwave brightness temperatures, and updating the model state variables. The microwave brightness at low frequencies is very sensitive to soil moisture in the top few centimeters in most vegetated surfaces. Most of the passive microwave brightness experiments for soil moisture retr ieval conducted in agri cultural terra ins have been short-term experiments that captured on ly parts of the growing season. Knowledge for the interactions between microwave brightness signatures and changes in soil moisture and temperatures for a dynamic agri cultural canopy, such as cotton, is very important during the whole growing seas on. Microwave brightness (MB) models simulating the terrain emission provide the opportunity to re late microwave signatures to soil moisture information. An integrated S VAT and MB model prov ides the opportunity to direct assimilate microwave remote sensing observations. The goal of this dissertation is to develop a MB model that can be used to simulate microwave brightness temperature ( TB) for the entire growing s eason of cotton. This MB model can be linked with existing SVAT mode ls such as the Land Surface Process (LSP) model for the cotton growing season to allow assimilation of passive microwave observations.

PAGE 18

1 CHAPTER 1 INTRODUCTION Soil moisture is a key variable governing land-surface processes. It determines the amount of precipitation that contributes to infiltration, surface r unoff, and groundwater recharge. Additionally, soil moisture is important in partitioning the available energy into the latent and sensible heat fluxes at the land surface. The control of soil moisture is the key mechanism for feedback mechanisms between land and atmospheric fluxes (Brubaker and Entekhabi, 1995; Entekhabi and Brubaker, 1995; Entekhabi et al., 1996). Accurate estimates of these land surface fluxes are essential for understanding and quantifying the local, regional, and global hydrological pathways . Many numerical SoilVegetation-Atmosphere-Transfer (SVAT) mode ls have been used for simulating the water and energy balance processes at the su rface and in vadoze zone. Some examples of such models include, BATS (Dickinson et al., 1986), ISBA (Noilhan and Mahfouf, 1989), SSiB (Xue et al., 1991), VIC (Wood et al., 1992), SiSPAT (Braud et al., 1995), SWEAT (Burke et al., 1997), PATTERN (M ulligan, 1998), and LSP (Judge, 1999; Judge et al., 2003). SVAT models pr ovide estimates of these fluxes to the Global Circulation Models (GCMs), to provide longand near-t erm weather predictions (Roy et al., 2001; Pruski and Nearing, 2002; Schmidt and Glad e, 2003; Shepherd and McGinn, 2003; Peres and DaCamara, 2004; Voldoire and Royer, 2004; Meier et al., 2005; Von Bloh et al., 2005; Kay et al., 2006a; Kay et al., 2006b). GCMs have also been used to investigate climate anomalies. The results from the GCM simulations demonstrated that soil

PAGE 19

2 moisture anomalies are closely related to anomalies in estimation of runoff and evapotranspiration (ET) (E ntekhabi et al., 1996). Even though flux estimates in SVAT mode ls are well represented and understood, the model estimates still diverge over time due to accumulated errors in numerical approximation, parameter estimation, and mode l initialization. One promising way to decrease the errors in flux estimates is to assimilate independent observations periodically into the SVAT models. Such data assimilation is based on a probabilistic framework that accounts for the uncertainty when combining different information sources, e.g., remotely sensed observations , in-situ measurements, and mathematical models (McLaughlin, 1995). McLaughlin (1995) discussed the early development in data assimilation for SVAT modeling. He concluded that remotely sensed observations could initiate dramatic changes in hydrologic pr actice. Microwave brightness observations provide complimentary data sources and can be used with in-situ measurements. Since then, a number of studies have been conducte d using different obser vations and different assimilation approaches (McLaughlin, 1995; Re ichle et al., 2001a; Re ichle et al., 2001b; Margulis et al., 2002; McLaugh lin, 2002; Reichle et al., 2002a; Reichle et al., 2002b; Aubert et al., 2003; Crow a nd Wood, 2003; Crow et al., 200 3; Crow, 2003; Heathman et al., 2003; Montaldo and Alberts on, 2003; Reichle and Koster , 2003; Houser et al., 2004; Margulis and Entekhabi, 2004; Reichle a nd Koster, 2004; Rodell and Houser, 2004; Walker and Houser, 2004; Berg et al., 2005; Crow et al., 2005a; Crow et al., 2005b; Dunne and Entekhabi, 2005; Gottschalck et al., 2005; Moradkhani et al., 2005; NiMeister et al., 2005; Reichle and Koster, 2005; Dunne and Entekhabi, 2006). Because data assimilation must balance information fr om different sources, proper representation

PAGE 20

3 of uncertainty is essential (McLaughlin, 2002). Remotely sensed observations by the space-borne, airborne, and ground-based sens ors provide the possibility for data assimilation in SVAT models at different spatial and temporal scales. Microwave observations at low frequencies (< 10 GHz) are very sensitive to soil moisture in the top few centimeters in most vegetated surfaces (Bruckler et al., 1980; Mo et al., 1980; Newton and Rouse, 1980; Ulaby et al ., 1981; Wang and Choudhury, 1981; Burke and Schmugge, 1982; Schmugge, 1985; Schmugge et al ., 1985; Ulaby et al., 1986; Jackson et al., 1992; Jackson et al., 1993; Jackson et al., 1995; Jackson et al., 1998; Owe and van de Griend, 1998; Laymon et al., 2001; Le Vine et al., 2001; Paloscia et al., 2001; Jackson et al., 2002; Jackson et al., 2005; McCabe et al., 2005; Schwa nk et al., 2005; Wen et al., 2005). The microwave observations , brightness temperatures (TB), are estimated by a Microwave Brightness (MB) model using the te mperature and moisture properties of the surface. Recently studies have been conducted to develop SVAT models that are linked with microwave emission or microwave br ightness (MB) models. These integrated models allow for assimilation of microwave obs ervations (Judge et al ., 1999; Burke et al., 2001; Burke et al., 2002; Burke et al., 2003; J udge et al., 2003; Montaldo and Albertson, 2003; Burke et al., 2004; Demart y et al., 2005; Crow et al., 2005b). Figure 1-1 shows the linkage between SVAT-MB models and da ta assimilation using remote sensing observations. The SVAT models use weather data as forcing to estimate the moisture and energy fluxes, and profiles of soil temperature and moisture in the vadoze zone. Using the soil temperature and moisture profiles as inputs, MB models estimate brightness temperatures (TB) of the terrain. Data assimilation utilizes the differences between TB

PAGE 21

4 estimated by the MB model and those obs erved by remote sensing, along with the associated errors to mini mize the uncertainty in TB. Finally, by model inversion the optimized TB can be used to update the moisture and temperature profiles in the SVAT model. Figure 1-1. Linkage between SVAT-MB models and data assimilation using remote sensing observations. The goal of this dissertation is to develop a MB model that can be used to simulate microwave TB for the entire growing season of cott on as shown in the red box. This MB model can be linked with existing SVAT mode ls such as the Land Surface Process (LSP) model for the cotton growing season to allow assimilation of passive microwave observations. Historical Development of Microwave Remote Sensing for Soil Moisture Microwave remote sensing provides observa tions that are independence of solar illumination and of cloudy and low precipitation conditions. Due to longer wavelengths,

PAGE 22

5 microwaves also penetrate into the soil and vegetation canopy. There are two fundamental types of microwave sensing a pproaches: active and passive. Both types utilize the large contrast of the dielectric properties between the atmosphere and land to estimate microwave signatures from the te rrains. Compared to passive microwave sensing, active sensing is more sensitive to the geometric and dielec tric properties of the land surface (Ulaby et al., 1981). Many ground-based, air-borne, and space-born e observations have been used for passive microwave remote sensing studies. Most of these studies involve retrieval of soil moisture and other surface variables using pa ssive microwave observations. For example, Wigneron et al. (1996) used microwave observations at 1.4 ( = 21 cm) and 5 GHz ( = 6 cm) to monitor hydrological variables over agricultural land. Njoku and Li (1999) used passive microwave observations at 6 to 18 GHz ( = 1.6 to 5 cm) to retrieve land surface variables, e.g., soil moisture, vegetation wate r content, and surface temperature. Jackson (2001) studied the relation be tween brightness temperature at low frequency at 1.4 GHz ( = 21 cm) and soil moisture over the spatia l resolutions of 800 m and 1600 m for the retrieval algorithms for the near-surface soil moisture. Calvet et al. (1996), Jackson and Le Vine (1996), Jackson (2001), Kerr et al. (2001), Njoku et al. (2002), Njoku et al. (2003), Drusch et al. (2004), and Crow et al . (2005a) used airand space-borne passive microwave imagery at 1.4, 5.5, 6.9, and 36.5 GHz ( = 21, 5.5, 4.4, and 0.8 cm) to retrieve soil moisture in regional scales. They found that although the spatial resolution of the space-borne microwave observation is quite coarse (~50 km), it is still within the range of most regional hydrologic models. Pard e et al. (2004) invest igated the possibility of implementing a multiple parameter retrieval approach based on the ground-based L-

PAGE 23

6 band ( f = 1.4 GHz or = 21 cm) radiometer observations . Crosson et al. (2005a, 2005b) analyzed the parameter sensitivity of soil mo isture retrievals at L-, Cand X-band ( f = 1.4, 6.7, and 10 GHz or = 21, 5.5, and 3 cm, respectively) empirically using airborne radiometers. They concluded that the accur acy for model parameters in the current retrieval approach, using single-frequency a nd single-polarization retrieval algorithm, might not be able to meet the hydrologic requ irements of the soil moisture retrieval from remote observations ( 4 % in volumetric soil moisture), given the spatial and temporal resolution of space-borne microwave radiomet ers. Davenport et al. (2005) analyzed the parameter sensitivity of soil moisture retrievals at L-band ( f = 1.4 GHz or = 21 cm) theoretically using a simple zero-order MB model for a terrain consisting of a layer of vegetation canopy on top of a semi-infinite soil layer. Their findings showed that vegetation canopy properties produced the most uncertainty in soil moisture retrievals. The low microwave frequencies are bette r for increased sampling depth and for reduced noise effects caused by the vegeta tion canopy and soil surface roughness. The sensitivity of TB to soil moisture changes decrease s with increasing vegetation. The soil contribution is attenuated significantly th rough the vegetation. Therefore, a better understanding of modeling emission from dynamic vegetation canopies during the growing season is required (Jackson et al ., 1999; Njoku and Li, 1999; Ferrazzoli et al., 2000; Crosson et al., 2005a; Davenport et al., 2005). Dissertation Objectives This dissertation aims to understand th e interactions between the passive microwave signatures, and soil and vegeta tion dynamics during a growing season of cotton. Cotton is one of the most important agricultural crops in the SE-US region in

PAGE 24

7 terms of both acreage and economic value. It occupies about 13.2 million acres in this region with a related value exceeding $ 50 b illion nationwide and supplies about 20% of the world output (US Cotton Market, 2002). Beca use it is increasingly an irrigated crop, its management influences water cycling sign ificantly, including the changes in aquifer storage and recharge, and evapotranspira tion (ET), causing changes in land-surface atmosphere exchange. The dissertation focuses on the developm ent of a physically based MB-Cotton model which simulates TB at C-band ( f = 6.7 GHz; = 4.4 cm) during the entire growing season, and on comparison of the TB estimated by the MB -Cotton model to the TB observed by the University of Florida Cband Microwave Radiometer (UFCMR) during the Microwave, Water, and Energy Balance Experiments (MicroWEXs). The research questions addressed in th is dissertation include the following: 1. What are the accuracy and precision of the TB measurements used in this study? (Chapter 3) 2. What is the effective depth of C-band under different soil moisture conditions? (Chapter 4) 3. How well does the MB-Cotton model cap ture the observed brightness using soil moisture at 2 cm? (Chapter 5) 4. How well does the MB-Cotton model captu re the observed brightness if the detailed soil moisture information is available for the effective depth? (Chapter 5) 5. Is scattering in the canopy impor tant for simulating accurate TB when cotton reaches maturity? (Chapter 5) 6. How does the sensitivity of TB to soil moisture changes as cotton mature? (Chapter 6) Dissertation Format In this dissertation, Chapter 2 contains the description and observations made during the first, second, and third Microw ave Water and Energy Balance Experiments (MicroWEX-1, 2, and 3). Chapter 3 contai ns the discussion for the radiometric calibration for the C-band microwave radiom eter, UFCMR, using the calibration data

PAGE 25

8 collected during the MicroWEXs. Chapter 4 includes the description of Microwave Brightness for the Cotton (MB-Cotton) model. Chapter 5 contains the model calibration for MB-Cotton model using data collected during MicroWEX-3. Ch apter 6 includes the discussion of the sensitivity analysis of the modeled TB using MB-Cotton and observed TB to the changes in moisture and temper ature during MicroWEX-1 and 3. Chapter 7 provides a summary of the results, origin al contributions, and recommendations for future research.

PAGE 26

9 CHAPTER 2 MICROWAVE WATER AND ENERGY BALANCE EXPERIMENTS This chapter contains the description of a series of field experiments, called Microwave, Water, and Energy Balance Expe riments (MicroWEXs) and the observations during the experiments that are used in this dissertation. These experiments were conducted by the Center for Remote Sensi ng, Agricultural and Bi ological Engineering Department at the Plant Science Research and Education Unit, IFAS during growing seasons of cotton (MicroWEX-1 in 2003) , corn (MicroWEX-2 in 2004), and cotton (MicroWEX-3 in 2004). The observations of mi crowave brightness of terrain, microwave absorber, and micrometeorological conditi ons collected during MicroWEX-1, 2, and 3 were used for microwave radiometer calib ration in Chapter 3. The observations of vegetation and soil conditions, micrometeorol ogical parameters, a nd terrain microwave brightness during MicroWEX-3 were used to calibrate the microwave brightness model for cotton (MB-Cotton) developed in Chapte r 5 and to discuss the sensitivity of brightness temperatures to changing soil and vegetation conditions in Chapter 6. The First Microwave Water and Energy Balance Experiment (MicroWEX-1) Figures 2-1 and 2-2 show the location of the PSREU and the study site for the MicroWEX-1, respectively. The area of th e study site was a 130 m X 75 m. A linear move system was used for irrigation. The co tton was planted on July 9 (Day of Year in 2003, DOY 190) at an orientation 60 from Ea st as shown in Figure 2-3. The plant spacing was about 4 cm and the row spacing was 90 cm. Instrument installation took place on July 17 (DOY 198).

PAGE 27

10 Figure 2-1. Location of PSREU/IFAS. Figure 2-2. Location of the fiel d site for MicroWEX-1 at the UF/IFAS PSREU where the MicroWEXs were conducted.

PAGE 28

11 The instruments consisted of a ground-ba sed microwave radiometer system and micrometeorological stations. The ground-ba sed microwave radiometer system was installed at the middle of the north edge of the site facing south to avoid the radiometer shadow interfering with the field of view (see Figure 2-3). The micrometeorological station was installed at the center of the fiel d and included soil moisture and temperature probes, soil heat flux plates, thermal infrared sensor, net ra diometer, and raingauges. Two additional raingauges also were installed at the east and we st edge of the radiometer footprint to capture the irriga tion. Later, an eddy covariance system was installed at the northeast corner of the site on August 15 ( DOY 227). Detailed descriptions of field observations and a field log during MicroWEX-1 can be found in the experimental data report (Tien et al., 2005) at Electronic Data Source of UF/IFAS Extension (http://edis.ifas.ufl.edu/AE288 ). Because the lack of soil moisture data due to lightening damage to the sensors during MicroWEX-1, observations during this experiment could only used for the microwave radiometer calibration (Chapter 3). Figure 2-3. Layout of the se nsors during MicroWEX-1.

PAGE 29

12 Sensors MicroWEX-1 had two major instrument subsystems: the ground-based University of Florida C-band Radiometer (UFCMR) a nd the micrometeorological subsystems, a collection of commercially available instruments. University of Florida C-band Microw ave Radiometer (UFCMR) System Microwave brightness temp eratures at 6.7 GHz ( = 4.48 cm) were measured every 30 minutes using the University of Florid a’s C-band Microwave Radiometer system (UFCMR) (Figure 2-4 (a)). The radiometer system consiste d of a dual polarized total power radiometer operating at the center frequency of 6.7 GHz housed atop a 10 m tower installed on a 16’ trailer bed. UFCMR wa s designed and built by the Microwave Geophysics Group at the University of Michig an (UM-MGG). It ope rated at the center frequency at 6.7 GHz that is similar to one of the center frequenc ies on the space-borne Advanced Microwave Scanning Radiometer (AMSR) aboard the NASA Aqua Satellite Program. Given the antenna beam width of 23 and the height between the antenna aperture and ground surface of 7.6 m, the si ze of the footprint was about 11 X 11 m2. A rotary system was used to rotate the in cidence angle of the UFCMR both for field observations and sky measurements. The terr ain brightness temperat ures were observed at an incidence angle of 55 matching that of the space-borne AMSR-E sensor. The radiometer was calibrated every two weeks with a microwave absorber as warm load and measurements of sky at zenith angles of 15, 30, 45, and 60 as cold loads. The calibration of UFCMR will be discussed in de tail in Chapter 3. Figure 2-4 (b) and (c) show the close-up of the rotary system a nd the antenna of the UFCMR, respectively. Table 2-1 lists the speci fications of UFCMR.

PAGE 30

13 Table 2-1. Radiometer specifications for UFCMR. Parameter Value Antenna type Potter Horn Frequency Center 6.7 GHz Bandwidth 3 dB 20 MHz 3 dB V-pol elev & azim 23 & 21 Beamwidth 3 dB H-pol elev & azim 21 & 23 Isolation > 27 dB Polarizations Sequential V/H Noise Figure From Trec 3.99 dB RF gain 85 dB 1 sec 0.71 K NEDT 8 sec 0.25 K First Side-lobe 47 from bore-sight -26 dB Figure 2-4. University of Florida C-band Mi crowave Radiometer (UFCMR): (a) field setup, (b) front view showing the ante nna, and (c) side view showing the rotary system.

PAGE 31

14 Figure 2-5 shows the Vand H-pol bright ness signatures observed by the UFCMR during MicroWEX-1. The TB at H-pol were lower than that at V-pol early in the season and rapidly increased with the increasing c ontribution from the growing vegetation. The lowest TB at H-pol was around 160 K on DOY 206.5 at the beginning of the season. The H-pol TB reached to 260 K on DOY 240. The TB at H-pol are more sensitive to changes in moisture than at V-pol for bare soil while V-pol primarily responds to soil temperature fluctuations at the incidence angle of 55 , close to Brewster angle at microwave frequencies (Ulaby et al., 1981). The difference in TB at Hand V-pol before DOY 240 is primarily due to low emissivities at H-pol than those at V-pol, when the land surface is sparsely vegetated and the canopy contribu tion is minimal. The maximum difference between the TB at two polarizations was 90 K. Figure 2-5. Observed brightness temperatures at vertically (V-) a nd horizontally (H-pol) polarizations during MicroWEX-1. The TB at Vand H-pol converged on DOY 240, 50 days after planting corresponding to LAI of 1.4, plant water content (PWC) of 0.75 kg/m2 (see Figure 2-35 and 2-36), and the canopy height of 80 cm (Figure 2-34). After DOY 240, a maximum difference of 10 K between the TB at two polarizations was observed. The terrain TB at

PAGE 32

15 both polarizations were domi nated by emission from the canopy. Most of the polarization dependent energy emitted from the soil was attenuated by the vegetation canopy and the observed TB were primarily due to emission fr om the canopy. The canopy emission was polarization independent because the mois ture distribution within the canopy is statistically random at 6.7 GHz, resulting in similar TB at the two polarizations. The most important goal for microwave radi ometer operation is to maintain thermal control inside the radiometer to assure the measurement consistency. UFCMR uses a thermoelectric cooler (TEC) for thermal contro l of the Radio Frequency (RF) stages for the UFCMR. This is accomplished by the Oven Industries “McShane” thermal controller. McShane is used to cool or heat by a Pr oportional-Integral-Derivative (PID) algorithm with a high degree of precision at 0.01C. The aluminum plate to which all the RF components are attached is chosen to have su fficient thermal mass to eliminate short-term thermal drifts. All components attached to this thermal plate, including the TEC, use thermal paste to minimize thermal gradients across junctions. The majority of the gain in the system is provided by a gain and filtering block designed by the University of Michigan for the STAR-Light instrument (De Roo, 2003). The main advantage of this gain block is the close proximity of all the amplifiers, simplifying the task of thermal control. This gain block was designed for a radiometer working at the radio astronomy window of 1400 to 1427 MHz, and so the receiver is a heterodyne type with downc onversion from the C-band RF to L-band. To minimize the receiver noise figure, a C-band low-noise amplifier (LNA) is used just prior to downconversion. To protect the amplifier from saturation due to out of band interference, a relatively wide bandwidth, but low insertion lo ss, bandpass filter is used just prior to the

PAGE 33

16 amplifier. Between the filter and the antenna are three com ponents: a switch for choosing polarization, a switch for mon itoring a reference load, and an isolator to minimize changes in the apparent system gain due to differences in the reflections looking upstream from the LNA. The electrical penetrations use commer cially available weatherproof bulkhead connections (Deutsch connector s or equivalent). The heat sinks have been carefully located employing RTV (silicone sealant) to seal the bolt holes. The radome uses 15 mil polycarbonate for radiometric signa l penetration. It is sealed to the case using a rubber gasket held down to the case by a square retainer. The first electomechanical latching RF switch switches between Hand Vpolarization. The second latching RF switch, sw itches between the analog signal from the first switch and the reference load. From here, the signal passes through an isolator, which limits the coherent interference of the receiver noise with itself as it is reflected by the switches. While this type of coherent interference oc curs everywhere in the front end, the switches have different return loss de pending on the switch position. The nearly equal temperatures of the isolator termination and the switches themselves insure that the sum of the noise generated by the switches and reflected by the switches is nearly constant with respect to switch pos ition. The signal then passes th ru a bandpass filter with a 6.7GHz center frequency, which protects the amplifier from saturation by out of band Radio Frequency Interference (RFI). A Low No ise Amplifier (LNA) is used to amplify the weak signals captured by the antenna, a nd thereby limits the contributions of downstream components to the receiver noise figure. A mixer takes the input from the LNA and a local oscillator to output a 1.4 GHz signal to the STAR-Lite block. After the

PAGE 34

17 Power Amplifier and Filtering Block (the ST AR-Lite back-end), the signal is passed through a Square Law Detector and a Post-D etection Amplifier. UFCMR is equipped with a Z-World BL1720 microcontroller that ha s responsibility for taking measurements, monitoring the thermal environment, and st oring data until a downl oad is requested. A laptop computer is used for running th e user interface, named FluxMon, which communicates with the radiometer with a Radiometer Control Language (RadiCL). The radiometer is configured to maintain a part icular thermal set point, and make periodic measurements of the brightness. Each measurement is a sequence of brightness observations of the reference load, then horizon tal and vertical polarizations and then the reference load again. The data collected by th e radiometer is not calibrated within the instrument, since calibration erro rs could corrupt an otherwis e useful dataset. Figure 2-6 shows the block diagram of UFCMR. Figure 2-6. Block diagram of the University of Florida C-band Radiometer (De Roo, 2003).

PAGE 35

18 Micrometeorological Subsystems The micrometeorological subsystems de ployed during MicroWEX-1 includes eddy covariance system, net radiometer, thermal infr ared sensor, soil moisture and temperature probes, soil heat flux plates , and the Florida Automated We ather Network station located at the PSREU (http://fawn.ifas.ufl.edu/). Eddy covariance system A Campbell Scientific eddy covariance system (Campbell Scientific, 1998a; Campbell Scientific, 1998b) was located at the s outheast corner of the field (see Figure 27). The system included a CSAT3 anemometer and KH20 hygrometer. CSAT3 is a three dimensional sonic anemometer, which meas ures wind speed and the speed of sound on three nonorthogonal axes. Orthogonal wind spee d and sonic temperature are computed from these measurements. KH20 measures the water vapor in the atmosphere. Its output voltage is proportional to the water vapor density flux. During MicroWEX-1, latent and sensible heat fluxes were measured every 30 minutes at the height of 2.1 m from the ground, with the CAST3 pointed at 209 with respect to the s outhwest direction. Figure 2-8 illustrates the raw latent (LE) and se nsible heat fluxes (H) observed by the eddy covariance system during MicroWEX-1. Figure 2-9 and 2-10 show the horizontal wind speed and direction observed by the eddy covariance system during MicroWEX-1, respectively. Figure 2-7. Eddy covariance system.

PAGE 36

19 Figure 2-8. Latent (LE) and se nsible heat fluxes (H) obser ved by the eddy covariance system during MicroWEX-1. Figure 2-9. Horizontal wind speed observed by the eddy covariance system during MicroWEX-1. Figure 2-10. Horizontal wind direction observed by the e ddy covariance system during MicroWEX-1.

PAGE 37

20 Net radiometer A Kipp and Zonen CNR-1 four-component net radiometer (Campbell Scientific, 2006) was located at the center of the field to measure upand down-welling shortand long-wave infrared radiation. The sensor c onsists of two pyranometers (CM-3) and two pyrgeometers (CG-3) as shown in Figure 2-11. Th e sensor was installed at the height of 2.5 m above ground and facing south. Figure 212 to 2-15 illustrates the upand downwelling solar (shortwave) wave radiation, upand down-welling far infrared (longwave) radiation, net total radiation, and so lar albedo observed during MicroWEX-1, respectively. Figure 2-11. Kipp and Zonen CNR-1 net radiometer. Figure 2-12. Downand up-welling sola r radiation observed by CNR-1 during MicroWEX-1.

PAGE 38

21 Figure 2-13. Downand up-welling far infr ared radiation observed by CNR-1 during MicroWEX-1. Figure 2-14. Net total radiation obs erved by CNR-1 during MicroWEX-1. Figure 2-15. Solar albedo observe d by CNR-1 during MicroWEX-1.

PAGE 39

22 Thermal infrared sensor An Everest Interscience thermal infrared (TIR) sensor (4000.3ZL) was collocated with the net radiometer to observe skin temp erature at nadir. Given that the sensor was installed at the height of 2.5 m and a field of view of 15, th e size of the footprint for the thermal infrared sensor was 66 cm by 66 cm. Figure 2-16 shows the surface thermal infrared temperature obser ved during MicroWEX-1. The data gap between DOY 219 and 290 is due to TIR sensor failure fr om condensation inside the sensor. Figure 2-16. Surface temperature observe d by TIR sensor during MicroWEX-1. Soil moisture and temperature probes Five standard Vitel Hydra soil moisture and temperature probes and five Campbell Scientific time-domain water content reflec tometer (CS616) were used to measure soil volumetric water content (VSM) and temperatur e at depths of 4, 8, 12, and 20 cm within the rows every 15 minutes. The observations we re duplicated at the depth of 4 cm near the root zone. Figure 2-17 and 18 show the volumetric soil moisture content observed by the Hydra and TDR probes, respectively. Figur e 2-19 shows the volumetric soil moisture content observed at 4 cm in the middle of two rows and near the root area by the Hydra and TDR probes, respectively. The probes were installed ~1 m apart to avoid

PAGE 40

23 interference. However, it s hould be bear in mind that the two locations of the VSM profile measured by the Hydra and TDR probe s were separated by a distance of ~15 meters. During the late season, the VSM peaks at 8 cm were sometimes higher than those at 4 cm, for example, during the precip itation events on DOY 301, 323, 331, and 332; but sometimes lower than or equal to those at 4 cm, for example, during the precipitation events on DOY 309 and 348. This is mainly because of the heterogeneity of the infiltration due to the water intercepti on by the spatially va riable canopy. The VSM measured by the Hydra probes were higher than those measured by the TDR probes, while the diurnal variation of the VSM meas ured by the Hydra probes was less than those measured by the TDR probes. Figure 2-20 show s the soil temperatures observed by the Hydra probes. Figure 2-21 shows the soil temper atures observed at 4 cm in the middle of two rows and near the root area by the H ydra probes during Micr oWEX-1. The data gap between DOY 215 and 302 was caused by damage to the Hydra probes due to lightening. After DOY 293, the TDR probes were installe d alongside the Hydra probes to measure VSM at the same depths as those of TDR probes. Figure 2-17. Volumetric soil moisture c ontent (VSM) observed by Hydra probes during MicroWEX-1.

PAGE 41

24 Figure 2-18. Volumetric soil moisture content (VSM) observed by TDR during MicroWEX-1. Figure 2-19. Volumetric soil moisture cont ent (VSM) at 4cm observed by the Hydra and TDR probes during MicroWEX-1. Figure 2-20. Soil temperatures observed by the Hydra probes during MicroWEX-1.

PAGE 42

25 Figure 2-21. Soil temperatures at 4 cm observed by Hydra probes during MicroWEX-1. Soil heat flux plates Two Campbell Scientific soil heat flux pl ate (HFT-3) were used to measure soil heat flux at the depths of 4 and 8 cm, in ro w and near the root ar ea, respectively. Figure 2-22 shows the soil heat fluxes observed duri ng MicroWEX-1. The soil heat fluxes were high in the beginning of the season when the surface was mostly bare soil and cotton canopy had low biomass. As the canopy bioma ss increased, the soil h eat fluxes decreased during the growing season. Figure 2-22 Soil heat fluxes obser ved by HFT-3 during MicroWEX-1.

PAGE 43

26 Tipping buck raingauges Four raingauges were used to collect pr ecipitation and irri gation across the study site, as shown in Figure 2-3. Fi gure 2-23 and 2-24 shows the ra infall/irrigation at the east and west edge of the study site obser ved by the tipping bucket raingauge during MicroWEX-1, respectively. Figure 2-23. Rainfall/irrigation at the east edge of the study site observed by the tipping bucket raingauge during MicroWEX-1. Figure 2-24. Rainfall/irrigation at the west e dge of the study site observed by the tipping bucket raingauge during MicroWEX-1. Florida automated weather network (FAWN) Data from one of the Florida Automated Weather Network (FAWN, 2006) sites were also available during the MicorWEX-1. Table 2-2 shows the parameters measured

PAGE 44

27 by FAWN. Figure 2-25 to 2-29 show s the air temperature, soil temperatures at 10 cm, net radiation, relative humidity, and precipitati on observed by FAWN during MicroWEX-1. Table 2-2. Micrometeorological parame ters measured by the FAWN station. Parameter Unit and Description Air temperatureC, at 60 cm height Soil temperature C, at 10 cm depth Rainfall mm Relative humidity % Solar radiation W/m2 Figure 2-25. Air temperature obser ved by FAWN during MicroWEX-1. Figure 2-26. Soil temperature at 10 cm observed by FAWN during MicroWEX-1.

PAGE 45

28 Figure 2-27. Net radiation observe d by FAWN during MicroWEX-1. Figure 2-28. Relative humidity obs erved by FAWN during MicroWEX-1. Figure 2-29. Rainfall observed by FAWN during MicroWEX-1.

PAGE 46

29 Soil Sampling Extensive soil sampling was conducted to provide additional information of the spatial distribution of the surface soil mois ture, temperature, a nd surface roughness. The locations of the soil sampling sites are shown in Figure 2-3. Gravimetric Soil Moisture The gravimetric soil moisture (GSM) wa s sampled at 0-4,4-8, 8-12, and 12-20 cm by a coring tool. The soil samples were weight when wet and then were oven-dried at 105C for 24 hours. The inner diameter of the coring tool cylinder was 4.3 cm (d), the VSM and GSM of the soil samples were calculated as follows, L d s s VSMd w 22 (2.1) d d ws s s GSM (2.2) where sw and sd are the wet and dry we ights of a soil sample, L is the length of the coring tool section. Figure 2-30 a nd 2-31 show the volumetric soil moisture content derived by the gravimetric soil samples measured during MicroWEX-1. Soil Temperature A Max/Min waterproof digital thermometer from Forestry S upplier was used to measure the soil temperature at the depths of 2, 4, 8, and 16 cm at the same locations and time as the soil moisture sampling. The near surface soil temperature at the depth of 2 cm changed rapidly. This was primarily due to the fact that canopy cover was not uniform throughout the field. Figure 2-32 and 33 s how the soil temperature observed during MicroWEX-1.

PAGE 47

30 Figure 2-30. Observed volumetric soil moisture content (VSM) derived by the gravimetric soil samples during MicroWEX-1. Figure 2-31. Observed volumetric soil moisture content (VSM) derived by the gravimetric soil samples near the UFCMR footprint during MicroWEX-1.

PAGE 48

31 Figure 2-32. Observed soil temp eratures during MicroWEX-1. Figure 2-33. Observed soil temperatures ne ar the UFCMR footprin t during MicroWEX-1.

PAGE 49

32 Vegetation Sampling Vegetation properties such as stand dens ity, row spacing, height, biomass, and LAI were measured weekly during the field expe riment. The crop density derived from the stand density and row spacing was measured during the first two samplings since the cotton seeds were planted in the fixed sp acing and the germination rate was over 90% throughout the field. The specific bi-weekly me asurements included height, biomass, and LAI. In the early season, th e first four vegetation samplings were conducted on seven spatially distributed sampling locations (Figure 2-3). The locations were chosen to characterize the spatial variability of the vegetation propert ies in the study site on July 29, July 31, August 8, and August 17 (DOY 210, 212, 220, and 229, respectively). After that, one vegetation sampling location was chosen to better represent the crop height within the radiometer footprint at least 30 meters from the field boundary. Canopy Height Crop height was measured by placing a m easuring tape at the soil surface to average height of the crop. The height insi de the UFCMR footprint and at the vegetation sampling area were taken for each vegetation sampling. Figure 2-34 shows the crop height observed during MicroWEX-1. Leaf Area Index (LAI) LAI of the footprint was approximate d by measuring 24 locations around the footprint every week. A Sunscan canopy analysis system was used for the first two LAI samplings. After August 24 (DOY 236), LAI was measured with a Li-Cor LAI-2000 in the inter-row region with 4 cross-row measurements. The LAI-2000 was set to average 4 locations into a single value so one ob servation was taken above the canopy and 4 beneath the canopy; in the row, of the wa y across the row, of the way across the

PAGE 50

33 row, and of the way across the row. This ga ve a spatial average for row crops of partial cover. Figure 2-35 shows the LA I observed during MicroWEX-1. Figure 2-34. Canopy height inside the radiom eter footprint observ ed during MicroWEX1. Figure 2-35. LAI inside the UFCMR f ootprint observed during MicroWEX-1. Green and Dry Biomass Each biomass sampling was conducted along one row. Along the row, a distance of one meter was recorded as the sampling length. The plants within this length were picked and put in plastic bags to prevent moisture loss. In the laborato ry, the canopy samples were separated into leaves, stems, and bolls to measure their wet we ights. The vegetation samples were put into paper bags and dried in an oven at 75C for 48 hours. Then the

PAGE 51

34 vegetation samples were removed from the ove n and the dry weights were measured. The plant water content (PWC) then was calculated as: d wB B PWC (2.3) where Bw and Bd are the green and dry biomass (kg/m2), respectively Figure 2-36 shows the green biomass and PWC observed during MicroWEX-1. Figure 2-36. Green biomass and plant water content (PWC) observed during MicroWEX1. The Second Microwave Water and Energy Balance Experiment (MicroWEX-2) MicroWEX-2 was a similar experiment to MicroWEX-1 conducted during the corn growing season from March 17 (DOY 77) to June 3 (DOY 155) in 2004 (Judge et al., 2005). The study site was near that of MicroWEX-1, with an area of 183 X 183 m2 (Figure 2-37). During MicroWEX-2, the UFCMR were cal ibrated every two weeks. In this dissertation, only observations of sky brightness signatures, microwave absorber, and micro meteorological conditions were used to calibrate UFCMR and to investigate the accuracy and precision of the calibration in Ch apter 3. Detailed descriptions of field observations and field log during MicroWEX-2 ar e given in the experimental data report

PAGE 52

35 (Judge et al., 2005) at th e Electronic Data Source of UF/IFAS Extension (http://edis.ifas.ufl.edu/AE360 ). Figure 2-37. Location of the field site for MicroWEX-2 and 3 at the UF/IFAS PSREU. The Third Microwave Water and Energy Balance Experiment (MicroWEX-3) MicroWEX-3 was conducted utilizing the same study site as during MicroWEX-2 (Figure 2-37) during the gr owing season of cotton from June 16 (DOY 168) through December 21 (DOY 356) in 2004 (Lin et al., 2005). The crop spacing was about 8 cm and the row spacing was 76.2 cm. The inst ruments consisted of the UFCMR and micrometeorological stations (Figure 2-38). The UFCMR was deployed at the same incidence angle of 55 and height of 7.6 m as during Mi croWEX-1. The observations were made every 15 minutes.

PAGE 53

36 The micrometeorological stations include d measurements of soil moisture and temperature profiles, surface thermal infrared temperature, downand up-welling shortand long-wave radiation, so il heat flux, latent and se nsible heat fluxes, and precipitation/irrigation de pths. Detailed descriptions of field observation and field log for MicroWEX-3 can be found in the field data re port (Lin et al., 2005) at Electronic Data Source of UF/IFAS Extension (h ttp://edis.ifas.ufl.edu/AE361). Figure 2-38. Layout of the sens ors during MicroWEX-2 and 3. The observations of sky brightness si gnatures, microwave absorber, and micrometeorological conditions collected dur ing MicroWEX-3 were used for microwave radiometer calibration in Chapter 3. The fiel d observations of soil surface temperature and soil temperature and moistu re at depth of 2 cm during MicroWEX-3 were used to

PAGE 54

37 develop and evaluate the microwave brightne ss model in Chapter 5, due to the lack of soil moisture data from lightening damage to the sensors during the MicroWEX-1. The observations of terrain bright ness signatures, soil moisture , and temperature were also used to understand the sensitivity of brightness signatures to changing soil and vegetation conditions in Chapter 6. Figure 2-39 shows the observed TB during MicroWEX-3. The lowest TB at H-pol was around 120 K on DOY 205 in the early season. The H-pol TB reached to 260 K on DOY 245. The maximum difference between the TB at two polarizations was 110 K. The TB at Vand H-pol converg ed on DOY 262, 71 days after planting corresponding to LAI of 1.9, PWC of 0.60 kg/m2 (see Figure 2-39,2-44, and 246), and the canopy height and width of 60 a nd 50 cm, respectively (see Figure 2-42 and 2-43). After DOY 262, a maximum difference of 20 K between the TB at two polarizations was observed. Toward the late season during a period of 5 days from DOY 310 to 315, both TB at Hand V-pol decreased by ~15 K. This is because the decrease in Wc and LAI during the late season (Figure 2-43 and 2-44). Although the values of biomass and PWC still increased during this period mainly because the significantly increases of biomass and PWC in cotton bolls , those values decreased for leaves and stems (Figure 2-45 to 2-54). TB at both polarizations responded to the precipitation and/or irrigation in the whole growing season during MicroWEX-3. Figure 2-40 and 41 show the surface temperature, soil temperature and moisture at 2 cm observed during MicroWEX-3, respectively. Figure 2-40 shows th e surface and soil temperatures at 2 cm observed by thermistor during MicroWEX3. Figure 2-41 shows the volumetric soil moisture content (VSM) at 2 cm obser ved by the TDR probe during MicroWEX-3.

PAGE 55

38 The data collected during vegetation sampling during MicroWEX-3 included canopy height and width, LAI, wet and dry we ight of canopy components, leaves, stems, and bolls. Figure 2-42 to 2-44 show the ca nopy height, width, and LAI observed from four vegetation sampling areas during Micr oWEX-3, respectively. Figure 2-45 and 46 shows the total biomass and PWC observed during MicroWEX-3, resp ectively. Figure 247 to 50 show the biomass for each compone nt during MicroWEX-3. Figure 2-51 to 54 show the PWC for each component during Mi croWEX-3. Table 2-3 lists the growth stages of the cotton planted during MicroWEX -3. Figure 2-55 shows pictures of growth stages of cotton. The detailed information a bout the growth and development of cotton can be found in Wright and Sprenke l (2005) and Ritchie et al. (2004). Table 2-3 Growth stages of the cotton observed during MicroWEX-3. Growing Stage Occurring Time (DOY in 2004) Planting 191 Emergence 195 Square formation 234 50 % Bloom 253 75 % Bloom 258 10 % Boll formation 272 40 % Boll formation 281 50 % Boll formation 288 Open boll 314 200 210 220 230 240 250 260 270 280 290 300 310 120 140 160 180 200 220 240 260 280 300 320 TB (K)DOY in 2004 (EST) Vpol Hpol Figure 2-39. Observed brightne ss temperatures at vertically (V-) and horizontally (H-pol) polarizations during MicroWEX-3.

PAGE 56

39 200 210 220 230 240 250 260 270 280 290 300 310 280 290 300 310 320 Temperature (K)DOY in 2004 (EST) Surface 2 cm Figure 2-40. Surface and soil temperatures at 2 cm observed by thermistor during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.1 0.2 0.3 0.4 VSMDOY in 2004 (EST) 2 cm Figure 2-41. Volumetric soil moisture cont ent (VSM) at 2 cm observed by the TDR probe during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.2 0.4 0.6 0.8 1 1.2 Hc (m)DOY in 2004 (EST) NW NE SW SE Figure 2-42. Canopy height observed during MicroWEX-3.

PAGE 57

40 200 210 220 230 240 250 260 270 280 290 300 310 0 0.2 0.4 0.6 0.8 1 Wc (m)DOY in 2004 (EST) NW NE SW SE Figure 2-43. Canopy width obs erved during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 2 2.5 3 LAIDOY in 2004 (EST) NW NE SW SE Figure 2-44. LAI observed during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 2 2.5 3 Biomass (kg/m2)DOY in 2004 (EST) NW NE SW SE Figure 2-45. Green biomass observed during MicroWEX-3.

PAGE 58

41 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 2 PWC (kg/m2)DOY in 2004 (EST) NW NE SW SE Figure 2-46. Plant water content (PWC) observed during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 Biomass (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-47. Green biomass observed from NW area during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 Biomass (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-48. Green biomass observed from NE area during MicroWEX-3.

PAGE 59

42 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 Biomass (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-49. Green biomass observed from SW area during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.5 1 1.5 Biomass (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-50. Green biomass observed from SE area during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.2 0.4 0.6 0.8 1 PWC (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-51. Plant water c ontent (PWC) observed from NW area during MicroWEX-3.

PAGE 60

43 200 210 220 230 240 250 260 270 280 290 300 310 0 0.2 0.4 0.6 0.8 1 PWC (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-52. Plant water c ontent (PWC) observed from NE area during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.2 0.4 0.6 0.8 1 PWC (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-53. Plant water c ontent (PWC) observed from SW area during MicroWEX-3. 200 210 220 230 240 250 260 270 280 290 300 310 0 0.2 0.4 0.6 0.8 1 PWC (kg/m2)DOY in 2004 (EST) Leaves Stems Bolls Figure 2-54. Plant water c ontent (PWC) observed from SE area during MicroWEX-3.

PAGE 61

44 Figure 2-55. (a) Square, (b) bloom, (c ) boll, and (d) open boll of cotton.

PAGE 62

45 CHAPTER 3 RADIOMETRIC CALIBRATION FOR C-BAND MICROWAVE RADIOMETERS In this chapter, the calibration of UF CMR is described using widely used calibration techniques for ground-based C-band radiometers. The accuracy and errors associated with the UFCMR observations are also quantified. Introduction Ground-based microwave radiometers have been used extensively to measure upwelling terrain emission in the field e xperiments for hydrology, agriculture, and meteorology (Jackson and O’Neill, 1990; Jackso n et al., 1997; Judge et al., 2001; Shi et al., 2002; Lemaitre et al., 2004; Schneeberge et al., 2004; Memmo et al., 2005). The total power radiometer is of the simplest designs compared to other designs such as the Dicke and noise injection (Ulaby et al., 1981; S kou, 1989). The stability and consistency of the relation between the output voltage and the an tenna temperature, i.e. system gain and offset, are critical for radiometer operations. The system gain is highly sensitive to fluctuations in the physical temperature inside the radiometer requiring frequent calibration during the radiometer operation for reliable and accurate observations. Many calibration techniques have been de veloped for microwave radiometers for space-borne and air-borne (Njoku et al., 1980; Ruf, 2000; Ruf and Li, 2003; Ruf et al., 1994; Ruf et al., 1995; Corbella et al., 2002; Bonnefond et al., 2003) and ground-based radiometers (Han and Westwater, 2000; Al-A nsri et al., 2002; Cimini et al., 2003; Deuber et al., 2004; Corbella et al., 2005; Pham et al., 2005; Goodberlet and Mead, 2006). In general, calibration techniques include obser vations of radiometer output voltages for

PAGE 63

46 cold and hot targets with known brightne ss temperatures (Ula by et al., 1981; Skou, 1989). For the radiometers operating at low fr equencies away from the water vapor and oxygen absorption bands, such as C-band (6.7 GHz), commonly used cold targets are liquid nitrogen or the sky. Hot targets incl ude microwave absorbers or matched load inside the radiometers. For a C-band groundbased microwave radiometer, the simplest calibration technique using the microwave absorb er at ambient temperature as a hot target is called “external calibration” (EC). Another widely used calibration technique which utilizes the sky measurements at different a ngles to calculate the optical depth of the atmosphere and the brightness temperature of the sky is called “tipping curve calibration” (TC) (Han and Westwater, 2000; Cimini et al ., 2003; Deuber et al., 2004). Either EC and TC can be used exclusively, or TC can be used to provide better estimate of the sky measurement for EC. Both these techniques ar e inconvenient and labor intensive to be performed frequently for long-term soil mo isture studies using C-band ground-based radiometers. Moreover, the utility of TC at C-band can be hampered by the high atmospheric transparency at low microwav e frequencies (Ulaby et al., 1981). Another technique, “internal calib ration” (IC), uses an internal matc hed load as the hot target. This technique has been used for space-borne mi crowave radiometers, e.g. SMMR (Njoku et al., 1980), TMR (Ruf et al., 1994; Ruf et al ., 1995), and JMR (Bonnefond et al., 2003), airborne radiometers (Corbe lla et al., 2002), and groundbased radiometers (De Roo, 2003; Pham et al., 2005; Goodberlet and Mea d, 2006). Unlike EC and TC, IC can be performed faster than gain fl uctuations. Furthermore, IC is neither sensitive to operator technique, nor to weathering of the delicat e microwave absorber, nor does it require any additional hardware exclusively for the pur pose of calibration. However, IC does not

PAGE 64

47 account for the losses in the antenna and tran smission lines before the internal switch used to observe the matched load. In this section, the performa nce of IC is quantified and validated using EC and TC for long-term observations of soil moisture using ground-based C-band radiometers. The ground-based total power radiometer with si milar design to the UFCMR, the C-band unit on the Truck Mounted Radiometer System 3 (TMRS-3); and the ca libration experiments conducted in Alaska are described. Thr ee different calibration techniques are summarized, and the consistency of the cal ibration among these techniques during the experiments are compared. The absolute a ccuracy of the brightness observations are evaluated by comparing the observed bright ness temperatures of a lake with those obtained using a lake emission model. C-band Radiometers The UFCMR and TMRS-3C were both developed by the Microwave Geophysics Group at the University of Michigan (UMMGG). Both are dual-polarized, unbalanced total power radiometers operating at the cen ter frequency of 6.7 GHz, near the frequency of the Advanced Microwave Scanning Radiometer – EOS (AMSR-E) aboard the NASA Aqua Satellite. The UFCMR is mounted on a 10 m tower, whereas the TMRS-3C is mounted on the hydraulic arm of a Norstar truck, that can extend to 12 m. TMRS-3 consists of a suite of dual-polarized radiometers operating at 1.4, 6.7, 19, and 37 GHz mounted on an elevation positioner which allo ws for approximately 300 rotation in the elevation axis. A major difference between the UFCMR and TMRS-3C designs is the use of two receivers for Vand H-pol in TMRS -3C, compared to only one receiver in the UFCMR that switches between the two polarizat ions. Table I lists the specifications of the C-band radiometers.

PAGE 65

48 Field Experiments Microwave Water and Energy Balance Experiments (MicroWEXs) MicroWEXs were conducted by the Center for Remote Sensing, Department of Agricultural and Biological Engineering, Univ ersity of Florida, at the Plant Science Research and Education Unit (PSREU), IFAS, Citra, FL during the growing seasons of cotton (MicroWEX-1 (Tien et al., 2005) and -3 (Lin et al., 2005)) and corn (MicroWEX2 (Judge et al., 2005)). During the MicroW EXs, the UFCMR measured microwave brightness temperatures ever y 15 minutes and was calibrated every two weeks. We conducted 10, 4, and 11 calibrations during the 1 40, 80, and 190 days of the MicroWEX1, 2, and 3, respectively. Each cal ibration included meas urements of sky at zenith angles of 15, 30, 45, and 60; of a microwave absorber at ambient temperature; and of a matched load inside the radiometer. The Tenth Radiobrightness and Ener gy Balance Experiment (REBEX-10) REBEX-10 was conducted by the UM-MGG from May 6 to July 1, 2004, at a site about 1 km north of Toolik Field Station on the North Slope of Al aska. In addition to conducting twice daily EC calibrations duri ng REBEX-10, validation data was obtained by driving the Norstar truck to a beach on th e NE shore of Toolik Lake, and extending the radiometer systems over the open wa ter on June 21 (DOY 173) and 22 (DOY 174). The boom was extended to the West from the s hore, in the direction of the smallest solid angle of land presented at the opposite shore of the lake. The calibration targets included sky, absorber, and lake surface. The sky measur ements were recorded at zenith angles of 0, 10, 23, 30, 32, 40, and 55. The lake surface measurements were obtained at incidence angles of 23, 30, 32, 40, and 55. The lake temperature was measured once

PAGE 66

49 on DOY 173 at 1502 hrs (AKDT) to be 13.7 C and on DOY 174 at 0342 hrs (AKDT) to be 10 C. These are expected to be extr eme lake temperatures during this period. Calibration Methodology The relationship betwee n the output voltage ( Vout) and the antenna apparent temperature ( T'B) of a total power radiometer with a square-law detector such as the UFCMR and the TMRS-3C, can be expressed as follows: I V S Tout B (3.1) where S and I are the slope and intercept of the calibration curve, respectively. External Calibration The calibration targets of EC included th e microwave absorber at ambient air temperature and the sky measurement at ze nith angle of 15 and 0 for UFCMR and TMRS-3C, respectively. The S and I are abs out sky out abs ant sky ant abs sky BV V T T T T S, , , , ,) 1 ( ) ( ) ( (3.2) sky out sky ant sky BV S T T I, , .) 1 ( (3.3) where Tabs is the physical temperature of the absorber (K), is antenna efficiency, equal to 0.86 0.01, as estimated in the labora tory using one-port measurements with a network analyzer, Tant,sky and Tant,abs are the physical temperatur es of antenna during the sky and absorber measurements, respectively (K), and Vout,sky and Vout,abs are the output voltages during the sky and absorber measurements, respectively (volt). TB,sky (K) given by (Ulaby et al., 1981) is sec exp0 , , extra atm B sky BT T T (3.4) and

PAGE 67

50 0 ,sec , 0 exp sec z d z z T z Ta atm B (3.5) where Textra is the extraterrestrial brightness temperature (K) which is ~2.7 K, 0 is the total zenith opacity (Np), is the zenith angle, a is the atmospheric absorption coefficient (Np/m), T is the temperature profile (K), and (0, z ) is the optical thickness of the atmosphere between the surface and height z (Np). Given the atmospheric temperature, pressure, and water vapor dens ity, the sky brightness temperature can be calculated based on the 1962 U.S. Standard At mosphere (Ulaby et al., 1981). At C-band, the sensitivity of sky brightne ss to changes in atmospheric conditions can be ignored due to the high atmospheric transp arency (Ulaby et al., 1981). The sources of error using EC include the measurement errors due to the antenna sidelobes, sl, the insertion loss variability of the radiometer switches, sw, and the uncertainty in the physical temperat ure measurements of the absorber, at. The effect of these errors using UFCMR and TMRS-3C will be discussed in later section. Tipping-Curve Calibration TB,sky can be obtained by TC assuming a horizontally stratified atmosphere (Janssen, 1993) and (Han and Westwater, 2000) as A T A T Tatm extra sky Bexp 1 exp, (3.6) where A is the airmass at observation angle ( ), is the atmospheric opacity (Np), and Tatm is the mean atmospheric temperature (K). In a plane-stratified atmosphere, airmass A is defined as (Han and Westwater, 2000): sin 1 A (3.7)

PAGE 68

51 For UFCMR and TMRS-3C, the antenna temperature is linearly related to the output voltage such that rec abs A abs A sky A ofst abs out abs out sky outT T T T V V V V , , , , , , (3.8) where Vofst is the system offset voltage when the system input noise temperature is zero K ( Tsys = T'A + Trec), TA,sky and TA,abs are the apparent antenna temperatures for the sky and absorber measurements, respectively (K), and Trec is the receiver noise temperature (K). The equations for S and I are the same as (3.2) and (3.3), with TB,sky estimated by the radiative transfer equation using the leas t-square technique from 0 to 45. The atmospheric temperature was approximated by the air temperature at the earth surface (Han and Westwater, 2000). The sources of erro r using TC include sl, sw, and at, similar to those in EC. Internal Calibration IC uses an internal matched load or a fixed temperature source inside the radiometer as the hot target. The cold targ et is the sky measurement at 15 for UFCMR and at 0 for TMRS-3C. The S and I using IC are cal out sky out cal sky ant sky BV V T T T S, , , ,) 1 ( (3.9) cal out calV S T I, (3.10) where Tcal is the physical temperature of the matched load (K), is the antenna efficiency, Vout,cal is the output voltage at the matched load at Tcal (volt), and TB,sky is estimated, similar to EC (Ulaby et al., 1981).

PAGE 69

52 The sources of error using IC include the sl and sw, similar to those in EC, the error due to the uncertainty in the antenna efficiency estimation, ae. and the uncertainty in the physical temperature measur ements of the internal load, lt. Lake Emission Model (LEM) For an open, calm water surface, the brightness temperature observed by a microwave radiometer can be modeled as: water p sky B p p BT T T 1, , (3.11) where p is the reflectivity at polarization p and Twater is the physical temperature of water (K). TB,sky is ~5 K for C-band. The reflectivity of the specular water surface is determined by the incidence angle (Rose et al., 2002) and the dielectric constant of the water. The empirical dielectric models for pure wate r can be found in Ulaby et al. (1986) and Meissner and Wentz (2004). Results and Discussion The calibration data from MicroW EXs and REBEX-10 provided a unique opportunity to compare the performance of two C-band radiometers with similar design in different environmental conditions. Duri ng each experiment, the radiometers were maintained at constant temperatures with 0. 1 K standard deviation at the RF circuitry. Table II shows the means and the standard de viations of the calibra tion curves at H-pol during MicroWEXs and REBEX-10. These incl uded 25 data points during MicroWEXs, as well as 80 points for EC and IC, and 2 points for TC during REBEX-10. IC produced the most consistent calibration curves in terms of the lowest standard deviations of the slopes, although the differences among the ca libration techniques were not statistically significant.

PAGE 70

53 Table 3-1. Mean and standard deviation of the slopes (S) and Intercepts (I) for the H-and V-pol calibration curve during Micr oWEX-1 (MW-1) and REBEX-10 (RB10). EC IC TC Unit: S [K/v] and I [K] Mean Std. Mean Std. Mean Std. S H-pol MW-1 189.37 3.81 188.62 2.00 182.48 3.12 S H-pol RB-10 152.35 3.68 152.62 0.84 150.34 3.40 I H-pol MW-1 -131.10 5.37 -130.37 3.66 -116.24 6.58 I H-pol RB-10 -69.18 5.38 -69.47 4.23 -65.70 6.68 S V-pol MW-1 175.52 5.06 174.56 6.94 167.04 6.38 S V-pol RB-10 157.36 4.92 156.41 3.37 152.05 1.94 I V-pol MW-1 -99.56 14.92 -98.84 16.67 -81.40 14.70 I V-pol RB-10 -19.52 2.64 -19.00 2.31 -17.28 1.78 Fig. 3-1 to 3-4 show the gain fluctu ations of the calibration curves for MicroWEXs and REBEX-10 at H-pol. Simila r results also were found for V-pol. The mean absolute difference (MAD) between the slopes of EC and IC was 2.8 K/volt during MicroWEXs. Figure 3-1. Slopes for the calibration curve at H-pol during MicroW EX-1. RMSE of EC = 1.20, TC = 1.84, and IC = 1.10 K/volt.

PAGE 71

54 Figure 3-2. Slopes for the calibration curve at H-pol during MicroW EX-2. RMSE of EC = 1.12, TC = 1.43, and IC = 1.02 K/volt. Figure 3-3. Slopes for the calibration curve at H-pol during MicroW EX-3. RMSE of EC = 1.12, TC = 1.43, and IC = 1.02 K/volt.

PAGE 72

55 Figure 3-4. Slopes for the calibration curve at H-pol during REBEX-10. RMSE of EC = 1.20, TC = 1.84, and IC = 1.10 K/volt. The difference between the slopes of TC and EC was 4.4 K/volt; and the difference between the slopes of TC and IC was 3.6 K/volt during MicroWEXs. The MADs for REBEX-10 were not calculate d because there were only two TC measurements. During MicroWEXs, the EC and IC calibration curves were closer to each other, while TC produced slightly dissimilar results from EC and IC. This was primarily because at C-band, TC is based on multiple measurements with small differences in brightness temperatures (TB) of the sky, compared to measurements at higher frequencies at which the differences are larger. Due to th e high atmospheric transparency, the utility of TC at C-band was reduced. Applying the calibration curves over the output voltages for the terrains observed duri ng MicroWEXs and REBEX-10, the MAD of the calibrated TB using the three calibration techniques wa s 1.14 K. Table 3-2 gives the root-meansquare errors (RMSE) estimates in the accuracy of observed TB using UFCMR and

PAGE 73

56 TMRS-3C due to ae, sl, sw, at, and lt as mentioned in section IV. The RMSE of EC, TC, and IC are 1.20, 1.84, and 1.10 K/volt, respectively. Table 3-2. RMSE estimation for the opera tion and calibration for UFCMR and TMRS3C. Symbol Source Value (K) ae Antenna Efficiency < 0.14 Antenna Sidelobe (Incidence angle 0 to 45), H-pol 1.06 to 1.75 sl Antenna Sidelobe (Incidence angle 0 to 45), V-pol 0.97 to 1.32 sw Radiometer Switch < 0.25 at Absorber Temperature Measurements < 0.50 lt Load Temperature Measurements < 0.10 To assess the accuracy of the calibration, the observed TB of a calm lake at different incidence angles between 20 to 55 during REBEX-10 was used to compare with the LEM modeled TB. The RMSE due to antenna sidelobe during the lake observations decreases slightly as the incidence angle increases. Figure 3-5 illustrates the RMSE due to antenna sidelobe during lake observation. The RMSE due to antenna sidelobes during the lake observations were 0.28, 0.26, 0.25, 0.25, and 0.24 K at H-pol whereas those were 0.47, 0.39, 0.37, 0.31 and 0.26 K at V-pol at the incidence angles of 23, 30, 32, 40, and 55, respectively. Th ese RMSE are included in the error bars shown in Figure 3-6 and 3-7. Observed TB were compared with those of a smooth water surface simulated by LEM (Figure 3-6 to 3-7). Figure 3-5. RMSE due to antenna side lobe during the lake observations

PAGE 74

57 Figure 3-6. Comparison of the brig htness temperature differences ( TB) by observations and LEM at H-pol on DOY 173. The total RMSE in observations and simulation are 1.46, 2.02, and 1.39 for EC, TC, and IC, respectively. Figure 3-7. Comparison of the brig htness temperature differences ( TB) by observations and LEM at H-pol on DOY 174. The total RMSE in observations and simulation are 2.58, 2.73, and 2.54 for EC, TC, and IC, respectively.

PAGE 75

58 Two dielectric models for pure water, Ulaby et al. (1986) and Meissner and Wentz (2004) were used. Figure 3-8 shows the dielectric constant of pure water simulated by two models at C-band. At C-band the MAD between the simulated LEM TB by two models were insignificant at ~0.0095 K. Th e uncertainty in the water temperature measurement was 3.0 K resulting in an RMSE of 0.8 and 2.3 K at Hand V-pol in the simulated TB, respectively. The MAE between the observed and modeled TB at H-pol were 2.5 .46, 3.9 .02, and 2.4 .39 K for EC, TC, and IC, respectively. The MAE between the observed and modeled TB at V-pol were 1.3 .58, 2.4 .73, and 1.3 .54 K for EC, TC, and IC, respectively. For soil mo isture applications, an accuracy of about 2 K at C-band is adequate (Calvet et al., 1996). Figure 3-8. Dielectric consta nt of pure water simulate d by Ulaby et al. (1986) and Meissner and Wentz (2004).

PAGE 76

59 Conclusion The calibration experiments during the MicroWEXs and REBEX-10 were designed to assess the calibration consistenc y of two C-band radiometers with similar design. The three most widely used techni ques, EC, TC, and IC, were compared to understand their performance for long-term soil moisture studies using ground-based Cband radiometers. Even though IC produced th e most consistent calibration curves, the differences among the three calibration techni ques were not statis tically significant. Applying the calibration curves over th e output voltages observed during the MicroWEXs and REBEX-10, the MAD of the TB calibrated among th e three calibration techniques was 1.14 K. This chapter provides results to the fi rst research question in Chapter 1: “What are the accuracy and precision of the TB measurements used in this study?” as follows: The absolute accuracy of calibration techni ques was investigated by comparing the observed and modeled TB of a calm lake. The MAE between the observed and modeled TB at H-pol were 2.5 .46, 3.9 .02, and 2.4 .39 K for EC, TC, and IC, respectively. The MAE between the observed and modeled TB at V-pol were 1.3 .58, 2.4 .73, and 1.3 .54 K for EC, TC, and IC, respectively. Due to the high atmospheric transparency, the utility of TC at C-band is greatly reduced. Because IC was found to have an MAE of ~2 K that is suitable for soil moisture applications and was consistent during our experiments under significantly different en vironmental conditions, it can be used to augment less frequent calibrations obta ined by the EC or TC techniques.

PAGE 77

60 CHAPTER 4 MICROWAVE BRIGHTNESS MODEL FOR COTTON In this chapter, the Microwave Brightness for Cotton (MB-Cotton) model developed to simulate the brightness temperature (TB) for the whole growing season of cotton is described. Microwave Brightness Model for Cotton (MB-Cotton) Overall terrain emission of cotton is estimated as a linear combination of bare soil and vegetation canopy TB because the footprint is a mixt ure of bare soil and vegetation emission for row crops such as cotton (Figure 4-1): p C B c p S B c p BT C T C T, , , , ,1 (4.1) where p is the polarization, Cc is the fraction of canopy cover, TB,S,p and TB,C,p are the bare soil and canopy components, respectively. Figure 4-1. Schematic of the micr owave brightness model for cotton.

PAGE 78

61 The Cc is the ratio of canopy width to row width. r c cw w C (4.2) Bare Soil Component (TB,S,p) The soil is modeled as a semi-infinite, horizontally uniform, and smooth surface medium. The bare soil component consists of emission from the atmosphere and soil layers as (Figure 4-1) p eff p sky B p S BT T T 1, , , (4.3) where TB,sky is the downwelling atmospheri c brightness temperature (K), p is the reflectivity of the soil at polarization p, and Teff is the effective phys ical temperature of the soil (K). At C-band, TB,sky is ~5 K. The reflectivity at vertical ( v) and horizontal ( h) polarizations are given by Fresne l equations (Ula by et al., 1981): 2 2 2sin cos sin cos ) ( soil soil soil soil v (4.4) 2 2 2sin cos sin cos ) ( soil soil h (4.5) where is the incidence angle from zenith (see Figure 4-1), and soil is the dielectric constant of the wet soil. Teff is estimated using Radiative Transfer equation (Ulaby et al., 1981) as (Figure 42): 0 , ,, 0 exp , 0 expeff effZ s s e eff s Z soil effz d z z T z Z T T (4.6)

PAGE 79

62 where Tsoil,Zeff is the soil physical temperature at the depth of effective depth (Zeff), s(0,Zeff) is the optical depth of th e effective soil layer (Np), e,s is the power extinction coefficient of the soil medium (Np/m), and T is the physical temperat ure of the soil (K). Figure 4-2 shows the schematic of the Teff approximation. Figure 4-2. Schematic of Teff approximation. The effective depth (Zeff) of soil emission is approximated by the penetration depth of soil ( p), which is calculated as (Ulaby et al., 1982) as: 1 0Im 4soil p (4.7) where Im represents the imaginary part of the quantity, 0 is the free-space wave length and soil is dielectric constant of the wet soil. The soil is modeled as a semi-infinite medium, with layered constitutive properties. The dielectric properties within a layer are assumed to be constant. The multiple reflections between the soil layers and volume scattering within the soil layers were a ssumed to be zero. Figure 4-3 shows the penetration depth, i.e. Zeff as a function of volumetric soil moisture content (VSM). Because the values of observed VSM during th e field experiments were between 5 and 35 % (Figure 2-17, 18, and 40), the effective depth at C-band for MB-Cotton model was determined to be 2 cm (Figure 4-3). Teff was calculated by integr ating the emission from layered soil based upon thei r dielectric and thermal profiles from 0 to 2 cm.

PAGE 80

63 5 10 15 20 25 30 3 5 0 1 2 3 4 VSM (m3/m3)Zeff (cm) Figure 4-3. Penetration depth at C-band as a function of vol umetric soil moisture content using Equation 4.7. The power extinction coefficient was equal to the absorption coefficient of the soil medium because the volume scattering with in soil layers was assumed to be zero, e,s = a,s (Np/m) calculated as (Ulaby et al., 1981), 2 1 2 1 2 01 1 2 4 r r r r s a (4.8) where 0 is the free-space wavelength, r is the relative permeability, 'r and "r are the real and imaginary part of th e complex dielectric constant. s is defined as (Ulaby et al., 1981): z z s a sdz z z z , (4.9) At C-band, the dielectric prope rty of soil layer can be tr eated as dielectric mixture including soil particles, air, bound water, and free water usin g the soil dielectric model developed by Dobson et al. (1985). fw fw bw bw a a s s soilv v v v (4.10)

PAGE 81

64 where is the dielectric constant, v is the volumetric content, and the subscripts s, a, bw, and fw denote soil particle, air, bound water, and free water, respectively, and is the adjustable factor ( = 0.65 from Ulaby et al., 1986). For the soil of particle density around 2.65 g/cm3, the dielectric constant are s = 4.7, a = 1, bw = 3.3 (Dobson et al., 1985). The dielectric constant of free water is simulated by the Debye equation (Ulaby et al., 1986). f jfw2 10 (4.11) where and 0 are the high-frequency lim it and static dielectric constants of pure water, f is the electromagnetic frequency (Hz), is the relaxation time of pure water (s), and j is 1 . The real and imaginary parts of the pure water dielectric constant can be written as: 2 02 1 ffw (4.12) 2 02 1 2 f ffw (4.13) The high-frequency limit dielectric constant of pure water is found to be a constant, = 4.7 (Lane and Saxton, 1952) whereas the sta tic dielectric constant of pure water is a function of the water physical temperature, Tw (C) and simulated by (Klein and Swift, 1977) 3 5 2 4 010 075 . 1 10 295 . 6 4147 . 0 045 . 88w w wT T T (4.14) The soil surface is modeled as a smooth-surfa ced dielectric layer whose reflectivity at vertical ( v) and horizontal ( h) polarizations are given by Fresnel equations (Ulaby et al., 1981).

PAGE 82

65 Recently, Mironov et al. (2004) developed a new dielectric model for soil in the microwave band from 0.6 to 18.0 GHz. The key point of their soil dielectric model is to use the complex dielectric constant directly as a function of volumetric soil moisture. The model also provided a more sophisticated method to account for the values of bound water fraction and dielectric constant. The soil complex dielectric constant values modeled by Mironov et al. (2004) model matc h well with the soil complex dielectric constant modeled by the Dobson et al. (1985) model. The comparison was done on silty clay soil with texture are of 5.02 % sand, 47.60 % silt, and 47.38 % clay. This soil is different from the soils at the study site with 92 % sand. The difference between the dielectric constants simulate d by two models for the sandy soils at study site should be even smaller. At 6.7 GHz, the differences between the complex dielectric constants predicted by the two models were ~1.0 for real and imagery parts between soil volumetric moisture content of 10 and 30 %. Vegetation Component (TB,C,p) The canopy is modeled as dielectric cloud a bove the soil of width and thickness as canopy width and height, respectively. The canopy component, TB,C,p in Equation 4-1, can be expressed as (Figure 3-1): Canopy p B Soil p B Sky p B p C BT T T T, , , , , (4.15) where TB,p Sky, TB,p Soil, and TB,p Canopy are the terms for sky, so il, and vegetation emission, respectively. The sky emission term account s for downwelling atmospheric brightness temperatures (TB,sky), attenuated through the canopy, reflected at the canopy-soil interface, and attenuated through the canopy again. c p sky B Sky p BT T2 exp, , (4.16)

PAGE 83

66 where c is the optical depth of the vegeta tion canopy (Np). The soil emission term accounts for the contribution from the soil br ightness temperature emitted at the canopysoil interface and attenuated by the canopy. c p eff Soil p BT T exp 1, (4.17) TB,p Canopy accounts for the emission from the vegetation canopy in the upward and downward directing wi thin the canopy. c p c canopy Canopy p BT T exp 1 exp 1, (4.18) where Tcanopy is the effective canopy physical temperature (K) and c is the optical depth of the canopy (Np) calculate d as (Ulaby et al., 1981): z z c e cdz z z z,, (4.19) where e,c is the power extinction coefficien t of the vegetation canopy (Np/m). The vegetation canopy is trea ted as a two-component mi xture including vegetation materials and air as host. Assuming there is no volume scattering in the vegetation canopy, the power extinction coefficient can be calculated using Equation (4.8). The dielectric constant of the vegetation canopy is modeled using the Dual-dispersion model (Ulaby and El-Rayes, 1987). 5 . 018 . 0 1 0 . 55 9 . 2 86 . 22 18 1 0 . 75 9 . 4 jf v f j jf vbw fw r c (4.20) where r is the non-dispersive residual compone nt, f is the electromagnetic frequency (GHz), and vfw and vbw are the free and bound water frac tion in the vegetation canopy. These variables are empirically related to the gravimetric moisture content of the vegetation canopy (Mg) (Ulaby and El-Rayes, 1987).

PAGE 84

67 216 . 6 74 . 0 7 . 1g g rM M (4.21) 076 . 0 55 . 0 g g fwM M v (4.22) 2 236 . 7 1 64 . 4g g bwM M v (4.23) The canopy dielectric properties were deri ved assuming the canopy is composed of vegetative materials only. This might lead to an overestimation in the canopy attenuation. Empirical dielectric models, such as E ngland and Galantowicz (1995), estimating canopy dielectric constant based upon the volume fraction of wet vegetative materials may provide a better approximation. Summary In this chapter, the MB-Cotton model wa s developed based upon radiative transfer equation to simulate the brightness temperat ures for a growing season of cotton. MBCotton model simulated overall terrain emi ssion as a mixture of bare soil and canopy emissions. The proportion of canopy emission is determined by the fraction of canopy cover whereas the rest of the emission is contributed by the bare soil emission. The soil is modeled as a semi-infinite, horizontally uni form, and smooth surface medium. The soil at the effective depth is divided into three layers at the depths of 0.5, 0.5, and 1.0 cm from the surface. The dielectric property of the so ils is modeled using literature-based model (Dobson et al., 1985). The cotton canopy is mode led as a rectangular dielectric cloud of width and thickness as canopy widt h and height. The dielectric property of the canopy is modeled using empirical model developed for vegetation constituents. This chapter provides results to the se cond research question in Chapter 1, “What is the effective depth of C-band un der different soil moisture conditions?” as follows:

PAGE 85

68 For sandy soil under the soil moisture c onditions (5 to 35 %) observed during MicroWEXs, the effective depth of C-band is from 2.3 (5 %) to 0.7 cm (35 %).

PAGE 86

69 CHAPTER 5 MODEL EVALUATION In this chapter, the evaluation of the MB -Cotton model developed in Chapter 4 for a growing season of cotton is described. The gr owing season is divided into three stages, viz. early, mid, and late, based upon the phe nological stages of cotton ranging from germination to mature canopy. The input variables for the MB-Cotton model are discussed, including the vegetation properties, surface temperature, and soil moisture and temperature profiles simulated by the Land Surface Process (LSP) model and observed soil moisture and temperature during MicoWEX-3. Finally, the TB simulated by the MBCotton model are compared to those obs erved for the dry down periods during MicroWEX-3 to evaluate the results. Input Variables The forcing for the MB-Cotton model can be categorized into soil and vegetation variables. Two simulations were conducted; th e first using input soil variables from field observations from MicroWEX-3 and second us ing input soil variable s from estimations by an SVAT model called, Land Surface Proces s (LSP) model. The field observed soil variables included surface thermal infrared te mperature, volumetric soil moisture (VSM) and temperature at depth of 2 cm (see Fi gure 2-40 and 41). The LSP model estimated variables included surface temperature at 0.05 cm, VSM and temperature profiles at depths of 0.25, 0.76, and 1.26 cm (Figure 51). A brief description of the LSP model and its preliminary calibration results used in this dissertation are provided below.

PAGE 87

70 Figure 5-1. Soil profiles es timated by the LSP model. Input Variables for Soil MicroWEX-3 field observations Figure 2-40 and 41 shows the thermal infr ared temperature, volumetric soil moisture (VSM) and temperature at depth of 2 cm for the input variables for soil to the MB-Cotton model. Land Surface Process (LSP) model The LSP model was developed by th e Microwave Geophysics Group at the University of Michigan to simulate c oupled one-dimensional energy and moisture transport at the land surface and in the vadose zone when forced with observed weather (Liou and England, 1998; Judge et al., 1999; Judge et al., 2003). The governing equations for the energy and mo isture balance are given as (Judge et al., 2003): m mq t X (5.1) e eq t X (5.2)

PAGE 88

71 where Xm and Xe are the total moisture and ener gy contents per unit volume, (kg/m3 and J/m3), respectively, and qe and qm are the moisture and energy fluxes (kg/sm2 and J/sm2), respectively. The soil profile is defined with layers of different thermal and hydraulic properties. The thickness of each soil layer increases e xponentially with depth. A block-centered, forward-time finite difference scheme is used to calculate the moisture and temperature profiles. The model forcings include mi crometeorological parameters, e.g. air temperature, relative humidity, downwe lling shortand long-wave radiation, irrigation/precipitation, and wind speed. The initial conditions of the temperature and moisture profiles are obtained from the MicroWEX-3 observations. The upper boundary conditions for moisture and energy fluxes ar e given by the energy and moisture balance at surface as: R E E D z qtr s c l m 0 (5.3) s s ns eL H R z q 0 (5.4) where 0 z qm and 0 z qe are the moisture and ener gy fluxes at the upper boundary, l is the density of liquid water (kg/m3), Dc is the rate of drainage (= total precipitation or irrigation – interception by the canopy) from the canopy (m/s), Es is the rate of evaporation from the soil (m/s), Etr is the rate of transpiration from the soil (root zone) (m/s), R is runoff (m/s), Rns is the net radiation absorbed by the soil (W/m2), and Hs and Ls are the sensible and latent heat fl uxes from the soil, respectively (W/m2). Moisture and energy fluxes at the lowest boundary conditions are set to gravity drainage and uniform energy flux, respectively. Tabl e 5-1 shows the physical and hydraulic properties of soil and Table 5-2 shows the canopy parameters us ed in the LSP model (Yan et al., 2006).

PAGE 89

72 Figure 5-2 and 5-3 show the temperature a nd moisture profiles estimated by the LSP model during MicroWEX-3, respectively. Table 5-1. The physical and hydraulic proper ties of soil used in the LSP model (Yan, 2006). Parameter Description Value zob Bare soil roughness length 0.01 (m) s Soil IR emissivity 0.98 Pore-size index 0.5 0 Air entry pressure 0.076 (of m H2O) Ksat Saturated hydraulic conductivity 1 10-4 (m/s) r Wilting point 0.0051 (m3/m3) sa Sand fraction 0.894 (m3/m3) si Silt fraction 0.034 (m3/m3) c Clay fraction 0.071 (m3/m3) o Organic fraction 0.0 (m3/m3) Porosity 0.34 (m3/m3) Table 5-2. The canopy parameters used in the LSP model. Parameter Description Value x Leaf angle distribution parameter 1.00 Leaf reflectance 0.3 c Canopy emissivity 0.98 cd Canopy drag coefficient 0.15 iw Canopy wind intensity factor 0.60 lw Leaf width 0.05 (m) Fb Base assimilation rate -0.2 10-7 (kgCO2/m2s) photo Photosynthetic efficiency 11.4 10-9 200 210 220 230 240 250 260 270 280 290 300 310 280 290 300 310 320 Temperature (K)DOY in 2004 (EST)Surface 0.25 cm 0.76 cm 1.26 cm Figure 5-2. Surface and soil temperature profil es in the top 2 cm simulated by the Land Surface Process (LSP) model during MicroWEX-3.

PAGE 90

73 200 210 220 230 240 250 260 270 280 290 300 310 0 0.1 0.2 0.3 0.4 VSMDOY in 2004 (EST)0.25 cm 0.75 cm 1.26 cm Figure 5-3. Soil moisture profiles in the t op 2 cm simulated by the Land Surface Process (LSP) model during MicroWEX-3. Input Variables for Vegetation The vegetation variables for both the simulations were obt ained from the biweekly vegetation samplings during MicroWEX-3. Am ong the four sampling areas, the NW site was found to be the most representative to th e vegetation properties in the footprint of UFCMR. The variables in clude canopy height (Hc), canopy gravimetric water content (Mg), and canopy cover (Cc). Mg was calculated as ratio of we ight of water in the canopy to weight of dry vegetation. Cc was calculated as ratio of canopy width to row width (equation 4.2). The observations during MicroW EX-3 were interpolat ed to provide the MB-Cotton model with inputs at every fifteen minutes, matc hing the temporal frequency of the other inputs. The interpolation used statistical models in the form of t Y Y exp 1max (5.5) where Y is the interpolated variable, Ymax is the maximum value of the variable observed during the season. Ymax was equal to the average of the last four data points from the vegetation sampling during MicroWEX-3. is the regression coefficient, and t is number of days since emergence. Th e 95% confidence interval ( ) on the nonlinear least squares parameter, given the residuals and the Jacobi an matrix at the solution for Mg and Cc

PAGE 91

74 using equation (5.5) were 0.0195 and 0.0312, respectively. However, the values for Hc was small at 10-5, suggesting that the non linear fit using equation (5.5) was inadequate. To simplify the nonlinear fit, a cubic spline method was used instead (see Figure 5-4(a)). Table 5-3 shows the regression coefficients and R2 values of the empirical models for Cc and Mg. Figure 5-4 shows the observed and inter polated vegetation input variables. The values of Hc and Cc were calculated as the mean of four measurements for each sampling location, while the values of Mg were the sum of the Mg in leaves, stems, squares, and bolls. Table 5-3. Regression coefficients and Ymax values for the canopy cover (Cc) and gravimetric water content of canopy (Mg) empirical models. Ymax Cc 0.0205 0.8766 Mg 0.0337 2.2324 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 35 0 0 0.2 0.4 0.6 0.8 1 1.2 (a)Hc (m) (a) 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 35 0 0 0.2 0.4 0.6 0.8 1 (b)Cc (b) 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 35 0 0 0.5 1 1.5 2 2.5 3 (c)MgDOY in 2004 (EST) Mean (MicroWEX) Interpolation Mean (MicroWEX) Interpolation Total (MicroWEX) Interpolation Figure 5-4. Input variables for vegetation: (a) Canopy height (Hc), (b) canopy cover (Cc), and (c) gravimetric water content of cotton (Mg) observed during MicroWEX3. Error bars show one standard deviation around the mean values.

PAGE 92

75 Dielectric Properties for Soil and Vegetation The dielectric properties of the soil a nd vegetation were estimated by the fourcomponent mixing model and dual-dispersion model as shown in equation (4.9) and (4.19), respectively. Figure 42 shows the effective depth (Zeff) at C-band as a function of volumetric soil moisture content (VSM) usi ng the mixing model. Minimum soil moisture observed during simulation period was 5 % so that the maximum Zeff was ~2.0 cm for the simulation period. Evaluation Methodology The simulation period of the MB-Cott on model started on DOY 196 (July 14) and ended on DOY 314 (November 9) in 2004. The simulation was conducted during the dry down periods only, beginning at 24 hours afte r irrigation or preci pitation events. The growing season was divided into s ub-seasons based upon the phenological stages of the cotton canopy (Chapter 2). Th e early season started with germination on DOY 196 to squares formation on DOY 234. The mid season was from DOY 234 to 50 % boll formation on DOY 288. The late season was between DOY 288 to the last day of simulation on DOY 314 with 12 % of the bolls opened (Figure 2-55). As mentioned earlier, the maximum Zeff at C-band with given soil moisture field condition was 2.0 cm. Smaller values of Zeff are expected during the dry down periods when VSM was between 10 to 20 %. For example, when VSM is 20 %, Zeff is < 1.0 cm (Figure 4-2). The closest measurement of VSM to the surface was at 2 cm during MicroWEX-3. This observation measures an average VSM between 1 and 3 cm. Continuous monitoring VSM at depths < 2 cm ar e not feasible using soil moisture probes that are currently available. Because Zeff = ~2.0 cm, detailed soil moisture and temperature profiles in the top 2 cm are very crucial for accurate simulation of TB at C-

PAGE 93

76 band because of its high temporal variation. To investigate this issue further, two simulations were conducted with the MB-Co tton model. First with observed VSM and temperature at 2 cm and second with estimat ed VSM and temperature profiles in the top 2 cm by the LSP model (Figure 5-2 and 5-3). The TB from these two simulations were compared with field observed TB during MicroWEX-3 (see Figure 2-39). Results and Discussion The Early Season For the early season, the MB-Cotton m odel was run for 39 days from DOY 196 (July 14) through 234 (August 21) in 2004. The biomass and plant water content of the canopy were less than 0.4 and 0.3 kg/m2, respectively (Figure 2-45 and 2-46). The cotton was heavily irrigated and there were seven dry down periods. Table 5-4 shows the start and end times for these dry down periods. Table 5-4. The start and end times of th e dry down period in the early season during MicroWEX-3. No Start Time End Time 1 196.0 196.4 2 203.6 205.6 3 207.6 209.6 4 213.5 215.4 5 218.6 219.4 6 229.6 230.6 7 232.6 233.6 Comparison using VSM and temperature from LSP Figure 5-6(a) shows the simulated and observed TB at H-pol using LSP estimated soil moisture and temperature profiles duri ng the early season. Overall, the MB-Cotton model captured the phases of th e diurnal variation well, as seen in Figure 5-6(a). Figure 5-6(b) to (c) shows the VSM, surface temper atures, and soil temperatures simulated by the LSP model and observed during MicroWEX-3, respectively.

PAGE 94

77 196 198 200 202 204 206 208 210 212 214 216 218 220 222 224 226 228 230 232 23 4 170 190 210 230 250 270 (a) T B (K) MicroWEX MBCotton Model (LSP 0cm) 196 198 200 202 204 206 208 210 212 214 216 218 220 222 224 226 228 230 232 23 4 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 (b)VSMMicroWEX 2cm LSP 0.5cm LSP 0.5.0cm LSP 1.0.0 cm 196 198 200 202 204 206 208 210 212 214 216 218 220 222 224 226 228 230 232 23 4 292 296 300 304 308 312 316 (c)Surface Temperature (K)MicroWEX TIR LSP 196 198 200 202 204 206 208 210 212 214 216 218 220 222 224 226 228 230 232 23 4 296 300 304 308 312 316 320 (d)Soil Temperature (K)DOY in 2004 (EST)MicroWEX 2cm LSP 0.5cm LSP 0.5.0cm LSP 1.0.0 cm Figure 5-6. Comparison of th e simulated and observed TB at H-pol at 6.7 GHz using (a) 0-2 cm LSP simulated soil temperature and moisture as inputs to the MBCotton model, (b) the 2 cm observed a nd 0 to 2 cm LSP simulated VSM, (c) the observed and LSP simulated surface temperatures, and (d) the 2 cm observed and 0 to 2 cm LSP simulate d soil temperatures during the early season of MicroWEX-3.

PAGE 95

78 The TB simulated by the MB-Cotton model matched well with the observed TB during the first 11 hours of dr y down with mean absolute differences (MAD) and root mean square differences (RMSD) of 3.2 and 3.6 K, respectively (Table 5-5). During the dry down periods from DOY 203 to 219, the MB-Cotton model estimated the phases well, but the estimated TB were higher by ~15 K than the observed TB during the day. The TB from the MB-Cotton model matched well w ith observations during the night. This could be due to the underestimation of th e VSM by the LSP model during the day by ~4 % (see Figure 5-6(b)). During the last two dry downs on DOY 230 and 233, the MBCotton model matched the observed TB well during the day with low MAD and RMSD, whereas the model overestimated the observed TB by ~20 K during the night. This could be due to the underestimation in VSM by th e LSP model during the night by ~6 % during this period (see Figure 5-6(b)). Table 5-5. Mean absolute differences (MAD) and root mean square differences (RMSD) using MicroWEX-3 measurements at 2 cm (2 cm) and LSP estimations between 0 and 2 cm (0-2 cm) during the dry down in the early season during MicroWEX-3. MAD (K) RMSD (K) MAD (K) RMSD (K) No. LSP: 0-2 cm LSP: 0-2 cm MicroWEX: 2 cm MicroWEX: 2 cm 1 3.2 3.6 1.3 1.6 2 6.2 7.9 8.9 11.3 3 7.2 9.6 6.9 8.5 4 4.9 6.6 7.6 9.0 5 8.0 9.4 9.5 11.1 6 9.7 10.4 8.7 10.3 7 10.8 11.5 7.4 8.6 Comparison using observed VSM and temperature at 2 cm Figure 5-7 shows the simulated TB at H-pol using field obs erved soil temperature and moisture at 2 cm and those observed during the early season. Overall, the amplitudes of diurnal variation were significantly underestimated by the MB-Cotton model. The

PAGE 96

79 diurnal amplitudes of the observed TB were ~60, 40, and 20 K, respectively; whereas the diurnal amplitude of the TB simulated was only ~15 K. This underestimation of diurnal amplitude is primarily because of the misrep resentation of VSM profile in the effective depth, Zeff. Because Zeff < 1cm when VSM is greater than 20 % (Figure 4-3), using VSM at 2 cm as input does not provide realistic soil moisture distribution from 0 to 1 cm necessary to estimate diurnal amplitude in TB simulation to those observed (see Figure 56(b)). 196 198 200 202 204 206 208 210 212 214 216 218 220 222 224 226 228 230 232 23 4 170 190 210 230 250 270 T B (K) DOY in 2004 (EST)MicroWEX MB Model (MicroWEX 2cm) Figure 5-7. Comparison of th e simulated and observed TB at H-pol at 6.7 GHz during the early season of MicroWEX-3 using 2 cm field observed soil temperature and moisture as inputs to the MB-Cotton model. During the early season, the average valu es of MAD and RMSD between observed and simulated TB using 0 to 2 cm soil moisture and temperature estimated by LSP were 6.9 and 8.5 K, respectively; whereas av erage values of MAD and RMSD between observed and simulated TB using 2 cm soil moisture and temperature measurements were 7.5 and 9.3 K, respectively. The Mid Season For the mid season, the MB-Cotton model was run for 55 days from DOY 234 (August 21) through 288 (October 14). During th is period, formation of cotton square began on DOY 234 and bolls formation began on DOY 253. About 40 and 50 % of

PAGE 97

80 cotton bolls forming were observed on DOY 281 and 288, respectively. Figure 5-8 shows the cotton canopy at the beginning and end of this period. The vegetation properties increased significantly (Figure 2-42 to 2-45). The biomass and plant water content of the canopy components, leaves and stems incr eased from DOY 234 and achieved maximum values on ~DOY 273, whereas the biomass a nd plant water content (PWC) of squares and bolls kept increasing from ~0.4 to 1 kg/m2 (Figure 2-47 to 2-54). The crop was irrigated five times and several precipitation events occurred, resulting in eleven dry down periods in the mid season. Table 5-6 s hows the start and end times of these dry down. Figure 5-8. Cotton canopy in the early season. Photo on the left and right were taken on DOY 237 (August 24) and DOY 278 (October 4), respectively. Table 5-6. The start and end times of th e dry down period in the mid season during MicroWEX-3. No Start Time End Time 1 240.6 241.6 2 242.6 245.8 3 254.6 257.4 4 258.6 261.0 5 262.4 264.6 6 265.6 269.3 7 273.6 275.6 8 276.6 277.8 9 279.6 282.6 10 283.0 284.5 11 286.6 288.0

PAGE 98

81 Comparison using VSM and temperature from LSP Figure 5-9(a) shows the simulated and observed TB at H-pol using LSP estimated soil temperature and moisture profiles at 0 to 2 cm as model inputs during the mid season, from DOY 234 and DOY 288 when ~50 % of the bolls were formed. Overall, the phases of the TB estimated by the MB-Cotton model matched well the phases of the observed diurnal variations. During the first two dry downs, the model overestimated the TB compared to the observed TB by average of 6 K. This could be caused by an overestimation in canopy emission due to an overestimation in gravimetric water content (Mg) of the vegetation (Figure 54(c)) during this period. Mg was calculated using the total water content in the canopy without accoun ting for the distribution of moisture in different components. The effective depth of the canopy, only a few cm, is very sensitive to the optical depth ( ) of the canopy. As the cotton square formation begins, usually underneath the leaves, the emission from the canopy could be more complicated and using one effective may not be realistic. Alt hough the assumption of canopy cloud model with homogeneous dielectric prope rties was a good approximation for cotton canopy, information about the distribution of th e moisture in different components of the canopy as the reproductive stages began w ould help to improve the accuracy of simulation during this period. As bolls formation began on DOY 258, the model estimated the phases and amplitudes of the di urnal variations very well during the third dry down period as suggested by the values of MAD and RMSD of 4.0 and 5.3 K, respectively (Table 5-7). Af ter DOY 262 until the end of the mid season, the MB-Cotton model matched the observed TB well during the day by the average MAD and RMSD of 6.3 and 7.2 K, respectively. But th e model underestimated the observed TB at night by ~8 K.

PAGE 99

82 235 240 245 250 255 260 265 270 275 280 285 235 240 245 250 255 260 265 270 275 280 (a) T B (K) Squares Forming Bolls Forming (a) Squares Forming Bolls FormingMicroWEX MBCotton Model (LSP 0cm) 235 240 245 250 255 260 265 270 275 280 285 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 (b)VSMMicroWEX 2cm LSP 0.5cm LSP 0.5.0cm LSP 1.0.0 cm 235 240 245 250 255 260 265 270 275 280 285 290 294 298 302 306 310 314 (c)Surface Temperature (K)MicroWEX TIR LSP 235 240 245 250 255 260 265 270 275 280 285 292 296 300 304 308 312 316 320 (d)Soil Temperature (K)DOY in 2004 (EST)MicroWEX 2cm LSP 0.5cm LSP 0.5.0cm LSP 1.0.0 cm Figure 5-9. Comparison of th e simulated and observed TB at H-pol at 6.7 GHz using (a) 0-2 cm LSP simulated soil temperature and moisture as inputs to the MBCotton model, (b) the 2 cm observed a nd 0 to 2 cm LSP simulated VSM, (c) the observed and LSP simulated surface temperatures, and (d) the 2 cm observed and 0 to 2 cm LSP simulate d soil temperatures during the mid season of MicroWEX-3.

PAGE 100

83 Table 5-7. Mean absolute differences (MAD) and root mean square differences (RMSD) using MicroWEX-3 measurements at 2 cm (2 cm) and LSP estimations between 0 and 2 cm (0-2 cm) during the dry down in the mid season during MicroWEX-3. MAD (K) RMSD (K) MAD (K) RMSD (K) No. 0-2 cm 0-2 cm 2 cm 2 cm 1 6.3 6.6 3.2 3.5 2 6.0 7.4 2.9 3.5 3 4.0 5.3 3.2 4.0 4 2.9 4.3 5.7 5.9 5 7.3 8.4 10.0 10.3 6 6.7 7.5 9.3 9.6 7 6.1 7.4 7.1 7.6 8 6.7 7.9 8.9 9.2 9 6.1 7.0 9.3 9.5 10 6.9 8.4 7.2 7.4 11 3.9 4.6 8.2 8.3 Because Cc was ~0.6 during this period, i.e. the bare soil cover fraction was 0.4, the signal from the bare soil fraction was overe stimation by ~6 % in VSM by the LSP model (Figure 5-9(b)). Figure 5-9(c) shows the comparison between the surface temperatures observed by the thermal infrared (TIR) sensor and simulated by the LSP model. During the night, the observed surface temperatures were higher than those simulated by the LSP model by ~4 K. Comparison using observed VSM and temperature at 2 cm Figure 5-10 shows the simulated and observed TB at H-pol using field observed soil temperature and moisture profiles at 2 cm as model inputs during the mid season. Between DOY 240 and 245, the MB-Cotton mode l captured the phases and amplitudes of diurnal variation. After DOY 253, the model c onsistently underestimated the observed TB by ~15 K. This could be because the soil at 2 cm was consistently wetter than that at 0 to 1 cm during this time by ~6 % (Figure 5-9(b)).

PAGE 101

84 235 240 245 250 255 260 265 270 275 280 285 235 240 245 250 255 260 265 270 275 280 T B (K) DOY in 2004 (EST) Squares Forming Bolls Forming Squares Forming Bolls FormingMicroWEX MB Model (MicroWEX 2cm) Figure 5-10. Comparison of th e simulated and observed TB at H-pol at 6.7 GHz during the mid season of MicroWEX-3 using 2 cm field observed soil temperature and moisture as inputs to the MB-Cotton model. During the mid season, the average valu es of MAD and RMSD between observed and simulated TB using 0 to 2 cm soil moisture and temperature profiles simulated by the LSP model were 5.7 and 6.7 K, respectivel y; whereas average values of MAD and RMSD between observed and simulated TB using 2 cm soil moisture and temperature measurements were 6.8 and 7.6 K, respectiv ely. During the mid season, the assumption of non-scattering canopy cloud model is valid be cause the mean values of the simulated TB using both field observations a nd LSP estimations as inputs were less or equal to that of the observed TB. The Late Season For the late season, the MB-Cotton m odel was run for 27 days from DOY 288 (October 14) through 314 (November 9). During this period, cotton bolls began to open on DOY 292. On DOY 314, about 12 % of th e cotton bolls were opened. Figure 5-11 shows the cotton canopy during the third sub-se ason. During this period, the crop height and width reached constant values of ~1.0 and 0.6 m, respectively (Figure 2-42 and 243). The biomass and plant water content kept increasing (Figure 2-45 and 2-46). This is because the biomass of the cotton boll increa sed exponentially in the late season (Figure

PAGE 102

85 2-47 to 2-54). There was no irrigation applie d to the crop during the late season. Table 58 shows the start and end times of the dry down periods in the late season. Figure 5-11. Cotton canopy in the late season. The photo was taken on DOY 306 (November 1). Table 5-8. The start and end times of the dry down period in the late season during MicroWEX-3. No Start Time End Time 1 288.0 289.3 2 289.6 294.2 3 295.6 300.5 4 300.6 307.7 5 309.5 315.0 Comparison using VSM and temperature from LSP Figure 5-12(a) shows the simulated and observed TB at H-pol using LSP estimated 0 to 2 cm soil temperature and moisture as model inputs during the late season. Between DOY 288 and 294, the MB-Cotton model matched the phases and amplitudes of diurnal variation of the observed TB well and average values of MAD and RMSD were 5.1 and 5.8 K, respectively (Table 5-9). Starting from DOY 295 until the end of season, the model matched the observed TB well during the night fairly well. But the model overestimated the observed TB during the day by an average of ~10 K.

PAGE 103

86 290 295 300 305 310 31 5 235 240 245 250 255 260 265 270 275 280 285 (a) T B (K) Bolls Opening (a) Bolls OpeningMicroWEX MBCotton Model (LSP 0cm) 290 295 300 305 310 31 5 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 (b)VSMMicroWEX 2cm LSP 0.5cm LSP 0.5.0cm LSP 1.0.0 cm 290 295 300 305 310 31 5 280 284 288 292 296 300 304 308 312 (c)Surface Temperature (K)MicroWEX TIR LSP 290 295 300 305 310 31 5 284 288 292 296 300 304 308 312 316 (d)Soil Temperature (K)DOY in 2004 (EST)MicroWEX 2cm LSP 0.5cm LSP 0.5.0cm LSP 1.0.0 cm Figure 5-12. Comparison of th e simulated and observed TB at H-pol at 6.7 GHz using (a) 0-2 cm LSP simulated soil temperature and moisture as inputs to the MBCotton model, (b) the 2 cm observed a nd 0 to 2 cm LSP simulated VSM, (c) the observed and LSP simulated surface temperatures, and (d) the 2 cm observed and 0 to 2 cm LSP simulated so il temperatures during the late season of MicroWEX-3.

PAGE 104

87 Table 5-9. Mean absolute differences (MAD) and root mean square differences (RMSD) using MicroWEX-3 measurements at 2 cm (2 cm) and LSP estimations between 0 and 2 cm (0-2cm) during the dry down in the mid season during MicroWEX-3. MAD (K) RMSD (K) MAD (K) RMSD (K) No. 0-2 cm 0-2 cm 2 cm 2 cm 1 5.6 6.2 9.7 9.9 2 5.0 5.7 5.4 5.9 3 5.9 7.1 1.6 2.6 4 3.9 5.1 3.4 4.4 5 7.6 8.9 4.4 5.1 The average MAD and RMSD during this period were 5.8 and 7.3 K, respectively (Table 5-9). This is primarily because the LSP model underestimated the VSM at 0 to 2 cm during the day by ~3 % (Figure 5-12(b)). This might be due to the unrealistic representation of the moisture transport in LSP model process at night when the cotton canopy begins to senesce. If the VSM pr ofiles estimated by LSP model could be improved significantly, the accuracy of TB simulation should improve as well. Comparison using observed VSM and temperature at 2 cm Figure 5-13 shows the simulated and observed TB at H-pol using field observed soil temperature and moisture at 2 cm as m odel inputs during the la te season. During DOY 288 to 294, the MB-Cotton model captured the ph ases of the diurnal variation. However, the model underestimated the diurnal amplitude by ~10 K. This is because the underestimation in VSM measured at 2 cm when the cotton canopy was mature. During the period of three days after the beginning of formation of open bolls (DOY 295) and when ~10 % of the cotton bolls were opene d (DOY 307), the model captured the phases and amplitudes of the diurnal variation ve ry well. The MAD and RMSD during this period were 2.4 and 3.2 K, respectively. During DOY 309 and 315, the model captured the phases of the diurnal variation of the observed TB. But the amplitudes were

PAGE 105

88 underestimated from DOY 312 to DOY 315. This is because Cc was underestimated. Toward the end of season due to the defolia tion, defoliation decreased the density of leaves in cotton canopy. Therefore, the cont ribution from bare soil emission increased because of less extinction of the radiati on emitted from the bare soil underneath the canopy. The accuracy of canopy emission could be improved by a better representation of Cc. An empirical function that relates the Cc to LAI would account for the overestimation of Cc toward the end of the season when the density of the leaves decreases. 290 295 300 305 310 31 5 235 240 245 250 255 260 265 270 275 280 285 T B (K) DOY in 2004 (EST) Bolls Opening Bolls OpeningMicroWEX MB Model (MicroWEX 2cm) Figure 5-13. Comparison of th e simulated and observed TB at H-pol at 6.7 GHz during the late season of MicroW EX-3 using 2 cm field observed soil temperature and moisture as input to the MB-Cotton model. During the late season, the average values of MAD and RMSD between observed and simulated TB using 0 to 2 cm soil moisture and temperature profiles simulated by the LSP model were 5.5 and 6.9 K, respectivel y; whereas average values of MAD and RMSD between observed and simulated TB using 2 cm soil moisture and temperature measurements were 4.0 and 5.1 K, resp ectively. During the late season, the nonscattering canopy cloud model is also applic able to cotton canopy because the mean values of the simulated TB using both inputs were smaller than that of the observed TB.

PAGE 106

89 The Complete Growing Season Based upon the results found from the previ ous sections for early, mid, and late seasons, 0 to 2 cm LSP estimated soil temperatures and moisture profiles were used for simulation from the beginning of the season to DOY 294. Due to the underestimation in VSM by the LSP model after DOY 196, the obser ved VSM and temperature at 2 cm were used for the MB-Cotton model simulation for the rest of season. Figure 5-14 shows the comparison of the observed TB and simulated TB using 2 cm measured temperature and VSM and 0 to 2 cm temperature and moisture profiles estimated by the LSP model before DOY 196 for the whole growing season. 200 210 220 230 240 250 260 270 280 290 300 310 120 140 160 180 200 220 240 260 280 T B (K) DOY in 2004 (EST)MicroWEX MBCotton Model (MicroWEX 2cm) MBCotton Model (LSP 0cm+MicroWEX 2cm) Figure 5-14. Comparison of the observed and simulated TB at H-pol using 2 cm measured temperature and VSM and 0 to 2 cm temperature and moisture profiles estimated by the LSP model before DOY 196 during the whole season of MicroWEX-3. For the whole season, during the dr y down period, using LSP estimated temperature and moisture profiles at 0 to 2 cm produced the MAD and RMSD between the simulated and observed TB were 5.1 and 6.3 K, resp ectively. The canopy scattering was not significant for the cotton canopy at C-band. During the infiltration, MB-Cotton model using LSP estimated temperature and moisture profiles at 0 to 2 cm overestimated the TB in the early and mid seasons. During the late season, the MB-Cotton model

PAGE 107

90 overestimated the observed TB. This was primarily because th e infiltration process in the LSP model was yet to modified for sandy soils. Conclusion In this chapter, the MB-Cotton model was evaluated for a growing season. The whole growing season was divided into th ree sub-seasons based upon the phenological stages. The field observed VSM and temperature at 2 cm were used as inputs to the MBCotton model as well as those at 0 to 2 cm estimated by the LSP model. To evaluate the results, the simulated TB were compared to the TB observed for the dry down periods during MicroWEX-3. The MB-C otton model was evaluated only during the dry down periods primarily because the infiltration pro cess in the LSP model was yet to modified for sandy soils. This chapter provides results to the third research question mentioned in Chapter 1 follows: Third question: “How well does the MB-Cotton model capture the observed brightness using soil moisture at 2 cm?” During the early season between emerging and square forming, the TB simulated by the MB-Cotton model using 2 cm soil moisture and temperature measurements matched the phases of the observed TB. The average values of MAD and RMSD between the observed and modeled TB were 7.5 and 9.3 K, respectively. During the mid season, although the model captured the phases of diurnal variation, the model underestimated the diur nal amplitudes. The average values of MAD and RMSD were 6.8 and 7.6 K, respectively. During late season, the model captured the phases and amplitudes of the diurnal variati on very well. The average values of MAD and RMSD were 4.0 and 5.1 K, respectiv ely. Based upon the average values of sensitivities of TB to changes in VSM assuming uniform vertical distribution of VSM (see Chapter 6), the TB simulated by the MB-Cotton mode l using 2 cm soil moisture and

PAGE 108

91 temperature measurements produced an error of ~2 % in VSM during the early and mid season, while the errors reduced to ~1 % during the late season. The chapter also provides the Fourth question: “How well does the MB-Cotton model capture the observed brightness if th e detailed soil moisture information is available for the effective depth?” Based upon the results of model evaluation during the early season, the accuracy of MB-Co tton model simulation could be improved by calibrating the underestimation in VSM estimat ed by the LSP model of ~4 % during the day and ~6 % during the night. Because TB at C-band is not sensit ive to changes in VSM during the mid and late seas ons, the accuracy of MB-Co tton model simulation did not improve significantly using detailed VSM between 0 and 2 cm during these time. Forth question: “Is scattering in the canopy impo rtant for simulating accurate TB when cotton reaches maturity?” as follows: Based upon these results of comparison between modeled and observed TB, the volume scattering is not significant in mature cotton canopy at C-band. Because the mean values of the simulated TB using both field observations and LSP estimations as inputs were less or equal to that of the observed TB, inclusion of scattering will further decrease the model estimates of TB.

PAGE 109

92 CHAPTER 6 SENSITIVITY ANALYSIS OF BRIGHTNESS TEMPERATURE TO CHANGES IN SOIL MOISTURE In this chapter, the sensitivity of TB to changes in soil moisture and temperature for a growing season of cotton were investig ated. Observations from MicroWEX-1 and 3 were used to discuss changes in these se nsitivities with growing cotton. Because TB is a complex function of moisture and temperatur e distributions, the empirical relationships for plant water content (PWC) versus gravimetric water content (Mg), canopy height (Hc), and canopy cover (Cc) collected during MicroWEX-3 were utilized so that the MBCotton model described in Chapter 4 could be used to calculate sensitivities due to changes only in moisture or temperature as PWC increases. These simulated sensitivities were compared with those observed duri ng MicroWEX-3 for the growing season. Finally, the simulated and observed sensitivi ties during MicroWEX-3 were compared to the observed sensitivities during MicroWEX-1. MicroWEX-3 Season Simulated Sensitivity Using MB-Cotton Model The sensitivities of TB to changes in VSM using MB-Cotton model can be found by taking the partial deriva tive of equation (4.1) with respect to VSM as: VSM T C VSM T C VSM Tp C B c p S B c p B , , , , ,1 (6.1) The first term or the bare soil term can be expressed in detail as: VSM T VSM T VSM T VSM Tp eff p eff p sky B p S B 1, , , (6.2)

PAGE 110

93 and the second term or the canopy term can be expressed as: VSM T VSM T VSM T VSM TCanopy p B Soil p B Sky p B p C B , , , , , (6.3) where VSM T VSM Tp c sky B Sky p B 2 exp, , (6.4) VSM T VSM T VSM Tp eff p eff c Soil p B1 exp, (6.5) VSM T VSM Tp c canopy c Canopy p B exp exp 1, (6.6) Here, the vegetation properties, such as c and Tcanopy were assumed to be independent of the changes in VSM. Also, VSM Teff is zero under isothermal condition. Because PWC is independent of vegetati on type, it is a widely used canopy variable to express the sensitivity of TB. The values of PWC are not directly used in the MB-Cotton model, but they can be empirically related to the input vegetation variables (Mg, Cc, and Hc). Figure 6-1 (a), (b), and (c) sh ow the relationships between the vegetation properties to Cc from vegetation sampling data and regression models. The correlation between Mg was linear until Cc = 0.45. Then constant Mg was observed from Cc = 0.45 to 1.0. A similar trend was also observed for the correlation between Hc and Cc whereas a plateau occurred after Cc = 0.7. A linear relationship was used between Cc and PWC because PWC did not reach a constant value as Cc increased. Table 6-1 shows the R-square values for Mg and Cc (Figure 6-1(a)), Hc and Cc (Figure 6-1(b)), and PWC and Cc (Figure 6-1(c)).

PAGE 111

94 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.5 1 1.5 2 2.5 3 3.5 4 (a) Mg=4.92Cc (Cc<=0.45); Mg=2.21 (Cc>0.45)Mg 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.5 1 1.5 2 (b) Hc=1.30Cc (Cc<=0.7); Hc=0.91 (Cc>0.7)Hc (m) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.5 1 1.5 2 (c) PWC=1.35CcPWC (kg/m2)Cc Figure 6-1. Relationships between the vegeta tion properties derived using data collected during MicroWEX-3. Table 6-1. R-square (R2) values for the fits for the relationships between Mg, Hc, and PWC versus Cc. R2 Mg-Cc 0.3893 Hc-Cc 0.8711 PWC-Cc 0.4779 TB at Hand V-pol were calculated with incremental increases in VSM during the growing season as PWC increased from 0 (bare soil) to 1.2 kg/m2. The incremental increases in VSM ranged from dry (5 %) to saturated soil (35 %). The soil layers and

PAGE 112

95 vegetation canopy were isothermal at 300 K, si milar to the av erage effective temperature observed during MicroWEX-3. The VSM was assu med to be constant at the effective depth (Zeff), 0 to 2 cm, for the sensitivity of TB to changes in effective temperature calculation. The cotton canopy was assumed to c onsist of uniformly distributed dielectric materials. A set of input data was genera ted for the MB-Cotton model described in Chapter 4. The model simulated the Hand V-pol TB with incremental increases in VSM as PWC increased from 0 (bare soil) to 1.2 kg/m2 (Cc = 0.9). The sensitivities of TB to changes in VSM was calculated as: 1 2 1 , , 2 , , ,VSM VSM T T VSM Tp B p B p B (6.7) where TB,p,1 and TB,p,2 are the brightness temperatures calculated by the MB-Cotton model using soil moisture values of VSM1 and VSM2, respectively. The incremental increases from VSM1 to VSM2 ranged from 5-6, 10-11, 20-21, 30-31 and 35-36 % as the range observed during MicroWEX-3. The sensitivities of TB to changes in Teff was calculated as: 1 , 2 , 1 , , 2 , , , eff eff p B p B eff p BT T T T T T (6.8) where TB,p,1 and TB,p,2 are the brightness temperatures calculated by the MB-Cotton model using soil temperature values of Teff,1 = Tcanopy,1 = 300 K and Teff,2 = Tcanopy,2 = 301 K, respectively. To calculate the sensitivity of TB to changes in Teff, the VSM profile was constant at 20 %. Figure 6-2 and 6-3 show th e sensitivities of the TB at Hand V-pol to changes in soil moisture over the growing season as estimated by the MB-Cotton model.

PAGE 113

96 0 0.2 0.4 0.6 0.8 1 1 .2 0 1 2 3 4 5 6 Hpol TB/ VSM (K/%)PWC (kg/m2) 5% 10% 20% 30% 35% Figure 6-2. Sensitivity of the model brightness temperatures to changes in volumetric soil moisture (VSM) at different plant wa ter content (PWC) at H-pol. For this simulation, Tcanopy = Teff = 300 K. 0 0.2 0.4 0.6 0.8 1 1.2 0 1 2 3 4 5 6 Vpol TB/ VSM (K/%)PWC (kg/m2) 5% 10% 20% 30% 35% Figure 6-3. Sensitivity of the model brightness temperatures to changes in volumetric soil moisture (VSM) at different plant wate r content (PWC) at V-pol. . For this simulation, Tcanopy = Teff = 300 K.

PAGE 114

97 The TB are the most sensitive to soil moisture changes in bare soil conditions and when the soil is dry. The sensitivity decrease s linearly with increasing soil wetness for Hpol, whereas the decrease is not linear for Vpol (see Figure 6-2 and 6-3). The sensitivity of the TB at H-pol was 5.6 K/% for dry soil (VSM = 5-6 %) and the sensitivity decreased to 1.5 K/% for saturated soil (VSM = 35-36 %). The sensitivity of the TB at V-pol was low at 1.6 K/% for very dry soil, increased to its maximum of 2.0 K/% at VSM = 20-21 %, and decreased to 1.7 K/% for VSM = 3536 % (see Figure 6-3). The sensitivity was maximum at 20 % VSM because the reflectivit y at V-pol is a sec ond order function of dielectric property of the soils (see equation (4.4)), with the smallest value of emissivity occurring at VSM = 20 %. As the PWC increases, the sensitivities of TB at Hand V-pol decrease linearly. This is primarily because of the linear empirical relationships shown in Figure 6-1. The sensitivity of the TB to soil moisture decreased significantly to less than 1.0 K/% for both polarizations when ve getation water content is > 1.1 kg/m2. Finally, it should be noted that the sensitivity curves shown in Figure 6-2 and 6-3 were derived from the experimental data. Therefor e, the simulated sensitivity of TB to changes in VSM is diminished when Cc = 1. Figure 6-4 to 6-6 show the sensitivity of modeled TB to changes in effective physical temperature of bare soil (Teff) as VSM increases from 5 to 35 % when PWC = 0, 0.5, and 1.2 kg/m2, which are the average PWC values observed during the early, mid, and late season during MicroWEX-3. The sensitivity of TB to changes in Teff is obtained by calculating the differences between the TB at the Teff of 300 and 301 K for 5 to 35 % VSM. When PWC = 0 kg/m2, the sensitivity of TB to Teff is < 0.75 K/K with the sensitivity decreasing to 0.51 K/K at H-pol.

PAGE 115

98 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.5 0.6 0.7 0.8 0.9 1 TB/ Teff (K/K)VSM (m3/m3) Hpol Vpol Figure 6-4. Sensitivity of brightness temperatur es to effective physical temperature of soil (Teff) for bare soil condition, i.e. PWC = 0 kg/m2. 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.5 0.6 0.7 0.8 0.9 1 TB/ Teff (K/K)VSM (m3/m3) Hpol Vpol Figure 6-5. Sensitivity of brightness temperatur es to effective physical temperature of soil (Teff) for PWC = 0.5 kg/m2. 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.5 0.6 0.7 0.8 0.9 1 TB/ Teff (K/K)VSM (m3/m3) Hpol Vpol Figure 6-6. Sensitivity of brightness temperatur es to effective physical temperature of soil (Teff) for PWC = 1.2 kg/m2.

PAGE 116

99 At V-pol, the sensitivity of TB to Teff is < 0.99 K/K with th e sensitivity decreasing to 0.93 K/K (see Figure 64).. When PWC = 0.5 kg/m2, the sensitivity of TB to Teff is < 0.85 K/K with the sensitivity decreasing to 0. 71 K/K at H-pol. At V-pol , the sensitivity of TB to Teff is < 0.99 K/K with the sensitivity decreasing to 0.95 K/K (see Figure 6-5). When PWC = 1.2 kg/m2, the sensitivity of TB to Teff is < 0.97 K/K with the sensitivity decreasing to 0.95 K/K at H-pol. At V-pol, the sensitivity of TB to Teff is < 0.99 K/K (see Figure 6-6). As PWC increases, the sensitivities of TB to Teff and Tcanopy increase because the terrain emission is dominated by the canopy emission, which is independent upon VSM. Observed Sensitivity Figure 6-7 shows the close up of TB and VSM during a part of the early season. Between DOY 204 and 214, the values of LAI and PWC were < 0.5 and 0.0 kg/m2, respectively (Figure 2-44 and 2-46) and both crop height a nd width of 5 cm (Figure 2-42 and 2-43). On DOY 205.6, the TB at Hand V-pol decr eased by 100 K and 80 K, respectively. The decreases in TB corresponded to an increase in the VSM of 12 % at 2 cm depth. Another series of irrigation even ts occurred from DOY 209 to 211. During this period, corresponding to an average increase in VSM by 10 %, the decreases in TB at Hand V-pol were 97 and 42 K, respectively. Because TB is a function of both Teff and VSM, a part of the observed decrease in TB is due to a sudden change in Teff. The decrease in Teff during these irrigation events was ~2 K. Based upon the sensitivities of TB to Teff simulated by the MB-Cotton model (Figure 6-4), thus resulted in the decrease in TB at Hand V-pol of ~1.5 and 2.0 K, re spectively. Accounting for the decreases in Teff by the MB-Cotton model, the average values of TB/ VSM at Hand V-pol during the early season were 8.8 and 5.3 K/%, respectively.

PAGE 117

100 204 205 206 207 208 209 210 211 212 213 214 120 140 160 180 200 220 240 260 280 300 320 T B (K) DOY in 2004 (EST) 204 205 206 207 208 209 210 211 212 213 214 0 4 8 12 16 20 VSM (%)Vpol Hpol VSM@2cm Figure 6-7. Response of bright ness temperatures at Vand H-pol to the soil moisture changes at 2 cm during the early season. Figure 6-8 shows the close up of TB and VSM during a part of the mid season. Between DOY 253 and 256, the values of LAI and PWC were ~1.5 and 0.5 kg/m2, respectively (Figure 2-44 and 2-46) and crop height and widt h of 60 and 50 cm (Figure 242 and 2-43), respectively. On DOY 253.6, an irrigation occurred that increased VSM at 2 cm by 8 %. The decreases in TB at Hand V-pol correspondi ng to this event were 26 and 20 K, respectively. The decrease in Teff was ~1 K, resulted in the decrease in TB at Hand V-pol were 0.85 and 0.99 K, respec tively. Therefore, the values of TB/ VSM at Hand V-pol were 3.1 and 2.3 K, respectively. 253 254 255 256 240 250 260 270 280 290 T B (K) 253 254 255 256 0 4 8 12 16 20 VSM (%)DOY in 2004 (EST) Vpol Hpol VSM@2cm Figure 6-8. Response of bright ness temperatures at Vand H-pol to the soil moisture changes at 2 cm during the mid season.

PAGE 118

101 Figure 6-9 shows the close up of TB and VSM during a part of the late season. Between DOY 299 and 303, the values of LAI and PWC were ~2.2 and 1.2 kg/m2, respectively (Figure 2-44 and 2-46) and crop height and widt h of 90 and 70 cm (Figure 242 and 2-43), respectively. On DOY 300.6, th e VSM at 2 cm increased by 13 % due to irrigation. The observed TB at Hand V-pol decreased by 26 and 24 K, respectively. The decrease in Teff was ~2 K, resulted in the decrease in TB at Hand V-pol were ~1.9 and 2.0 K, respectively. The values of TB/ VSM at Vand H-pol were 1.9 and 1.7 K, respectively. 299 300 301 302 303 240 250 260 270 280 290 T B (K) DOY in 2004 (EST) 299 300 301 302 303 0 5 10 15 20 25 VSM (%)Vpol Hpol VSM@2cm Figure 6-9. Response of bright ness temperatures at Vand H-pol to the soil moisture changes at 2 cm during the late season. The sensitivities of TB derived from the MB-Cotton model at PWC = 0 kg/m2 was 5.6 K/% at H-pol and 1.6 K/% at V-pol for VSM =5-6 %, which is the typical value observed before irrigation dur ing MicroWEX-3. The observed TB was more sensitive to the changes in soil moisture than the modeled TB. This is because of the simulated sensitivity values were based upon inst antaneous VSM changes for a uniformly distributed VSM profile, whereas the observed sensitivity valu es were calculated using 15 minutes TB observations. Therefore, instead of a uniform vertical VSM distribution at the effective depth (~2 cm for C-band, see Chap ter 4), there is sharp gradient of moisture

PAGE 119

102 within the top 2 cm for sandy soil during infiltration, which might be needed for accurate sensitivity simulation us ing MB-Cotton model. During the mid season when LAI and PWC were ~1.5 and 0.5 kg/m2, the modeled sensitivities at Hand V-pol were 2.8 and 1.0 K/%, respectively. The modeled sensitivities matched the observed sensitivit ies. The differences between the observed and modeled sensitivities were of an averag e of 0.8 K/%. The better agreement between the simulated and observed values is because the overall terrain emission was dominated by the canopy emission, which is VSM i ndependent as the canopy developed. During the late season when the PWC was at 1.2 kg/m2, the sensitivities of the modeled TB to the changes in soil moisture were at average values of 0.6 K/% at H-pol and 0.2 K/% at V-pol. But the observed sensitiv ities were ~2.0 K/%. This is primarily because the contribution from the bare soil emission increased when the density of the leaves decreased due to defoliation. Theref ore, the MB-Cotton model might slightly overestimate the canopy emission due to th e assumption of uniformly distributed dielectric materials in the canopy. Thus, Cc should be slightly lower than it was assumed. The results suggested that the TB at both polarizations at C-band are still sensitive to VSM at 2 cm during the whole growing season. MicroWEX-1 Season The canopy biomass observed during Micr oWEX-1 was higher than MicroWEX3, which was a typical cotton crop (see Figur e 2-36 and 2-45 in Chapter 2). Because the canopy width data during the field experiment were not availabl e during MicroWEX-1, the model sensitivities simulated by the MB-C otton could not be computed correctly. Thus only the observed sensitivities during Mi croWEX-1 are discussed. The sensitivities simulated by the MB-Cotton model from the pr evious section are us ed to account for the

PAGE 120

103 effect due to changes in Teff. This section provides opportunity for the inter-seasonal comparison of the observe d sensitivities of TB to changes in VSM from MicroWEX-1 and 3. Figure 6-10 shows the close up of TB and VSM during a part of the early season. The TB at both polarizations responded to the pr ecipitation and/or irri gation in the early growing season. For example, the TB at Vand H-pol decreased by 40 K and 70 K, respectively, on DOY 205.5, corresponding to an increase in the VSM of 25 % at 4 cm depth. During the increase in VSM, from 12 % to 36 %, the soil temperature at 4 cm decreased by ~2 K. The TB/ Teff for initial VSM of ~12 % were 0.48 and 0.89 K/% at Hand V-pol, respectively (Figur e 6-4). This would result in a TB/ VSM of 1.5 and 2.8 K/% at Vand H-pol, respectively. An interm ediate response of about 15 K at V-pol and 20 K at H-pol on DOY 206.5 was observed for an increase of ~10 % in volumetric soil moisture. Accounting for the ~2.0 K decreased in Teff, the values of TB/ VSM for DOY 2065.5 were 1.3 and 1.9 K/% at Vand H-pol, respectively. Similar responses were also observed on DOY 212.5 and 213.8 corresponding to an increase of about 15%. 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 140 160 180 200 220 240 260 280 300 T B (K) DOY in 2003 (EST) 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 0 10 20 30 40 VSM (%)Vpol Hpol VSM@4cm Figure 6-10. Response of bright ness temperatures at Vand H-pol to the soil moisture changes at 4 cm during the early season.

PAGE 121

104 Figure 6-11 shows the close up of TB and VSM during a part of the late season. In the late season, when the TB were dominated by emission from vegetation canopy, the sensitivity of the TB to the soil moisture diminished. The TB at both polarizations were less sensitive to the change in soil moisture. The TB did not respond significantly to increases of 20, 10, and 20 % on DOY 302, 310, and 323, respectively. 230 240 250 260 270 280 T B (K) 300 305 310 315 320 325 330 335 340 0 8 16 24 32 40 VSM (%)DOY in 2003 (EST) Vpol Hpol VSM@4cm Figure 6-11. Response of bright ness temperatures of Vand H-pol to the soil moisture changes at 4 cm during the late season. Comparison Between MicroWEX-1 and 3 Seasons Table 6-2 shows the simulated and obse rved values of sensitivity during MicroWEX-1 and 3. Both observed TB responded to changes in VSM in the early seasons during MicroWEX-1 and 3. Howe ver, the values of observed TB/ VSM at two polarizations were considerably higher during MicroWEX-3 than those during MicroWEX-1. Higher values of the observed sensitivities during MicroWEX-3 than those during MicroWEX-1 were because the measured VSM was at 2 cm during MicroWEX-3 whereas the measured VSM was at 4 cm during MicroWEX-1, recall that Zeff at C-band is ~2 cm for the given VSM conditions. Therefore, TB at C-band is less sensitive to changes in VSM at 4 cm than to changes in VSM at 2 cm. The higher sensitivity values during the early season in MicroWEX-3 are also due to a drier field

PAGE 122

105 condition during the field experiment. During the late season, observed TB at both polarizations responded to the precipitation and/or ir rigation during MicroWEX-3, whereas there was no sensitivity of TB to changes in VSM during MicroWEX-1. This is because the cotton was under typical agricultu ral practice of the a pplication of growth regulators that prevented the canopy width gr owing from exceeding the row width during MicroWEX-3. The canopy biomass during MicroWEX-1 was higher and the canopy cover achieved unity very early in the season, as the cotton was not managed with typical growth regulators. The TB at C-band are still sensitive to changes in VSM with PWC = 1.2 kg/m2. Thus, it is still possible to use TB at C-band observations to retrieve soil moisture even during the late season of cotton. Table 6-2. The simulated and observed sens itivity values during MicroWEX-1 and 3. MicroWEX-1 MicroWEX-3 Observed (K/%) Observed (K/%) Simulated (K/%) H-pol V-pol H-pol V-pol H-pol V-pol Early 2.4 1.4 8.8 5.3 5.6 1.6 Mid N/A N/A 3.1 2.3 2.8 1.0 Late 0.0 0.0 1.9 1.7 0.6 0.2 Conclusion In this chapter, the MB-Cotton model desc ribed in Chapter 4 was used to simulate sensitivities of TB to changes in VSM during the gr owing season of MicroWEX-3 when PWC increased from 0 to 1.2 kg/m2. Because TB is a function of both Teff and VSM, a part of the observed decrease in TB observed is due to a sudden change in Teff. Therefore, the MB-Cotton model was used to simulate the change in TB due to changes in Teff under different PWC conditions. The sensitivities observed during MicroWEX-3 were discussed and the modeled sensitivities were compared to the observed sensitivities. The sensitivities observed duri ng MicroWEX-1 were also discussed and compare the

PAGE 123

106 observed values between MicroWEX-1 and 3 we re compared. The results suggested that the TB at both polarizations at C-band are still sensitive to VSM at 2 cm during the whole growing season. This chapter provides results to the si xth research question in Chapter 1 “How does the sensitivity of TB to soil moisture changes as cotton mature?” as follows: During MicroWEX3, the observed sensitivities of TB at H-pol decreased from 8.8 to 1.9 K/%, while those at V-pol decreased from 5.3 to 1.7 K/%. The TB at C-band are sensitive to changes in VSM during the whole growing season of cotton. The modeled sensitivities using MB-Cotton were lower than the observe d values and decreased from 5.6 K/% at Hpol and 1.6 K/% at V-pol to less than 1 K/% at both polarizati ons in the late season. Thus, TB at C-band can be used in data assimilati on for the growing season of cotton to improve estimates of moisture and energy fluxes by the LSP model.

PAGE 124

107 CHAPTER 7 CONCLUSIONS In this chapter, the results and contribu tions from this dissertation are summarized and recommendations for future research are provided. Summary This dissertation provides important in sights into microwave brightness modeling for dynamic vegetation and into sensitivity of the brightness temperat ures to changes in soil moisture. The major objective of the res earch was to develop a microwave brightness model at C-band (6.7 GHz) for the entire grow ing season of cotton ( one of the dominant agricultural crops in the south eastern region of the U. S.) a nd calibrate it using the data collected during fiel d experiments. Two extensive field experiments were c onducted, namely MicroWEX-1 and 3. The goal of MicroWEX-1 and 3 were to investig ate the interactions between the microwave brightness signatures, land surface cond itions, energy components, and hydrologic processes for the whole growing season of cotto n. A variety of data were collected during MicroWEX-1 and 3, including TB at C-band, soil moisture and temperature profiles, soil heat flux, surface thermal infrared temper ature, upwelling and downwelling solar and longwave radiation, latent a nd sensible heat fluxes, and precipitation and irrigation. MicroWEX-1 and 3 are unique datasets that can be used for various inter-disciplinary studies. The data are available in the UF /IFAS website (http://edis.ifas.ufl.edu). The observed TB were collected using the UFCMR. To assess the accuracy for the TB observations by the UFCMR during our fiel d experiments, microwave radiometer

PAGE 125

108 calibration experiments were conducted in Florida during MicroWEXs and in Alaska during REBEX-10 using two C-ba nd radiometers with nearly identical design. The three most commonly used calibration techniques fo r microwave radiometers, IC, EC, and TC, were tested. Accuracy of the measured TB was evaluated using a lake emission model. The results suggested that IC produced the most consistent and accurate TB. The MAE between the measured and simulated TB was ~2 K, which is suitable for soil moisture studies. A MB-Cotton model based upon the radia tive transfer equation was used to simulate the TB emitted from the terrain, which is a dynamic mixture of bare soil and cotton canopy. The bare soil emission was modeled as the emitted radiation from a specular, horizontally semi-i nfinite, vertically multiple-lay ered, non-scattering dielectric medium. The cotton canopy was modeled as an emissive, non-scatte ring dielectric cloud with homogeneous dielectric properties. The dielectric properties of the bare soil and canopy were obtained from literature-based di electric mixing and dispersion models, respectively. TB at H-pol simulated by the MB-C otton model developed in this dissertation were calibrate d against the observed TB during MicroWEX-3 for the whole growing season. To investigate the effect of a dynamic cotton canopy, the simulation period for the complete growing season was di vided into three sub-seasons, based upon the phenological stages. The soil moisture a nd temperature at 2 cm measured during the field experiment as well as the soil moistu re and temperature pr ofiles at 0 to 2 cm estimated by the LSP model were used as inputs to the MB-Cotton model to show the importance of detailed soil moisture and temp erature information at 0 to 1 cm for TB simulation.

PAGE 126

109 During the early season, the MB-Cotton mode l captured the phas es of the diurnal variation well using the 0 to 2 cm soil moisture and temperature profiles estimated by the LSP model. The TB simulated by the MB-Cotton mode l matched well with the observed TB during first dry down period. The MB-Cott on model estimated the phases well for the rest of dry downs and matched well with observations during th e night, although the model overestimated the observed TB by ~15 to 20 K during the day. This could be due to the underestimation of the VSM by the LSP model by ~4 and 6 %, respectively. Overall, the amplitudes of diurnal variation were si gnificantly underestimated by the MB-Cotton model using the observed 2 cm soil moisture and temperature as inputs. The diurnal amplitudes of the observed TB during the three day dry down periods were ~60, 40, and 20 K, respectively; whereas the diurnal amplitude of the simulated TB was only ~15 K. The underestimation of diurnal amplitude is primarily because the strong diurnal variation in of vertical VS M distribution within the top 2 cm. Using VSM at 2 cm as input does not represent the soil moisture dist ribution from 0 to 1 cm realistically during dry downs in the early season. The average values of MAE and RMSE between observed and simulated TB using 0 to 2 cm LSP estimations were 6.9 and 8.5 K, respectively. The MAE and RMSE between the observed and simu lated TB using 2 cm field measurements were higher at 7.5 and 9.3 K, respectively. During the mid season, the MB-Cotton m odel captured the phases of observed diurnal variations well using the LSP estimat ed 0 to 2 cm soil moisture and temperature profiles. During the first two dry downs when cotton square formation began, the model overestimated the observed TB by ~12 K during the day and ~8 K during the night. This could be caused by the overestimation in canopy emission due to a slightly

PAGE 127

110 overestimation in Mg of the vegetation. As bolls form ation began, the model estimated the phases and amplitudes of the diurnal varia tions very well. For the rest of the mid season, the MB-Cotton model matched the observed TB well during the day, although the model slightly underestimated the observed TB at night by ~8 K. This is due to the overestimation in VSM by ~6 %. During the mid season, the MB-C otton model captured the phases and amplitudes of diurnal variati on well. However, for the rest of the mid season, the model consistently underestimated the observed TB by ~15 K. This could be that the soil at 2 cm was consistently wetter than that at 0 to 1 cm during this time. The average values of MAE and RMSE between observed and simulated TB using 0 to 2 cm soil moisture and temperature profiles simu lated by the LSP model were 5.7 and 6.7 K, respectively; whereas average values of MAE and RMSE between observed and simulated TB using 2 cm soil moisture and temperature measurements were 6.8 and 7.6 K, respectively. The assumption of non-s cattering canopy cloud mode l is valid because the mean values of the simulated TB using both field observations and LSP estimations as inputs were smaller than that of the observed TB. During the late season until the open boll formation began, the MB-Cotton model matched the phases and amplitudes of diurnal variation of the observed TB very well using 0 to 2 cm soil moisture and temperature pr ofiles. For the rest of the late season, the model matched the observed TB well during the night fair ly well, although the model overestimated the observed TB during the day by an average of ~10 K. This is primarily because the LSP model underestimated the VSM at 0 to 2 cm during the day by ~3 %. If the VSM profiles estimated by LSP model coul d be improved significantly, the accuracy of TB simulation will be improved as well. Unti l the open boll formation began in the late

PAGE 128

111 season, the MB-Cotton model captured the phases of the diurnal variation using the observed 2 cm soil moisture and temper ature as inputs. However, the model underestimated the diurnal amplitude by ~ 10 K. The model captured the phases and amplitudes of the diurnal variation very well when ~10 % bolls were opened. For the rest of the late season, the model captured the pha ses of the diurnal variation of the observed TB. But the amplitudes were underestimat ed due to the underestimation in Cc. The accuracy of canopy emission could be improved by a better representation of Cc using an empirical function relating the Cc to LAI. The average values of MAE and RMSE between observed and simulated TB using 0 to 2 cm soil mo isture and temperature profiles simulated by the LSP model were 5.5 and 6.9 K, respectively; whereas average values of MAE and RMSE between observed and simulated TB using 2 cm soil moisture and temperature measurements were 4.0 a nd 5.1 K, respectively. The non-scattering canopy cloud model is also vali d during the late season. To investigate the sensitivities of TB to changes in VSM, the MB-Cotton model was used to simulate sensitivities of TB to changes in VSM during the growing season of MicroWEX-3. The MB-Cotton model was used to simulate the change in TB due to changes in Teff under different PWC conditions. Th e modeled sensitivities were compared to the observed sensitivities during MicroWEX-3. Finally, the sensitivities observed during MicroWEX-1 were used to compare the observed values between MicroWEX-1 and 3. During MicroWEX-3, the observed TB at both polarizations were more sensitive to the changes in VSM than the modeled TB. This is because of the simulated sensitivity values were based upon instantaneous VSM changes for a uniform vertical distribution in VSM, whereas the observed sensitivity values were calculated

PAGE 129

112 using 15 minutes TB observations. There is sh arp gradient of moisture within the top 2 cm for sandy soil during infiltration, which mi ght be needed for accurate sensitivity simulation using MB-Cotton model. The obser ved sensitivities duri ng MicroWEX-1 were lower than those observed during MicroWEX-3. Because Zeff at C-band is ~2 cm for the given VSM conditions, TB at C-band is less sensitive to changes in VSM at 4 cm. During the late season in MicroWEX-1, the sensitivities of TB at both polarizations diminished. This is because the canopy biomass dur ing MicroWEX-1 was higher and the canopy cover achieved unity very early in the seas on as cotton was not managed with typical growth regulators. The TB at C-band are still sensitive to changes in VSM with PWC = 1.2 kg/m2. Thus, it is still possible to use TB at C-band observations to retrieve soil moisture even during the late season of cotton. Contributions The major contributions of this dissertat ion are the developmen t, calibration, and sensitivity analysis of the dynamic MB mode l for the whole growing season of cotton. The MB model can be extended for row crops without any significan t volume scattering inside the canopy. Another significant contri bution is to the extensiv e dataset collection during MicroWEX-1 and 3 including TB at C-band, micrometeorolo gical data, and vegetation sampling for two growing seasons of cott on (~120 days each). These high temporal frequency datasets are very unique for land surface process, crop gr owth modeling, and microwave remote sensing studies.

PAGE 130

113 Recommendations for Future Research In this section, several recommendations to improve data co llection during field experiments and to improve modeling TB are provided. Some of the recommendations lead to improvement of our knowledge to the near surface moisture and energy processes. Improvements in Data Collect ion during Field Experiments Several improvements were made from Mi croWEX-1 to MicroWEX-3. First was automated processing of raw data in realtime immediately after download. This had significant impact on quality c ontrol of the dataset by identif ying sensor malfunction in a more timely manner, and decreasing the data gaps. Second, the vegetation sampling prot ocol was significantly improved and standardized for MicroWEX-3 . The protocol was approved by agronomists and is available on the web (http://edis.ifas.ufl.edu ). In spite of improvements from MicroWEX1 to MicroWEX-3, several areas coul d still be improved, which include 1. Detailed data for the cotton canopy, such as the spatial distributions of the branches, leaves, squares, and bolls is needed for better microwave brightness model development and calib ration. Also, more frequent LAI, biomass, canopy dimension observati ons are needed during the early growing season. 2. The field observations of VSM and temp erature profiles are needed for 0 to 1 cm soil depth at high temporal fr equency for MB-Cotton and LSP model calibration. The VSM observation cl osest to the surface for these experiments were taken at 2 cm usi ng currently available soil moisture sensors. The importance of 0 to 1 cm VSM for accurate simulation of diurnal variation in TB at C-band was demonstrated in this dissertation in Chapter 5. Improvements in Brightness Modeli ng using the MB-Cotton Model The calibration of the MB-Co tton model could be improved by using VSM at 0 to 1 cm. Although the LSP estimated 0 to 1 cm VS M data were used for calibration, the results suggested that the LSP model overestimated the diurnal amplitude in the 0 to 1 cm

PAGE 131

114 VSM. This is primarily because there was no available field observed data for VSM at 0 to 1 cm for LSP model calibration. Future research linking the MB-Cotton model to the LSP model will benefit from a robust calibra tion for 0 to 1 cm VSM estimated by the LSP model. Research efforts are also needed for si gnificant improvement to the LSP calibration during the late season, which would lead to improve the overestimation in the amplitude of diurnal variation of TB simulated by the MB-Cotton model. The calibration of the MB-Cotton model wa s only applied to the periods during dry down. Future research for model calibration/ validation during infiltration is needed for continuous simulations. Research efforts ar e underway to investig ate the changes in microwave brightness signatures duri ng irrigation and precipitation. Finally, the estimates of di electric constant of soil -water mixture should be improved through an experiment of dielectric property measurements for sandy soil. The dielectric model for wet soil used in this di ssertation was calibrated using only five types of soils (Dobson et al., 1985). The highest sand content of these soil types was only ~50 % compared to the 92 % sand observed in Mi croWEXs. A recently developed empirical dielectric model was calibrated using soil wi th sand content of only ~5 % (Mironov et al., 2004). However, the differences of the mode led dielectric constant of the wet soil between the dielectric mode l developed by Dobson et al. (1985) and Mironov et al. (2004) at C-band were less than 2 at both real and imaginary parts for sandy loam soils (sand = 51 %). They concluded that the differe nces would be even smaller for sandy soils because most of the uncertainty was in troduced by the fraction of bound water. A

PAGE 132

115 laboratory experiment of dielectric propert y measurements for sandy soil would verify and extend the applicability of the cu rrently existing dielectric models.

PAGE 133

116 LIST OF REFERENCES Al-Ansari K., Garcia P., Riera J. M., Bena rroch A., Fernandez D., and Fernandez L., 2002, “Calibration procedure of a mi crowave total-power radiometer,” IEEE Micro. Wireless Comp. Letters , 12: 93-95. Aubert D., Loumagne C., and Oudin L., 2003, “Sequential assimilation of soil moisture and streamflow data in a conceptual rainfall-runoff model,” J. Hydrol. , 280: 145161. Berg A. A., Famiglietti J. S., Rodell M., Reic hle R. H., Jambor U., Holl S. L., and Houser P. R., 2005, “Development of a hydrometeoro logical forcing data set for global soil moisture estimation,” Int. J. Climatol. , 25: 1697-1714. Bonnefond P., Haines B., Born G., Exertier P., Gill S., Jan G., Jeansou E., Kubitscheck D., Laurain O., Menard Y., and Orsoni A., 2003, “Calibrating the Jason-1 measurement system: initial results from the Corsica and Harvest verification experiments,” Adv. Space Res. , 32: 2135-2140. Braud I., Dantas-Antonino A. C., Vauclin M., Thony J. L., and Ruelle P., 1995, “A simple soil-plant-atmosphere transfer model (SiSPAT): development and field verification,” J. Hydrol. , 166: 231-250. Brubaker K. L. and Entekhabi D., 1995, “An analytical approach to modeling landatmosphere interaction. 1. Cons truct and equilibrium behavior,” Water Resour. Res. , 31: 619-632. Bruckler L., Witono H., and Stengel P., 1980, “Near-surface soil moisture estimation from microwave measurements,” Remote Sensing Environ. , 26: 101-121. Burke E. J., Gurney R. J., Simmonds L., and Jackson T. J., 1997, “Calibrating a soil water and energy budget model with remotely sensed data to obtain quantitative information about the soil,” Water Resour. Res. , 33: 1689-1697. Burke E. J., Shuttleworth W. J., Lee K., a nd Bastidas L. A., 2001, “Using area-average remotely sensed surface soil moisture in multipatch land data assimilation systems,” IEEE Trans. Geosci. Remote Sensing , 39: 2091-2100. Burke E. J., Bastidas L. A. and Shuttlewo rth W. J., 2002, “Explor ing the potential for multipatch soil moisture retrievals using multiparameter optimization techniques,” IEEE Trans. Geosci. Remote Sensing , 40: 1114-1120.

PAGE 134

117 Burke E. J., Shuttleworth W. J., and Ha rlow R. C., 2003, “Using MICRO-SWEAT to model microwave brightness temper atures measured during SPG97,” J. Hydrometeorol. , 4: 460-472. Burke E. J., Shuttleworth W. J., and Hosuer P. R., 2004, “Impact of horizontal and verticl heterogeneities on retrievals using multi angle microwave brightness temperature data,” IEEE Trans. Geosci. Remote Sensing , 42: 1495-1501. Burke H. K. and Schmugge T. J., 1982, “Eff ects of varying soil moisture contents and vegetation canopies on microwave emissions,” IEEE Trans. Geosci. Remote Sensing , GE-20: 268-274. Calvet J. C., Chanzy A., and Wigneron J. P., 1996, “Surface temperature and soil moisture retrieval in the Sahel from airborne multifrequency microwave radiometry,” IEEE Trans. Geo. Remote Sensing , 34: 588-600. Campbell Scientific, 1998a, Eddy Covariance System Operator’s Manual CA27 and KH20 , Campbell Scientific Inc., Logan, UT. Campbell Scientific, 1998b, CSAT3 Three Dimensional Sonic Anemometer Instruction Manual , Campbell Scientific Inc., Logan, UT. Campbell Scientific, 2006, CNR1 Net Radiometer Instruction Manual , Campbell Scientific Inc., Logan, UT. Cimini D., Westwater E. R., Han Y., and Ke ihm S. J., 2003, “Accuracy of ground-based microwave radiometer and balloon-bor ne measurements during the WVIOP2000 field experiment,” IEEE Trans. Geosci. Remote Sensing , 41: 2605-2615. Corbella I., Gasiewski A. J., Klein M., Leuski V., Francavilla A. J., and Piepmeier J. R., 2002, “On-board calibration of dual-cha nnel radiometers using internal and external references,” IEEE Trans. Micro. Theory Tech. , 50: 1816-1820. Crosson W. L., Limaye A. S., and Laymon C. A., 2005a, “Parameter sensitivity of soil moisture retrievals from airborne L-ba nd radiometer measurements in SMEX02,” IEEE Trans. Geosci. Remote Sensing , 43: 1517-1528. Crosson W. L., Limaye A. S., and Laymon C. A., 2005b, “Parameter sensitivity of soil moisture retrievals from airborne Cand X-band radiometer measurements in SMEX02,” IEEE Trans. Geosci. Remote Sensing , 43: 2842-2853. Crow W. T., 2003, “Correcting land surface model predictions for the impact of temporally sparse rainfall rate measurements using an ensemble Kalman filter and surface brightness temperature observations,” J. Hydrometeorol. , 4: 960-973.

PAGE 135

118 Crow W. T., Wood E. F., and Pan M., 2003, “Multiobjectiv calibration of land surface model evapotranspiration predictions using stream flow observations and spaceborne surface radiometric temperature retrievals,” J. Geophys. Res.-Atmos. , 108(D23). Crow W. T. and Wood E. F., 2003, “The assim ilation of remotely sensed soil brightness temperature imagery into a land surface m odel using Ensemble Kalman filtering: a case study based on ESTAR measurements during SGP97,” Adv. Water Resour. , 26: 137-149. Crow W. T., Bindlish R., and Jackson T. J., 2005a, “The added value of spaceborne passive microwave soil moisture retr ievals for forecasting rainfall-runoff partitioning,” Geophy. Res. Lett. , 32 (18). Crow W. T., Li F. Q., and Kustas W. P., 2005b, “Intercomparison of spatially distributed models for predicting surface ener gy flux patterns during SMACEX,” J. Hydrometeorol. , 6: 941-953. Davenport I. J., Fernandez-Galvez J., and Gurn ey R.J., 2005, “A sensitivity analysis of soil moisture retrieval from the tau-omega microwave emission model,” IEEE Trans. Geosci. Remote Sensing , 43: 1304-1316. Deuber B., Kampfer N., and Feist D. G., 2004, “A new 22-GHz radiometer for middle atmospheric water vapor profile measurements,” IEEE Trans. Geosci. Remote Sensing , 42: 974-984. Demarty J., Ottle C., Braud I., Olioso A., Frangi J. P., Gupta H. V., and Bastidas L. A., 2005, “Constraining a physically based Soil-Vegetation-Atmosphere Transfer model with surface water c ontent and thermal infrared brightness temperature measurements using a multiobjective approach,” Water Resour. Res. , 41(1). De Roo R., 2003, TMRS-3 Radiometer Tuning Procedures , available at http://www.eecs.umich.edu/~deroo/j asmeet.dir/0117x3_Tuning_Procedures.doc , December 2005. Dickinson R.E., Henderson-Sellers A., Kennedy P., and Wilson M., 1986, “Biosphereatmosphere transfer scheme (BATS) fo r the NCAR community climate model,” Tech. Report, NCAR/TN-275+STP , NCAR. Dobson M. C., Ulaby F. T., Hallikainen M. T., and El-Rayes M. A., 1985, “Microwave dielectric behavior of wet soil, II. Dielectric mixing models,” IEEE Trans. Geosci. Remote Sensing , GE-23: 35-46. Dunne S. and Entekhabi D., 2005, “An ensemble -based reanalysis approach to land data assimilation,” Water Resour. Res. , 41(2).

PAGE 136

119 Dunne S. and Entekhabi D., 2006, “Land su rface state and flux estimation using the ensemble Kalman smoother during th e Southern Great Plains 1997 field experiment,” Adv. Water Resour. , 42(1). Drusch M., Wood E. F., Gao H., and Thiele A ., 2004, “Soil moisture retrieval during the Southern Great Plain Hydrology Expe riment 1999: A comparison between experimental remote sensing data and operational products,” Water Resour. Res , 40. Entekhabi D. and Brubaker K. L., 1995, “An analytical approach to modeling landatmosphere interaction. 2. Stochastic extension,” Water Resour. Res. , 31: 635-643. Entekhabi D., Rodriguez-Itur be I., and Castelli F., 1996, “Mutual interaction of soil moisture and atmospheric processes,” J. Hydrol. , 184: 3-17. England A. W. and Galantowicz J. F., 1995, “Observed and modeled radiobrightness of prairie grass during early fall,” Processing IGRASS , 1: 513-515. Ferrazzoli P., Wigneron J. P.; Guerriero L., and Chanzy A., 2000, “Multifrequency emission of wheat: Modeling and applications,” IEEE Trans. Geosci. Remote Sensing , 38: 2598-2607. Goodberlet M. A. and Mead J. B., 2006, “Two -load radiometer precision and accuracy,” IEEE Trans. Geosci. Remote Sensing , 44: 58-67. Gottschalck J., Meng J., Rodell M., and Houser P., 2005, “Analysis of multiple precipitation products and preliminary asse ssment of their imp act on global land data assimilation system land surface states,” J. Hydrometeorol. , 6: 573-598. Han Y. and Westwater E. R., 2000, “Analysi s and improvement of tipping calibration for ground-based microwave radiometer,” IEEE Trans. Geosci. Remote Sensing , 38: 1260-1276. Heathman G. C., Starks P. J., Ahuja L. R., and Jackson T. J., 2003, “Assimilation of surface soil moisture to estimat e profile soil water content,” J. Hydrol. , 279: 1-17. Houser P. R., Hutchinson M. F., Viter bo P., Douville H., and Running S. W., 2004, “Terrestrial data assimilation,” Vege. Water Human Clim. , 273-287. Jackson T. J. and O’Neill P.E, 1990, “ A ttenuation of soil microw ave emission by corn and soybean at 1.4 and 5 GHz,” IEEE Trans. Geosci. Remote Sensing , 28: 978-980. Jackson T. J., Shutko A. M., Shiue J. C., Schm ugge T. J., Davis D. R., Parry R., Haldin A., Reutov E., Novichikhin E., Liberman B ., Kustas W. P., Goodrich D. C., Amer S., Bach L., and Ritchie J. C., 1992, “M ultifrequency passive microwave sensors observations of soil moisture in an arid rangeland environment,” Int. J. Remote Sensing , 13: 573-580.

PAGE 137

120 Jackson T. J., Le Vine D. M., Griffis A. J ., Goodrich D. C., Schmugge T. J., Swift C. T., and O’Neill P. E., 1993, “Soil moisture and rainfall estimation over a semiarid environment with the ESTAR microwave radiometer,” IEEE Trans. Geosci. Remote Sensing , vol. 41: 836-841. Jackson T. J., Le Vine D. M., Swfit C. T., Schmugge T. J., and Schiebe F. R., 1995, “Large area mapping of soil moisture using the ESTAR passive microwave radiometer in Washita,” Remote Sensing Environ. , 53: 27-37. Jackson T. J. and Le Vine D. E., 1996, “Mappi ng surface soil moisture using an aircraftbased passive microwave instru ment: Algorithm and example,” J. Hydrol. , 184: 8599. Jackson T. J., O’Neill P. E., and Swift C. T., 1997, “Passive microwave observation of diurnal surface soil moisture,” IEEE Trans. Geosci. Remote Sensing , 35: 12101221. Jackson T.J., Schmugge T. J., O’Neill P. E., and Parlange M. B., 1998, “Soil water infiltration observation with microwave radiometers,” IEEE Trans. Geosci. Remote Sensing , 36: 1376-1383. Jackson T. J., Le Vine D. M., Hsu A. Y., Olda k A., Starks P. J., Swif t C. T., Isham J. D., and Haken M., 1999, “Soil moisture mappi ng at regional scales using microwave radiometry: the Southern Great Plains Hydrology Experiment,” IEEE Trans. Geosci. Remote Sensing , 37: 2136-2151. Jackson T. J., 2001, “Multiple resolution of L-band brightness temperatures for soil moisture,” IEEE Trans. Geosci. Remote Sensing , 39: 151-164. Jackson T.J., Gasiewski A. J., Oldak A., Kl ein M., Njoku E. G., Yevgrafov A., Christiani S., and Bindlish R., 2002, “Soil moisture retrieval using C-band polarimetric scanning radiometer during the Sout hern Great Plain 1999 experiment,” IEEE Trans. Geosci. Remote Sensing , 40: 2151-2161. Jackson T. J., Bindlish R., Gasiewski A. J ., Stankov B., Klien M., Njoku E. G., Bosh D., Coleman T. L., Laymon C. A., and Star ks P., 2005, “Polarimetric scanning radiometer Cand X-band microw ave observations during SMEX03,” IEEE Trans. Geosci. Remote Sensing , 43: 2418-2430. Janssen M.A., 1993, Atmospheric Remote Sensing by Microwave Radiometry , New York, NY: Wiley Press. Judge J., 1999, Land surface process and radi obrightness modeling of the Great Plains, Ph.D. dissertation, University of Michig an, Department of Electrical Engineering and Atmospheric, Oceanic, and Space Sciences.

PAGE 138

121 Judge J., England A. W., Crosson W. L., Laym on C. A., Hornbuckle B. K., Borie D. L., Kim E. J., and Liou Y., 1999, “A growing season land surface process/radiobrightness model for wheat-s tubble in the Southern Great Plain,” IEEE Trans. Geosci. Remote Sensing , 37: 2152-2158. Judge J., Galantowicz J. F., and England A. W., 2001, “A comparison of ground-based and satellite-borne microwave radiomet ric observation in th e Great Plains,” IEEE Trans. Geosci. Remote Sensing , 39: 1686-1696. Judge J., Abriola, L. M., and England, A. W., 2003, “Numerical validation of the land surface process component of an LSP/R model,” Adv. Water Resour. , 26: 733-746. Judge J., Casanova J., Lin T., Tien K. C., Ja ng M., Lanni O., and Miller L., 2005, “Field Observations during the Second Microwave Water and Energy balance Experiment (MicroWEX-2) March 17 – June 3, 2004.” Circular Number: 1480 , Center for Remote Sensing, UF/IFAS, EDIS : http://edis.ifas.ufl.edu/AE360 Kay A. L., Reynard N. S., and Jones R. G ., 2006a, “RCM rainfall for UK flood frequency estimation. I. Method and validation,” J. Hydrol. , 318: 151-162. Kay A. L., Jones R. G., and Reynard N. S ., 2006b, “RCM rainfall for UK flood frequency estimation. II. Climatechange results,” J. Hydrol. , 318: 163-172. Kerr Y. H., Waldteufel P., Wigneron J. P., Ma rtinuzzi J. M., Font J., and Berger M., 2001, “Soil moisture retrieval from space: The Soil Moisture and Ocean Salinity (SMOS) mission,” IEEE Trans. Geosci. Remote Sensing , 39: 1729-1735. Laymon C. A., Crosson W. L., Jackson T. J., Manu A., and Tsegye T. D., 2001, “Ground-based passive microwave remote se nsing observations of soil moisture at S-band and L-band with insight into measurement accuracy,” IEEE Trans. Geosci. Remote Sensing , 39: 18441858. Le Vine D. M., Jackson T. J., Swift C. T., Haken M., and Bidwell S. W., 2001, “ESTAR measurements during the South Gr eat Plain Experiment (SGP99),” IEEE Trans. Geosci. Remote Sensing , 39: 1680-1685. Lemaitre F., Poussiere J. C., Kerr Y. H., Dejus M., Durbe R., Rosnay P., and Calvet J. C., 2004, “Design and test of the ground-ba sed L-band radiometer for estimating water in soil (LEWIS),” IEEE Trans. Geosci. Remote Sensing , 42: 1666-1676. Lin T., Judge J., Tien K. C., Casanova J. J., Jang M., Lanni O., Miller L. W., and Yan F., 2005, “Field Observations during the Thir d Microwave Water and Energy balance Experiment (MicroWEX-3) June 16 – December 21, 2004.” Circular Number: 1481 , Center for Remote Sensing, UF/IFAS, EDIS: http://edis.ifas.ufl.edu/AE361. Margulis S. A., McLaughlin D. B., and Ente khabi D., 2002, “Land data assimilation and estimation of soil moisture using measurements from the South Great Plains 1997 Field Experiment,” Water Resour. Res. , 38(12).

PAGE 139

122 Margulis S. A. and Entekhabi D., 2004, “B oundary-layer entrainment estimation through assimilation of radiosonde and micromet eorological data into a mixed-layer model,” Bound.-Lay. Meteorol. , 110: 405-443. McCabe M. F., Gao H., and Wood E. F., 2005, “Evaluation of AMSR-E derived soil moisture retrievals using ground-based and PSR airborne data during SMEX02,” J. Hydrometeorol. , 6: 864-877. McLaughlin D. B., 1995, “Recent advances in hydrologic data assimilation,” In US National report to the IUGG (1991-1994), Review. Geophy. , 977-984. McLaughlin D. B., 2002, “An integrated appr oach to hydrologic data assimilation: Interpolation, smoothing, and filtering,” Adv. Water Resour. , 25: 1275-1286. Meier P. R., Crowley G., Strickland D. J., Ch ristensen A. B., Paxton L. J., Morrison D., and Hackert C. L., 2005, “First look at the 20 November 2003 superstrom with TIMES/GUVI: Comparison with a thermospheric global circulation model,” J. Geophys. Res. Space , 110(49). Meissner T. and Wentz F. J., 2004, “The comp lex dielectric constant of pure and sea water from microwave sa tellite observations,” IEEE Trans. Geosci. Remote Sensing, 42: 1836-1849. Memmo A., Fionda E., Paolucci T., Cimini D., Ferretti R., Bonafoni S., and Ciotti P., 2005, “Comparison of MM5 integrated wa ter vapor with microwave radiometer, GPS, and radiosonde measurements,” IEEE Trans. Geosci. Remote Sensing , 43: 1050-1058. Mo T., Schmugge T. J., and Choudhury B. J., 1980, “Calculation of the spectral nature of the microwave emission from soils,” NASA Tech. Memo. , 82002: 1-64. Montaldo N. and Albertson J. D., 2003, “Multi -scale assimilation of surface soil moisture data for robust root zone moisture predictions,” Adv. Water Resour. , 26: 33-44. Moradkhani H., Sorooshian S., Gupta H. V., and Houser P. R., 2005, “Dual stateparameter estimation of hydrological mode ls using ensemble Kalman filter,” Adv. Water Resour. , 28: 135-147. Mulligan, M., 1998, “Modeling the geomorphological impact of climate variability and extreme events in a semi-arid environment,” Geomorph. , 24: 59-78. Newton R. W. and Rouse J. W., 1980, “Microwave radiometric measuremens of soil moisture content,” IEEE Trans. Antenna Propag. , 28: 680-686. Ni-Meister W., Walker J. P., and Houser P. R., 2005, “Soil moistu re initialization for climate prediction: Characterization of model and observation errors,” J. Geophys. Res.-Atmos. , 110(D13).

PAGE 140

123 Njoku E. G., Stacey J. M., and Barath F. T., 1980, “The Seasat Scanning Microwave Radiometer (SMMR): Instrument description and performance,” IEEE Oceanic Engi. , 5: 100-115. Njoku E. G. and Li L., 1999, “Retrieval of land surface parameters using passive microwave measurements at 6-18 GHz,” IEEE Trans. Geo. Remote Sensing , 37: 79-93. Njoku E. G.; Wilson W. J.; Yueh S. H.; Dinar do S. J.; Li F. K.; Jackson T. J.; Lakshmi V.; and Bolten J., 2002, “Observations of soil moisture using a passive and active low-frequency microwave airb orne sensor during SGP99”, IEEE Trans. Geosci. Remote Sensing , 40: 2659-2673. Njoku E. G., Jackson T. J., Lakshmi V., Chan T. K., and Nghiem S. V., 2003, “Soil moisture retrieval from AMSR-E,” IEEE Trans. Geosci. Remote Sensing , 41: 215229. Noilhan J. and Mahfouf, J. F., 1989, “A simple parameterization of land surface processes for meteorological models,” Mon. Wea. Rev. , 117: 536-549. Owe M. and Van de Griend A. A., 1998, “C omparison of soil moisture penetration depths for several bare soils at two mi crowave frequencies an d implications for remote sensing,” Water Resour. Res. , 34: 2319-2327. Paloscia S., Macelloni G., Santi E., and Ko ike T., 2001, “A multifre quency algorithm for the retrieval of soil moisture on a large scale using microwave data from SMMR and SSM/I satellites,” IEEE Trans. Geosci. Remote Sensing , 39: 1655-1661. Parde M., Wigneron J. P., Waldteufel P., Ke rr Y. H., Chanzy A., Sobjaerg S. S., and Skou N., 2004, “N-parameter retrievals from L-band microwave observations acquired over a variet y of crop fields,” IEEE Trans. Geosci. Remote Sensing , 42: 1168-1178. Peres L. F. and DaCamara C. C., 2004, “Land surface temperature and emissivity estimation based on the two-temperature method: Sensitivity analysis using simulated MSG/SEVIRI data,” Remote Sens. Environ. , 91: 377-389. Pham H., Kim E. J., and England A. W., 2005, “An analytical calibration approach for microwave polarimetric radiometers,” IEEE Trans. Geosci. Remote Sensing , 43: 2443-2451. Pruski F. F. and Nearing M. A., 2002, “C limate-induced changes in erosion during 21st century for eight US locations,” Water Resour. Res. , 38(12). Reichle R. H., Entekhabi D., and McLaugh lin D. B., 2001a, “Downscaling of radio brightness measurements for soil moisture estimation: A four-dimensional variational data assimilation approach,” Water Resour. Res. , 37: 2353-2364.

PAGE 141

124 Reichle R. H., McLaughlin D. B., and Entekhabi D., 2001b, “Variational data assimilation of microwave radiobrightn ess observations for land surface hydrology applications,” IEEE Geosci. Remote Sensing , 39: 1708-1718. Reichle R. H., McLaughlin D. B., a nd Entekhabi D., 2002a, “Hydrologic data assimilation with the ensemble Kalman filter,” Mon. Weather Rev. , 130: 103-114. Reichle R. H., Walker J. p ., Koster R. D., and Houser P. R., 2002b, “Extended versus ensemble Kalman filtering for land data assimilation,” J. Hydrometeorol. , 3: 728740. Reichle R. H. ad Koster R. D., 2003, “A ssessing the impact of horizontal error correlations in background fields on soil moisture estimation,” J. Hydrometeorol. , 4: 1229-1242. Reichle R. H., Koster R. D., Dong J. R., a nd Berg A. A., 2004, “Global soil moisture from satellite observations, land surface models, and ground data: implication for data assimilation,” J. Hydrometeorol. , 5: 430-442. Reichle R. H. and Koster R. D., 2004, “Bias reduction in short reco rds of satellite soil moisture,” Geophys. Res. Lett. , 31. Reichle R. H. and Koster R. D., 2005, “Gl obal assimilation of satellite surface soil moisture retrievals into the NASA Catchment land surface model,” Geophys. Res. Lett. , 3. Rodell M., and Houser P. R., 2004, “Evaluat ion of 10 methods for initializing a land surface model,” J. Hydrometeorol. , 6: 146-155. Rose L. A., Asher W. E., Reis ing S. C., Gasier P. W., St. Germain K. M., Dowgiallo D. J., Horgan K., Farquharson G., and Knapp E. J., 2002, “Radiometric measurements of the microwave emissivity of foam,” IEEE Trans. Geosci. Remote Sensing , 40: 2619-2625. Roy L., Leconte R., Brissette F. P., and Marc he C., 2001, “The impact of climate change on seasonal floods of southern Quebec river basin,” Hydrol. Proc. , 15: 3167-3179. Ruf C. S., Keihm S. J., Subramanya B., and Janssen M. A., 1994, “TOPEX/POSEIDON microwave radiometer performance and in-flight calibration,” J. Geophys. Res. , 99: 24915-24926. Ruf C. S., Keihm S. J., and Jansse n M.A., 1995, “TOPEX /Poseidon Microwave Radiometer (TMR). I. Instrument descrip tion and antenna temperature calibration,” IEEE Trans. Geosci. Remote Sensing , 33: 125-137. Ruf C.S., 2000, “Detection of calibration dr ifts in space borne microwave radiometers using a vicarious cold reference,” IEEE Trans. Geosci. Remote Sensing , 38: 44-52.

PAGE 142

125 Ruf C. S. and Li J., 2003, “A correlated noise calibration standard for interferometric, polarimetric, and autocorrelation microwave radiometers,” IEEE Trans. Geosci. Remote Sensing , 41: 2187-2196. Schmidt M. and Glade T., 2003, “Linking globa l circulation model outputs to regional geomorphic models: A case study of la ndslide activity in New Zealand,” Clim. Res. , 25: 135-150. Schmugge T. J., 1985, “Remote sensing of soil moisture,” in Hydrological Forecasting , ed. By M. G. Anderson and T. P. Burt , pp. 101-124, John Wiley, New York, NY. Schmugge T. J., Wang J. R., and Arsar G., 1985, “Results from the pushbroom microwave radiometer flight s over Konza prairie in 1985,” IEEE Trans. Geosci. Remote Sensing , 26: 590-596. Schneeberger K., Stamm C., Matzler C., and Fluhler H., 2004, “Ground-based dualfrequency radiometry of bare soil at high temporal resolution,” IEEE Trans. Geosci. Remote Sensing , 42: 588-595. Schwank M., Matzler C., Guglielmetti M., and Fluhler H., 2005, “L-band radiometer measurements of soil water under growing clover grass,” IEEE Trans. Geosci. Remote Sensing , 43: 2225-2237. Shepherd A. and McGinn S. M., 2003, “Assessment of climate change on the Canadian prairies from downscaled GCM data,” Atmos. Ocean. , 41(4): 300-316. Shi J., Chen K. S., Li Q., Jackson T. J., O'Neill P. E., and Tsang L., 2002, “A parameterized surface reflectivity mode l and estimation of bare-surface soil moisture with L-band radiometer,” IEEE Trans. Geosci. Remote Sensing , 40 (12): 2674-2686. Skou N., 1989, Microwave Radiometer Sy stems: Design and Analysis , Norwood, MA: Artech House, 13-19. Tien K. C., Judge J., Miller L., and Lanni O., 2005, “Field data report for the first Microwave Water and Energy Balance E xperiment (MicroWEX-1): July 17 – December 16, 2003,” Circular Number: 1470 , Center for Remote Sensing, UF/IFAS, EDIS: http://e dis.ifa.ufl.edu/AE288. Tien K. C., De Roo R. D., Judge J., a nd Pham H., 2006, “Comparison of calibration techniques for ground-base d C-band radiometers,” IEEE Geosci. Remote Sensing Letters , in press. Tien K. J. and J. Judge, 2006, “Passive micr owave remote sensing of soil moisture for growing vegetation,” Appl. Engi. Agri. , in press. Ulaby F. T., Moore R. K., and Fung A. K., 1981, Microwave Remote sensing: Active and Passive, vol. I , Artech House, Norwood, MA.

PAGE 143

126 Ulaby F. T., Moore R. K., and Fung A. K., 1982, Microwave Remote Sensing: Active and Passive, vol. II , Artech House, Norwood, MA. Ulaby F. T., Moore R. K., and Fung A. K., 1986, Microwave Remote Sensing: Active and Passive, vol. III , Artech House, Norwood, MA. Ulaby F. T. and El-Rayes M. A., 1987, “M icrowave dielectric spectrum of vegetationPart II: Dual-dispersion model,” IEEE Trans. Geosci. Remote Sensing , vol. GE-25, 5: 550-556. U.S. Cotton Market, 2002, Monthly Economic Letter, Cotton Incorporated ~ Market Research, June 12. Voldoire A. and Royer J. F., 2004, “Tropica l deforestation and c limate variability,” Clim. Dynam. , 22(8): 857-874. Von Bloh W., Romano M. C., and Thiel M., 2005, “Long-term predictability of mean daily temperature data,” Nonlinear Proc. Geoph. , 12(4): 471-479. Walker J. P. and Houser P. R., 2004,”Requireme nts of a global near-surface soil moisture satellite mission: accuracy, repeat time, and spatial reslution,” Adv. Water Resour. , 27(8): 785-801. Wang J. R. and Choudhury B. J., 1981, “Remote Sensing of soil moisture content over bare field at 1. 4 GHz frequency,” J. Geophy. Res. Ocean Atmo. , 86: 5277-5282 Wen J., Jackson T. J., Bindlish R., and Hsu A. Y., 2005, “Retrieval of soil moisture and vegetation water content using SSM/I data over corn and soybean region,” J. Hydromet. , 6: 854-863. Wigneron W. P., Calvet J. C., and Kerr Y ., 1996, “Monitoring water interception by crop fields from passive mi crowave observations,” Agri. Forest Meteol. , 80, 177-194. Wood E.F., Lettenmaier D., and Zart arian V., 1992, “A surface hydrology parametrization with subgrid variability in general circulation models,” J. Geophys. Res. , 97(D3): 2720-2728. Xue Y., Sellers P., Kinter J., and Shukla J., 1991, “A simplified biosphere model for global climate studies,” J. Clim. , 4: 345-364. Yan F., Lin T., and Judge J., 2006, “Dyna mic simulation of land surface fluxes for cotton,” Proc. AGU Joint Assembly 2006 , Baltimore, MD, May 23-26.

PAGE 144

127 BIOGRAPHICAL SKETCH Kai-Jen Calvin Tien was born in Taipei, Ta iwan. He received his bachelor’s degree from the Department of Agricultural Engin eering at the National Taiwan University in 1999. He received his master’s degree from the Department of Agricultural and Biological Engineering at the University of Florida in 2001. For his master’s thesis, Calvin Tien conducted research in land use/la nd cover classification and change detection using optical and thermal remote sensing obser vations obtained from satellites. In the fall of 2001, he began his Ph.D. studies in micr owave remote sensing. After six years of research, he graduated from the University of Florida with a Do ctor of Philosophy in agricultural and biolog ical engineering.