<%BANNER%>

Svat Calibration of Point and Regional Scale Water and Energy Dynamics


PAGE 1

SVAT CALIBRATION OF POINT AND REGIONAL SCALE WATER AND ENERGY DYNAMICS By BRENT WHITFIELD A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Brent Whitfield

PAGE 3

ACKNOWLEDGMENTS I would sincerely like to thank Dr. Jennifer Jacobs, chairperson of my supervisory committee, for all of her guidance and support throughout the project. I would also like to thank my committee members, Dr. Jasmeet Judge and Dr. Kirk Hatfield, for all of their valuable suggestions and encouragement. I am very appreciative for the help of Dr. Paul Houser and Dr. Matthew Rodell from NASAs Goddard Space Flight Center in Greenbelt, Maryland. I am grateful to Gerard Ripo, Sudheer Reddy Satti, Shirish Bhat, Aniruddha Guha, Enching Hsu, Siqing Liu and Jeremy Andrews for all of their help in data collection, analysis and peer review. I am indebted to my girlfriend, Erica Zingone, for all of her encouragement and for providing me with a justification for taking monthly vacations to Atlanta, Georgia. I am grateful for my parents, Randy and Martha Whitfield, for the emotional and financial support afforded me throughout my education at the University of Florida. I would also like to thank my dog Jake Whitfield for his consistent allegiance and watchful eye over all of my work. Research assistantship was provided by NASA NIP Grant NAG5-10567. iii

PAGE 4

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES...............................................................................................................v LIST OF FIGURES...........................................................................................................vi ABSTRACT......................................................................................................................vii CHAPTER 1. INTRODUCTION...........................................................................................................1 2. MODEL COMPARISONS..............................................................................................6 3. STUDY AREA..............................................................................................................11 4. SIMULATION DESIGN...............................................................................................15 Parameterization.........................................................................................................15 Initialization................................................................................................................16 Forcings......................................................................................................................17 5. RESULTS AND DISCUSSION....................................................................................18 Soil Moisture..............................................................................................................18 Soil Temperature........................................................................................................22 Surface Heat Fluxes....................................................................................................25 6. CONCLUSION..............................................................................................................34 LIST OF REFERENCES...................................................................................................36 BIOGRAPHICAL SKETCH.............................................................................................42 iv

PAGE 5

LIST OF TABLES Table page 1. Methodology for parameterization for CLM and LSP....................................................7 2. Characterization of land surface processes for CLM and LSP........................................8 3. Instrumentation used in this study.................................................................................14 4. Required atmospheric forcings for CLM and LSP........................................................17 v

PAGE 6

LIST OF FIGURES Figure page 1. Log scale plot of variation of parameterized saturated hydraulic conductivity with depth below the surface for CLM and LSP................................................................9 2. Water table depth with time from April 10 th the June 20 th Experiment period from May 4 th to May 17 th ..................................................................................................12 3. Comparison of measured soil water characteristic curve with three soil water characteristic curves defined by modeled soil matric potential (SMP) and seven hydraulic conductivity (K) curves based on the Clapp and Hornberger relationship within the CLM parameterization scheme...............................................................16 4. Comparison of measured soil moisture with CLM and LSP modeled soil moisture at a) 7.6 cm, b) 12.7 cm and c) 17.8 cm...........................................................................20 5. Comparison of measured subsurface soil temperature and CLM and LSP modeled soil temperature at a) 2.5 cm, b) 7.6 cm, c) 12.7 cm and d) 17.8 cm..............................24 6. Comparison of modeled and measured surface fluxes for a) CLM net radiation and b) CLM and LSP latent heat flux..................................................................................30 7. Comparison of modeled and measured surface fluxes for a) CLM and LSP sensible heat flux and b) CLM and LSP ground heat flux.....................................................31 8. Comparison of measured and modeled surface energy fluxes for a typical cloudy day, May 7 th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat fluxes....32 9. Comparison of measured and modeled surface energy fluxes for a typical clear day, May 13 th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat fluxes..33 vi

PAGE 7

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering SVAT CALIBRATION OF POINT AND REGIONAL SCALE WATER AND ENERGY DYNAMICS By Brent Whitfield May 2003 Chair: Jennifer Jacobs Major Department: Civil and Coastal Engineering The Community Land Model (CLM) and the Land Surface Process model (LSP) are typical soil vegetation atmosphere transfer models (SVAT) that were used to simulate the land surface processes for a wet prairie community in the southeastern US for May 4-17, 2001. The research objective is to provide a basis for calibration of SVAT modeling processes for the highly variable convective atmospheric conditions and the atypical hydrogeology of the southeastern US. Since CLM is a watershed or regional scale model and LSP is a field scale model, each model has distinctive parameterization schemes and methodologies for simulating land surface processes. The intercomparison of CLM and LSP provides perspective for the strengths and weaknesses of the generalized parameterization schemes of a regional scale model as compared to a more biophysically adept field scale model. The study area is Paynes Prairie State Preseve in north-central Florida, which can be characterized by the biological communities of freshwater marsh, wet prairie and pasture. vii

PAGE 8

Validation data were provided using standard meteorological instrumentation, an eddy correlation system for measurement of surface heat fluxes and subsurface measurements of moisture, temperature and matric potential. The CLM and LSP simulations of soil moisture at 7.6, 12.7 and 17.8 cm compared well with the measured results showing a RMSE less than 3.5% volumetric water content for both models at all three depths. Soil temperature simulations corresponding to 2.5, 7.6, 12.7 and 17.8 cm show good agreement with measured temperatures with a RMSE less than 1.5 K for all four depths in both simulations. Simulation of the components of the energy balance showed different trends for CLM than LSP. CLMs simulation showed the strongest agreement with the measured latent heat fluxes (RMSE of 62 Wm -2 ) as compared with the simulation of sensible (RMSE of 61 Wm -2 ) and ground heat fluxes (RMSE of 27 Wm -2 ). The LSP simulation of latent heat flux showed the poorest agreement (RMSE of 122 Wm -2 ) as compared with the simulation of sensible (RMSE of 29 Wm -2 ) and ground heat fluxes (RMSE of 5 Wm -2 ). An investigation of model performance on a typical cloudy day (May 7 th ) and a typical clear day (May 13 th ) demonstrated potential for the application of both CLM and LSP in the highly variable climatic conditions of the southeastern US. viii

PAGE 9

CHAPTER 1 INTRODUCTION Improved quantification of the energy, moisture, and momentum fluxes between the atmosphere and the Earths surface remains a critical component for studies in hydrology, meteorology and biogeoscience. Scientific endeavors, such as climate modeling, are often interested in quantifying surface fluxes at large spatial scales over a variety of land cover types that are not feasibly monitored using field based techniques. In the stead of complex direct measurement, soil-vegetation-atmosphere transfer (SVAT) models are often used to quantify these fluxes and stores. The most suitable characteristics for SVAT models are often determined by their desired application. The divergence in applications of SVAT models has lead to a variety of models that have structural differences for the characterization of land surface processes. Only by comparing the results of separate SVAT models with a known set of validation data can the utility of different modeling schemes be determined. The applicability of a given SVAT model is significantly dependent on its required parameterizations and model structure (Schaake et al. 2001). Increased parameterization allows for a greater description of the physical environment and theoretically creates a more robust simulation of surface fluxes. However, this increased robustness often leads to a decreased flexibility in usage. Often a less sophisticated model parameterization can prove advantageous, due to the characteristic heterogeneity of the land surface, as well as a general scarcity of insitu measurement of physical parameters. An effective SVAT 1

PAGE 10

2 model should incorporate adequate sophistication to accurately simulate the environment, but not be so complex as to reduce usability. In the Project for Intercomparison of Land surface Parameterization Schemes or PILPS, several scientific teams explored the value of different methods of model parameterization (Pitman and Henderson-Sellers 1998). The multiphase PILPS compared results from as many as 23 land surface models including BATS (Dickinson et al. 1993) and VIC-2L (Liang et al. 1996). The PILPS focus was a community-wide intercomparison of modeling schemes. Thus, none of the studies provided a detailed analysis of individual models. However, the intercomparisons did present results that can be used to guide more detailed intercomparisons (Chen et al. 1997). PILPS Phase 1 comparisons of annually-averaged latent heat flux, sensible heat flux, and radiative temperature, as well as monthly-averaged surface runoff and soil moisture with surface fluxes generated from NCAR climate data showed poor agreement in tropical forest and grassland settings. The revised experimentation framework demonstrated the importance of congruent forcings and parameterization schemes in model comparison studies (Pitman et al. 1999). Phase 2 provided a similar comparison using observed climate data from four different sites, the Cabauw experimental station in the Netherlands (Chen et al. 1997), the HAPEX-MOBILHY site in southwest France (Shao and Henderson-Sellers 1996), the Red-Arkansas River Basin in the Midwestern US (Liang et al. 1998), and a boreal site in Valdai, Russia (Slater et al. 2001). PILPS Phase 2 identified the differences in land surface model mechanisms and characterized the variability in the modeled results. Shao and Henderson-Sellers (1996) describe the difficulty in identifying the link between the models performance and the responsible mechanism. Because land surface

PAGE 11

3 models are non-linear systems with interacting components, the range of results in each of the Phase 2 intercomparisons is larger than acceptable measurement error. All of the Phase 2 intercomparisons, except those examining snow-covered terrain, exhibited a smaller disparity in latent heat fluxes with respect to the average annual measured latent heat flux than both the sensible heat flux and the ground heat flux (Shao and Henderson-Sellers 1996, Chen et al. 1997, Liang et al. 1998). PILPS demonstrated modeling schemes strengths in simulating general patterns of land surface processes. The overall scatter in simulation results, which existed between any set of models, may be categorized based on the structure of modeling schemes. However, PILPSs investigation determined that within categories of like modeling schemes there exists varying levels of sensitivity to parameterization (Slater et al. 2001). While PILPS characterized the overall status of SVATs, model studies that include fewer schemes allow more detailed examination of the models strengths and weaknesses for a variety of landscapes and climatic conditions. The Mohr et al. (2000) study of the untuned PLACE model in the Southern Great Plains experiment effectively simulated the near surface temperature, moisture and land surface fluxes in the first 48 hours following a heavy rainfall. However, they diverged from observations during the drying period due to simplifications in the parameterization of subsurface moisture, vegetation cover and soil texture. For the prairie grassland in the Radiobrightness Energy Balance Experiment (REBEX) near Sioux Falls, SD, the LSP/R model exhibited small errors when comparing predicted and observed soil and canopy temperatures in a northern prairie over a 14-day period in October (Liou et al. 1999). Energy fluxes were not examined. The van der Keur et al. (2001) study of the modified DAISY SVAT model had good agreement

PAGE 12

4 between the modeled and observed land surface fluxes in non-water limiting conditions for winter wheat in Jutland, Denmark. However, its application during water limiting conditions revealed that the simulated latent heat fluxes underestimated observed values and required a more complete parameterization of canopy resistance during stressed conditions. Nijssen et al. (1997) demonstrated good seasonal agreement with observed fluxes using the hydrology-soil-vegetation model DHSVM in forested sites in the Boreal region of Canada. However, the Nijssen simulation had a time lag for the simulated sensible heat fluxes that suggests the need for a more complete parameterization scheme for the soil thermal model. The importance of SVAT simulation efforts to provide validation in an array of climactic and vegetative conditions is exhibited in the broad range of locations found in existing research: mid-latitude grasslands and croplands (Wood et al. 1998, Liou et al. 1999, Mohr et al. 2000), boreal forests and arctic tundra (Njissen et al. 1997, Kim 1999, Slater et al. 2001), and tropical forests (Pitman et al. 1999). Although SVAT validation research conducted thus far has been representative of several regions and climate types, global extrapolation requires additional validation research across a greater number of climates and biomes. It is the purpose of this paper to provide a basis for calibration of SVAT modeling processes for the highly variable convective atmospheric conditions and the atypical hydrogeology in the southeastern US. The southeastern US is characterized by subtropical climatic conditions, that consist of high humidity, convective heating and high annual rainfall. The low topographic relief and high water tables create a unique environment for vegetative communities. These communities are distinguished by shallow root zones and variable stomatal resistances and reflectance properties (Mitsch

PAGE 13

5 and Gosselink 2000). A comparison of two SVAT models of contrasting design is used to determine the strengths and weaknesses of each methodology. The Common Land Model (CLM) (Dai et al. 2001) is a regional or watershed scale model, while the Land Surface Process (LSP) model (Liou et al. 1999) is a field scale model. This intercomparison examines the utility of increased sophistication in parameterization as it relates to differences in scale and its practical application in the southeastern US

PAGE 14

CHAPTER 2 MODEL COMPARISONS The Common Land Model (CLM) was developed as part of a multi-disciplinary and multi-institutional project designed to provide land surface forcings for the lower boundary of the Community Climate System Model (Blackmon et al. 2001). Dai et al. (2001) describe the technical elements of CLM. Zeng et al. (2002) and Bonan et al. (2002) demonstrate the coupling of CLM with CCM3. The Land Surface Process (LSP) model is designed to link traditional land surface models and satellite microwave observations to allow for data assimilation (Liou et al. 1999). Judge et al. (2003a and 2003b) describe recent modifications and further developments to the LSP model. CLM and LSP can be distinguished primarily by their model processes and required parameters that reflect their designed application scale. Table 1 summarizes the differences between their parameterization schemes. As CLM was designed to represent single columns with spatial extents as large as by it incorporates a minimum of insitu measurement of model parameters and initialization data. The only parameters that are user-provided are land cover class, soil texture, longitude, latitude, and soil color index. The remaining environmental parameters are derived from an internal parameterization scheme based on land cover classification. Land cover class is specified based on 18 different International Geosphere-Biosphere Programme (IGBP) classifications (Loveland et al. 2000) from which the remaining vegetation parameters are derived. Estimation of surface fluxes from multiple land cover classifications uses a tile6

PAGE 15

7 mosaic approach similar to Koster et al. (2000). Soil parameters, such as hydraulic conductivity and water retention curves, are generated from empirically based derivations using soil texture (Clapp and Hornberger 1978, Cosby et al. 1984). Table 1. Methodology for parameterization for CLM and LSP. ParametersCLMLSPSoil TextureUser DefinedUser DefinedEmpirical CalculationUser DefinedKsatClapp and HornbergerUser DefinedEmpirical CalculationUser DefinedWilting PointUser DefinedUser DefinedWater Retention CurveClapp and HornbergerUser DefinedRoot DepthEmpirical Calculation / IGBPUser DefinedLAIEmpirical Calculation / IGBPUser DefinedCanopy HeightIGBPUser DefinedRoughness LengthIGBPUser Defined In contrast, as LSP was designed as a research tool for field scale applications, it uses a more detailed parameterization scheme and requires insitu measurement of many environmental parameters. Field specific parameters such as soil texture, longitude, latitude, LAI, canopy height, and canopy biomass are user specified. Two physical soil parameters, air entry pressure ( o ) and pore size index (), are required to generate the soil water characteristic curve. LSP uses the Rossi and Nimmo (1994) relationship based on the Brooks and Corey water retention model. The mechanics of the hydrological processes: evapotranspiration, infiltration, and runoff vary significantly between the two models. Table 2 describes these differences for CLM and LSP. Both models determine bare soil evaporation using the Philip and De Vries (1957) diffusion model. CLMs transpiration process is an aerodynamic approach based on the BATS model (Dickinson et al. 1993) with a stomatal resistance component

PAGE 16

8 from the LSM model (Bonan, 1996). LSPs transpiration module uses an aerodynamic approach derived from the CLASS model (Verseghy et al. 1993). LSPs infiltration rate requires soil hydraulic and physical properties, soil temperature, vegetation cover, and surface characteristics like slope and roughness (Ghildyal and Tripathi, 1987). The maximum infiltration rate is estimated using a quasi-analytic solution to Richards equation for vertical infiltration in a homogeneous soil with a constant initial moisture profile (Green and Ampt, 1911, Philip 1957, Philip 1987a, Philip 1987b). Table 2. Characterization of land surface processes for CLM and LSP. ProcessesCLMLSPSoil EvaporationDiffusionDiffusionEvapotranspirationAerodynamic (BATS and LSM based)Aerodynamic (CLASS based)InfiltrationRichard's EquationRichard's Equation (modified)Subsurface Heat TransportFourier's EquationFourier's EquationRunoffSurface runoff and baseflow (TOPMODEL based)Hortonian Flow One of the distinguishing characteristics of CLM is the surface runoff generation process. CLM uses assumptions from the watershed scale model TOPMODEL (Beven et al. 1995) as adapted for land surface modeling (Stieglitz et al. 1997). These assumptions include: a baseflow component defined by the saturated hydraulic conductivity of the lower layers of the profile, an exponential decrease in the saturated hydraulic conductivity with depth and surface runoff based on the parameterization of saturated and unsaturated spatial extents. The exponential decrease of saturated hydraulic conductivity with depth is illustrated in Figure 1 using the conductivities defined by the soil profile used in this study. Figure 1 also illustrates the saturated conductivity used by LSP as parameterized by the soil profile of this study. The parameterization of the fraction of saturated and unsaturated landscape is based on an empirical relationship defined by the relative water table. Since LSP is designed for field scale use only, runoff is only

PAGE 17

9 generated from saturation excess and that depth of water not infiltrated at the end of each timestep. 0.00.51.01.52.02.53.00.00000.00000.00010.00100.01000.10001.0000Saturated Hydraulic Conductivity (mm/s)Depth (m) CLM LSP Figure 1. Log scale plot of variation of parameterized saturated hydraulic conductivity with depth below the surface for CLM and LSP Two modifications were made to CLM. For the low gradient topography and hydrogeology characteristic of much of the southeastern US, the TOPMODEL baseflow generation mechanism is not appropriate. The TOPMODEL baseflow generation mechanism was eliminated allowing the lower layers to remain saturated and eliminating excessive drainage. The parameterization of the relationship between soil moisture and root resistance was also modified for the CLM evapotranspiration mechanism based on measured stresses (Jacobs et al. 2002a). Both CLM and LSP discretize the soil profile into layers. Each allow soil properties to differ by layer. CLM has a 10-layer profile with the thickness of each layer determined by a unitless scaling factor and an exponential function that increases with depth. Given a typical vertical scaling factor of 0.025, the thickness of the uppermost

PAGE 18

10 layer of CLM is 1.75 cm, while the thickness of the lowest layer is 113.7 cm. LSP layers thicknesses also increase with depth. However, the LSP profile has 60 layers. The top layers are very fine, yet a deep soil layer may be simulated. Each layers thickness may be defined as appropriate for the application. CLM and LSP also have fundamental computational differences. CLM determines the water movement across layer interfaces using a first-order Taylor expansion and solving the resulting equation using a tridiagonal matrix solution. The flux of heat across CLM layer interfaces is solved using the Crank-Nicholson numerical scheme and a tridiagonal matrix solution (Dai et al. 2001). These methods generate soil moisture and temperature profiles at the same temporal resolution as the timestep of the simulation. LSP uses a block-centered finite difference approach that determines the temporal resolution based on the parameterized convergence criteria. The temporal resolution of the resulting LSP profile also differs from the timestep of the climate forcings.

PAGE 19

CHAPTER 3 STUDY AREA The Paynes Prairie State Preserve is a regional basin that contains biological communities such as freshwater marsh, wet prairie and pasture. This 5600 ha system in north-central Florida, USA is 13 km long (east-west) and ranges in the north-south dimension from 1.5 km to 7 km. The summer climate conditions are strongly influenced by surface heating with significant cumulus cloud cover and few cloud-free days. The basin is a large irregularly shaped bowl resulting from solution of the underlying limestone. The basin is either seasonally or perennially flooded based on a surficial aquifer that is separated from the highly transmissive Floridan aquifer by the underlying Hawthorne formation. This geologic formation acts as a semi-confining layer. The primary drain for the Preserve is a sinkhole breach (Alachua Sink) in the semi-confining layer. Based on rainfall patterns, two water-control structures are used to simulate natural basin water levels annually and inter-annually. Generally, water levels are highest during the summer months and lowest during the months of April, September, October, and November. However, no diversion of water into the Preserve through the control structures occurred during the study period due to a multi-year drought. The study was conducted in a wet prairie community located in the Paynes Prairie Preserve (29 o 34 14 N, 82 o 16 46 W). The study period is a 13-day rainless span from May 4 th until May 17 th of 2001. The study period occurs within a 40-day dry down from saturated conditions from April 17 th until May 27 th There is no precipitation during the 40-day dry down period with the exception of three storm events, of which only one 11

PAGE 20

12 is notable in depth. The first event is a 2.3 cm rainfall on April 27 th The second and third precipitation events are both 0.25 mm in depth and occur on May 2 nd and May 17 th As is seen in Figure 2, the water table depth reflects these rainfall patterns. The experiment period is during a steady dry down condition in which the depth to the water table increased from 62 cm to 86 cm. -1.4-1.2-1.0-0.8-0.6-0.4-0.20.0100110120130140150160170180Time (Day)Depth (meter) Experiment Period Figure 2. Water table depth with time from April 10 th the June 20 th Experiment period from May 4 th to May 17 th The prairie is a relatively flat, treeless plain with moderately dense ground cover. The mean canopy height during the study period is approximately 1.0 m. The sites soils include Emeralda fine sandy loam, Wauberg sand, and Ledwith Muck. The soils consist of sands with an organic surface layer that are underlain by clay. Field observations showed that the majority of the root zone was contained in the upper 10-cm soil layer with approximately 95% of the root zone contained in the upper 25-cm soil layer.

PAGE 21

13 Typically, the wet prairie is inundated for 50 to 100 days each year, burns every two to four years, and has low organic matter acculumulation. This prairie community is routinely burned in accordance with simulated natural periodicity in order to maintain the communities natural conditions and ecological diversity. Prior to the experiment, the study area was most recently burned in December 1999. The site is instrumented with a tower-based meteorological station installed onsite to provide several meteorological and micrometeorological measurements. Instrumentation onsite also provided subsurface measurements of water content, soil temperature and soil matric potential. Table 3 describes the instrumentation used for data collection. Jacobs (2002b) also provides a complete description of the instrumentation and study area. All atmospheric and flux instrumentation was mounted on a 6.1-m tower, and logged using a CR10X datalogger (CSI, Inc.). Land surface fluxes were measured using an eddy-flux approach. Evapotranspiration was measured directly using an energy-budget variant of the eddy correlation approach (Tanner and Greene 1989, Twine et al. 2000). Fluctuations in wind speed, virtual air temperature, and vapor density were sampled at 6 Hz, and 30-minute average covariances were calculated to estimate the fluxes. The latent heat fluxes were corrected for temperature-induced fluctuations in air density (Webb et al. 1980) and for the hygrometer sensitivity to oxygen (Tanner and Greene 1989). Sensible heat fluxes were corrected for differences between the sonic temperature and the actual air temperature (Schotanus et al. 1983). Both the sensible and latent heat fluxes were corrected for misalignment with respect to the natural wind coordinate system (Baldocchi et al. 1988). The Bowen-ratio method was used to close the surface energy balance relationship (Twine et al. 2000). Ground heat flux was measured

PAGE 22

14 approximately 10 cm below the surface. The measured soil heat flux was corrected for the energy stored in the upper 10 cm of soil using measurements of soil temperature. Near-surface volumetric soil water content was recorded at three depths: 7.6 cm, 12.7 cm and 17.8 cm. Soil water potential and soil temperature were also measured within the soil profile at points coinciding with the moisture probe depths, as well as an additional temperature measurements at 2.5 cm. Subsurface measurements were sampled every minute, and 30-minute averages were logged using an AM 416 multiplexer coupled with a CR10X datalogger (CSI, Inc.). Table 3. Instrumentation used in this study VariableInstrumentationHeight (m)Net RadiationREBS Q*7.16.5Sensible Heat FluxCSI 3-D Sonic Anemometer CSAT35.75Latent Heat FluxCSI Krypton Hygrometer KH205.75Ground Heat FluxREBS RFT 3.1-0.1Wind Speed and DirectionRM Young CS 800-L Anemometer6.5PrecipitationTexas Electronincs TE5256.5Relative HumidityVaisala HMP45c5.8TemperatureVaisala HMP45c5.8PressureVaisala PTB1001.0Subsurface TemperatureCSI 107-0.025,-0.076, -0.127, -0.178Subsurface MoistureCSI 615L-0.076, -0.127, -0.179Subsurface Matric PotentialCSI 257-0.025,-0.076, -0.127, -0.178

PAGE 23

CHAPTER 4 SIMULATION DESIGN Parameterization CLM and LSP soil parameters are characterized by soil texture composition from percentages of sand, clay, and loam. The study areas soil is predominantly Wauberg Sand. Parameters for CLM and LSP soil layers were determined by a weighted average based on the soil composition from the Soil Characterization Laboratory soil composition profile (UF-IFAS 1985). Figure 2 illustrates a comparison of the soil water characteristic curves generated from the Clapp and Hornberger relationship used by CLM with soil water characteristic curves derived from measured values. Figure 2 also includes the hydraulic conductivity as it varies with volumetric water content for each of the five CLM layers contained in the top 23 cm of the profile. Vegetation parameterization for CLM is based on the IGBP characterization of landscapes. The Land Process DAAC (edcdaac.usgs.gov/main.html) defines the study area as a cropland using the IGBP land classification. This characterization, derived from remote sensing data, may be indicative of agricultural pastures bordering the study area, but does not provide an accurate description of the vegetative communities in the region. Due to the hydrogeology and climatic conditions of Paynes Prairie, the plant species within the flooded prairie community are predominately grassland and wetland species. Therefore, the IGBP land cover classification chosen for CLM parameterization was wetland, while within the IGBP parameterization scheme the grassland parameters for leaf and stem reflectance and transmittance were used. CLMs leaf area index (LAI) 15

PAGE 24

16 parameters are generated from annual maximum and minimum values based on IGBP wetland properties. The temperatures of the lowest soil layer define the seasonal relationship of LAI. LSPs vegetation parameters are user defined. Here, LSPs LAI was proportionally increased during the experiment from 2.0 to 2.5, 0501001502002503003504000.050.100.150.200.250.300.350.40 (cm)0.0000.0050.0100.0150.0200.0250.030K (mm s-1) Measured SMP 0-9 cm depth SMP 9-16.5 cm depth SMP 16.5-23 cm depth K 1.76-4.51 cm depth K 0-1.75 cm depth K 4.52-9.05 cm depth K 9.06-16.56 cm depth K 16.57-23 cm depth Figure 3. Comparison of measured soil water characteristic curve with three soil water characteristic curves defined by modeled soil matric potential (SMP) and seven hydraulic conductivity (K) curves based on the Clapp and Hornberger relationship within the CLM parameterization scheme. Initialization The CLM and LSP initialization schemes require the initial soil temperature profile and the initial soil moisture content profile. The soil temperature profile is intialized using measured values for the top 23 cm. For the lower profile, the initial temperature is set equal to the temperature measured at 23 cm. The soil moisture profile is initialized using measurements in the top 17 cm of the profile. The profile is saturated below 62 cm. The initial soil moisture is assumed to increase linearly with depth between 17 cm and the water table.

PAGE 25

17 Forcings Table 4 summarizes the forcings required by CLM and LSP. All forcings excluding radiation, are from measured values. Estimates of incoming shortwave and longwave radiation were derived from measured net radiation. During the nighttime (generally 6:30 PM to 6:30 AM EDT), shortwave radiation is zero and net longwave radiation is equivalent to measured net radiation. Daytime values of the longwave radiation balance were estimated using Diak et al.s (2000) remotely-sensed radiation estimation scheme based on GOES data. Jacobs et al. (2002b) demonstrated the utility of GOES data in the study area. The diurnal radiation surface albedo as a function of time was determined for a cloud-free day using measured incoming shortwave radiation, measured net radiation and GOES estimated net longwave. The solar insolation is calculated from measured net radiation, GOES derived net longwave radiation and surface albedo. Table 4. Required atmospheric forcings for CLM and LSP CLMLSPIncoming Shortwave Radiation (Wm-2)Incoming Shortwave Radiation (Wm-2)Incoming Longwave Radiation (Wm-2)Incoming Longwave Radiation (Wm-2)Air Temperature (K)Air Temperature (K)Specific Humidity (-)Relative Humidity (%)Precipitation (mm 0.5hr-1)Precipitation (mm 0.5hr-1)Wind Speed (ms-1)Wind Speed (ms-1)Atmospheric Pressure (hPa)Model Forcings

PAGE 26

CHAPTER 5 RESULTS AND DISCUSSION The CLM and LSP simulation results are evaluated using three categories. The modeled soil moisture, soil temperature and surface heat fluxes are compared with measured validation data. The two main statistical quantities used for simulation assessment are mean absolute error (MAE) and root mean squared error (RMSE). The mean absolute error is defined as the average of the absolute differences between modeled and measured results: nXXMAEn1ii,measimod, Equation 1. Mean absolute error. The root mean squared error is the square root of the average of the differences between modeled and measured results squared, as follows: nXXRMSEn1i2i,measi,mod Equation 2. Root mean squared error. Soil Moisture The observed point measurements of soil moisture were compared to the simulated moisture for the layer that contains that measurement point. Each CLM and LSP soil 18

PAGE 27

19 layer has homogeneous moisture and temperature. The soil moisture measurements at 7.6, 12.7 and 17.8 cm correspond to CLM layers 3, 4 and 5, respectively and to LSP layers 11, 15 and 19. Figure 3 shows the evolution of modeled and measured soil moisture during the study period in terms of volumetric water content (VWC). CLMs layer 3 exhibits a more rapid dry down than the observed dry down. This layer has a higher modeled water content than observed during the first two days of the experiment. The model dry down rate results in the simulated layer having a lower moisture content than the observed for all periods after four days with an increasing bias for the duration. The MAE and RMSE for the CLM simulated results are 0.013 m 3 m -3 and 0.016 m 3 m -3 LSP matches the dry down rate of the measurement at 7.6 cm, however, there is an initial offset in the modeled results of approximately 0.03 m 3 m -3 This initial offset results in a high bias of simulated moisture contents for the entire experimental period. The LSP simulated results demonstrate a MAE and RMSE of 0.034 m 3 m -3 and 0.035 m 3 m -3 The CLM simulation results corresponding to the 12.7 cm measurement demonstrate good agreement with the measured dry down rate. A small bias results in simulated moisture content values that are lower than measured by approximately 0.03 m 3 m -3 throughout the study period. The LSP model shows very good agreement with the measured values over the period. The MAE and RMSE for the LSP modeled results are 0.009 m 3 m -3 and 0.011 m 3 m -3 Both CLM and LSP show a similar rate of dry down for the measurements at 12.7 and 17.8 cm and provide good moisture content estimates at the deepest measurement depth. CLMs moisture contents are slightly drier (0.01 m 3 m -3 ) than measured, while LSPs modeled water contents are slightly wetter (0.02 m 3 m -3 ).

PAGE 28

20 0.050.10.150.20.250.3124125126127128129130131132133134135136137Time (days) at 7.6cm Measured CLM Modeled LSP Modeleda) MAE RMSE CLM 0.013 0.016 LSP 0.034 0.035 0.050.10.150.20.250.3124125126127128129130131132133134135136137Time (days)at 12.7 cm Measured CLM Modeled LSP Modeledb) MAE RMSE CLM 0.032 0.033 LSP 0.009 0.011 0.050.10.150.20.250.3124125126127128129130131132133134135136137Time (days)at 17.8 cm Measured CLM Modeled LSP Modeledc) MAE RMSE CLM 0.011 0.013 LSP 0.020 0.022 Figure 4. Comparison of measured soil moisture with CLM and LSP modeled soil moisture at a) 7.6 cm, b) 12.7 cm and c) 17.8 cm.

PAGE 29

21 The measurement error associated with a CS615 moisture probe is 2.5% volumetric water content. CLM simulations at 7.6 and 17.8 cm depths have errors less than the instrument error. The CLM simulation at 12.7 cm is less than 1.5 times the instrument error. LSPs 7.6 cm depth exhibits error of less than 1.5 times the instrument error. LSPs 12.7 cm and 17.8 cm depths show better agreement with an error less than the instrument error. The volumetric soil moisture errors identified in this study are consistent with the results of other studies. In PILPS Phase 2(b), Shao and Henderson-Sellers (1996) recognized % volumetric water content error margins as reasonable, while Mohr et al. (2000) found error margins in the near surface soil simulation of % volumetric soil moisture. Comparisons of the modeled results to the observations suggest different strengths and weaknesses for each model. For all three depths, both models capture the diurnal fluctuations in moisture. The observations showed increasing soil moisture as the soil water profile reestablished during the evening. LSP did an excellent job of modeling the phase and amplitude of the measured soil moisture fluctuations throughout the experiment period. Both are able to capture the magnitude and location of soil water extraction due to evapotranspiration. While, CLMs daytime dry down appears to function appropriately, CLM was unable to replicate the recharge from the lower layers. The water retention curves used by the models appear to result in two distinct errors. For LSP, the soil water profile rapidly equilibriates to a biased profile. The LSP soil water retention curve appears shifted such that the soil matric potential for each depth corresponds to a higher moisture content than the moisture content that is measured. CLM overestimates the daily dry down rate, because it does not redistribute water from

PAGE 30

22 lower layers to upper layers during the nighttime. This pattern can be explained by examining the CLM soil water characteristic curves in Figure 2. For soil moisture contents between 0.15 m 3 m -3 and 0.17 m 3 m -3 the slope of the measured retention curve is very steep, however, the CLM water retention curve is relatively flat in this range. Daytime root water extraction results in an observed soil water profile that increases in depth from the surface. This profile results in a large gradient of matric potentials in response to the moisture content profile that drives soil water upward in the soil profile during the evening. However, in CLMs modeled soil column, comparable moisture profiles exhibit a much smaller gradient of matric potential. Thus, CLM has a reduced upward movement of moisture and increasingly dryer surface conditions. Soil Temperature The comparison of the modeled soil temperature profile with observed soil temperatures follows that methodology for the soil moisture comparison. Soil temperature and moisture were measured at the same depths with an additional temperature measurement at 2.5 cm depth. The time evolution of modeled CLM and LSP soil temperatures compared with measured values for the observation period are seen in Figure 4. Overall, both models show good agreement with the timing and magnitude of the actual diurnal fluctuations. Simulated soil temperatures of both CLM and LSP for all depths show a warming trend over the study period. For the 2.5 cm depth, the maximum daily soil temperature increases by 2 K over the two-week period. However, the simulated maxima are somewhat warmer than the measured maximum during the latter parts of the experiment. For the second depth, the maxima of both simulations are within 1 K of the measured maximum early in the experiment and greater than 3.0 K and 2.0 K for CLM and LSP, respectively, at the end of the experiment. At the 7.6 cm

PAGE 31

23 measurement depth, both simulations lead the measured diurnal fluctuations by 2.0 to 3.0 hours throughout the study period. While the temperatures are quite similar, the shift increases in errors to 1.2 K and 1.4 K for CLM and 0.8 K and 0.98 K for LSP. The 12.7 and 17.8 cm comparisons do not have a temporal discrepancy. Both comparisons exhibit the same enhanced warming trend that is evident at both 2.5 and 7.6 cm, as shown in Figures 4 c and d. At 12.7 cm, the early diurnal cycles show slightly lower modeled temperature maxima than the measured temperature maximum. The agreement improves during the experimental period with the final diurnal cycle exhibiting good agreement as compared to the measurements. The MAE and RMSE are 0.45 K and 0.54 K for CLM and 0.52 K and 0.67 K for LSP. The 17.8 cm measurements compare well with both simulated time series for the first 10 days with the final three cycles showing a small warm bias of 1.5 K between the measured and modeled temperatures at all points during the cycle. The lowest layer had the smallest error for both CLM with a MAE and RMSE of 0.35 K and 0.44 K and LSP with errors of 0.33 K and 0.41 K. For both models, the modeled error decreases with depth, as the diurnal fluctuations are damped by the increasing soil thickness. The measurement error associated with a CS107 temperature probe is 0.5 K. Analysis of the CLM modeled soil temperatures shows an agreement with the 2.5 cm and 7.6 cm measured values within 2.5 times the instrument error. The comparison of the CLM modeled temperatures with the 12.7 cm and the 17.8 cm measured values show errors less than the instrument error. The LSP simulation shows a similar trend. From the 2.5 to 17.8 cm depths, the ratio of LSPs MAEs compared to the instrument errors are less than 2.5, 2.0, 1.5 and 1.0, respectively. The errors of the simulated soil temperatures compare reasonably

PAGE 32

24 285287289291293295297124125126127128129130131132133134135136137Time (days)Temp (K) at 2. Measured CLM Modeled LSP Modeled 285287289291293295297299301303305124125126127128129130131132133134135136137Time (days)Temp (K) at 7.6 cm CLM Modeled Measured LSP Modeledb) MAE RMSE CLM 1.17 1.40 LSP 0.84 0.98 285287289291293295297299301303305124125126127128129130131132133134135136137Time (days)Temp (K) at 12.7 cm Measured CLM Modeled LSP Modeledc) MAE RMSE CLM 0.45 0.54 LSP 0.51 0.67 d)285287289291293295297299301303305124125126127128129130131132133134135136137Time (days)Temp (K) at 17.8 cm Measured CLM Modeled LSP Modeled MAE RMSE CLM 0.35 0.44 LSP 0.33 0.40 MAE RMSE CLM 1.02 1.28 LSP 1.07 1.34 a) 2993013033055 cm Figure 5. Comparison of measured subsurface soil temperature and CLM and LSP modeled soil temperature at a) 2.5 cm, b) 7.6 cm, c) 12.7 cm and d) 17.8 cm.

PAGE 33

25 well with similar modeling studies. Both Chen et al. (1997) and Mohr et al. (2000) found errors within K for subsurface temperatures. The offline simulations of PILPS Phase 1 found ranges between models of 1.4 K for tropical forest and 2.2 K for grassland in a multi-year study. Judge et al. (2003b) found a 1.8 K average difference between modeled and measured temperature for bare soil and a 1.0 K mean difference for brome grass. In general, the modeled temperatures for both models show a warming trend over the duration of the period at all depths. This trend reflects the observed dry down and warming conditions. CLMs simulated diurnal temperature fluctuations are greater in magnitude than LSPs. This likely reflects the decreased thermal heat capacity of CLMs relatively drier soil. Despite the higher moisture content, the LSP layers temperatures generally show higher fluctuations in temperature than the measurements. This could be due to the soil texture parameterized thermal capacity of dry soil within LSP being lower than the actual thermal capacity of the soil solids. Surface Heat Fluxes The comparison of modeled surface fluxes with observations from field data demonstrates some disparity in the modeled energy balance for both models. Figure 5a demonstrates the agreement of CLM modeled net radiation and measured net radiation. As both downwelling longwave and shortwave radiation force CLM, the agreement of the net radiation demonstrates the accuracy of the simulated reflected radiation or albedo and upwelling longwave radiation. Figures 5b, 6a and 6b illustrate the differences between the observed fluxes and the CLM and LSP modeled fluxes for three components of the energy balance: latent heat flux, sensible heat flux and ground heat flux. The ground heat flux of CLM is compared to the measured ground heat flux at the surface (10 cm depth measurement corrected for soil heat storage), while the LSP ground heat flux

PAGE 34

26 for model layer 13 is comparable to the actual 10 cm flux measurement (Figure 6b). The MAE and RMSE were calculated using only daytime values (6:30 am until 6:30 pm) due to missing nighttime measurements resultant from dew on the lens of the krypton hygrometer. The CLM results for latent heat fluxes show the best agreement, while the sensible heat and ground heat fluxes have larger errors. The LSP results have the largest discrepancy between simulated and measured fluxes for latent heat fluxes, whereas the sensible heat and ground heat fluxes demonstrate better agreement. The diurnal variation in all energy fluxes is well captured by both models, with the exception of the LSP latent heat flux. The diurnal variation of the simulated latent heat flux increasingly lags the measured diurnal variation as the experimental period progresses. At the beginning of the experiment, the diurnal variation of the simulated latent heat fluxes lags the measured by 2 hrs. The lag is approximately 3 hrs during the last day of the experiment. The MAE and RMSE derived from comparisons of the measured and modeled latent heat for CLM is 44.79 Wm -2 and 61.69 Wm -2 The corresponding statistics for the LSP simulation are 99.43 Wm -2 and 122.32 Wm -2 A MAE of 47.09 Wm -2 and a RMSE of 60.92 Wm -2 is derived from comparisons of the measured and CLM simulated sensible heat flux. LSP has much better agreement with measured sensible heat flux with respective MAE and RMSE values of 23.50 Wm -2 and 29.08 Wm -2 CLM has relatively high errors of 20.83 Wm -2 and 26.19 Wm -2 for the MAE and RMSE of the simulated ground heat flux as compared to LSPs errors of 4.27 Wm -2 and 5.31 Wm -2 The MAE associated with the simulated daytime CLM latent heat flux represents 19% of the average 232.11 Wm -2 measured latent heat flux. The MAE of the LSP

PAGE 35

27 simulation is 43% of the average measured latent heat. The magnitude of the CLM daytime sensible heat MAE is approximately 48% of the magnitude of the average measured sensible heat flux, 98.7 Wm -2 A similar examination of LSP shows the magnitude equal to approximately 24% of the average measured sensible heat flux. The MAE resulting from comparisons of the measured and CLM daytime simulated ground heat flux is nearly equal to the average magnitude of the measured ground heat flux of 22.55 Wm -2 representing approximately 92% of the magnitude. The MAE of LSP daytime simulated ground heat flux is only 19% of the average measurement magnitude. Simulation of surface fluxes often provides the largest discrepancies from measurements. Chen et al. (1997) found reasonable agreement with ranges across modeling schemes of 30 Wm -2 and 25 Wm -2 for sensible and latent heat flux, respectively. Chang et al. (1999) also found good agreement using the CAPS model, with a monthly discrepancy of 7.1 Wm -2 in latent heat flux and 7.7 Wm -2 for sensible heat flux. However the same study found discrepancies in diurnal amplitude of ground heat flux around 20 Wm -2 with a phase difference of 2.5 hours. Both Chen et al. (1997) and Chang et al. (1999) calculated error margins based on the full diurnal cycle. A trend evident in several SVAT model validation studies (Acs and Hantel 1998, Chang et al. 1999, Gonzalez-Sosa et al. 2001) is that latent heat fluxes are more accurately simulated than sensible heat fluxes and ground heat fluxes. This trend suggests that the model physics representing the processes of latent heat flux are more comprehensive than the model physics describing the sensible or ground heat flux. The results of the CLM simulation provide another example of this tendency. In contrast, the LSP model has a tendency to underpredict the latent heat fluxes as the soil column dries out and

PAGE 36

28 evapotranspiration conditions drop below potential, while accurately simulating the sensible and ground heat fluxes. As the climatic conditions of the southeastern US may significantly influence model performance, a comparison of modeled fluxes were examined for two days that exhibit large differences in cloud cover. Within the experimental period, May 7 th represents a typical cloudy day, whereas May 13 th is a typical clear-sky day. The average measured net radiation is 210.4 Wm -2 on May 7 th and 394.2 Wm -2 on May 13 th Figures 7 a, b and c illustrate the latent heat flux, sensible heat flux and ground heat fluxes for May 7 th Both CLM and LSP underpredict the latent heat fluxes during the first 5.0 hours of the cloudy diurnal cycle. In order to maintain the energy balance, both CLM and LSP overpredict the measured sensible heat during the same period. CLM also overpredicts ground heat fluxes for the first half of the day, while LSP shows very good agreement throughout the day. During midday, the net radiation shows high variability that is indicative of cumulus atmospheric conditions. CLM closely tracks the variation in net radiation, but overpredicts the effect of that variability on the measured latent and sensible heat fluxes. While LSP demonstrates similar patterns, LSP better captures the effect of the variability. This may be a function of LSPs higher temporal resolution of the simulation. Figures 8 a, b and c demonstrate the latent heat, sensible heat and ground heat fluxes for the relatively clear skies observed on May 13 th CLMs simulated latent heat fluxes underpredict measured latent heat fluxes for the first three hours of the diurnal cycle. After the initial period, the measured values are overpredicted until the peak. CLMs latent heat and sensible heat fluxes show similar agreement with the

PAGE 37

29 measurements. Whereas the modeled latent heat fluxes overpredict midday values, the modeled sensible heat fluxes underpredict during the same period. LSPs simulated latent heat fluxes underpredict the measured latent heat fluxes for the first 10 hours of the day. The diurnal variation of LSPs simulated latent heat fluxes has a 3 hr phase shift. LSPs simulated sensible heat and ground heat fluxes show very good agreement with the measured fluxes over the course of the day. Overall, both models seem to simulate the surface energy fluxes well in both clear sky and convective climatic conditions. These results support the validity of the CLM and LSP model physics for determination of high temporal resolution applications for climates typical of the southeastern US.

PAGE 38

30 MAE RMSE CLM 19.20 23.90 -200-1000100200300400500600124125126127128129130131132133134135136137Time (days)Net Radiation (Wm-2) Measured CLM Modeled b)-100010020030040050060024125126127128129130131132133134135136137Time (days)Latent Heat Flux (Wm-2) Measured CLM Modeled LSP Modeled MAE RMSE CLM 44.79 61.69 LSP 99.44 122.32 a)700800 Figure 6. Comparison of modeled and measured surface fluxes for a) CLM net radiation and b) CLM and LSP latent heat flux. 1

PAGE 39

31 a)-1000100200300400500600124125126127128129130131132133134135136137Time (days)Sensible Heat Flux (Wm-2) Measured CLM Modeled LSP Modeled MAE RMSE CLM 47.09 60.92 LSP 23.50 29.08 b)-1000100200300400500600124125126127128129130131132133134135136137Time (days)Ground Heat Flux (Wm-2)-400-300-200-1000100200300Ground Heat Fulx (Wm-2) Measured Ground Heat at Surface CLM Modeled Measured Ground Heat at 10 cm LSP Modeled at 10 cm MAE RMSE CLM 20.83 26.19 LSP 4.27 5.31 Figure 7. Comparison of modeled and measured surface fluxes for a) CLM and LSP sensible heat flux and b) CLM and LSP ground heat flux.

PAGE 40

32 -200-100010020030040050060070080090010000123456789101112131415161718192021222324EDST (Hours)Latent Heat Flux (Wm-2) Net Radiation Measured Latent Heat CLM Modeled LSP Modeleda) -200-100010020030040050060070080090010000123456789101112131415161718192021222324EDST (Hours)Sensible Heat Flux (Wm-2) Net Radiation Measured Sensible Heat CLM Modeled LSP Modeledb) -200-100010020030040050060070080090010000123456789101112131415161718192021222324EDST (Hours)Ground Heat Flux (Wm-2)-800-700-600-500-400-300-200-1000100200300400Ground Heat Flux (Wm-2) Measured Ground Heat at Surface CLM Modeled LSP Modeled at 10 cm Measured Ground Heat at 10 cmc) Figure 8. Comparison of measured and modeled surface energy fluxes for a typical cloudy day, May 7 th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat fluxes.

PAGE 41

33 -200-100010020030040050060070080090010000123456789101112131415161718192021222324EDST (Hours)Sensible Heat Flux (Wm-2) Net Radiation Measured Sensible Heat CLM Modeled LSP Modeledb) -200-100010020030040050060070080090010000123456789101112131415161718192021222324EDST (Hours)Latent Heat Flux (Wm-2) Net Radiation Measured Latent Heat CLM Modeled LSP Modeleda) -200-100010020030040050060070080090010000123456789101112131415161718192021222324EDST (Hours)Ground Heat Flux (Wm-2)-800-700-600-500-400-300-200-1000100200300400Ground Heat Flux (Wm-2) Measured Ground Heat at Surface CLM Modeled LSP Modeled at 10 cm Measured Ground Heat at 10 cmc) Figure 9. Comparison of measured and modeled surface energy fluxes for a typical clear day, May 13 th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat fluxes.

PAGE 42

CHAPTER 6 CONCLUSION The purpose of this research was to compare two SVAT models applicability in a high water table environment characteristic of the southeastern US. As all SVAT models rely on similar mathematical approximations of biophysical reality, the main difference between a field scale model and a watershed scale single-column model is parameterization. LSPs strength is its meticulous reproduction of biophysical mechanisms at the land surface. In order to provide an exhaustive simulation of land surface processes, the LSP model can use all available data to parameterize the canopy and soil profile. Alternatively, CLM has an inherent database that contains default parameterization schemes for each IGBP land cover. The results indicate that both CLM and LSP can be used with success in the convective climate and high water table environment of the southeastern US. The two models were used to simulate a 13-day dry drown. The simulations show that subsurface moisture content and surface energy fluxes provide good agreement with measured values. Both models simulated soil temperatures show very good agreement as compared to measured temperatures. The models were able to appropriately replicate the soil warming that occurred over the experimental period as well as the diurnal fluctuations. Both models simulated energy fluxes within expected error ranges. When considering the field scale of the experiment, CLM performs quite well in comparison with the biophysically more sophisticated LSP model. However, for use in the southeastern US, it is important to alter CLMs baseflow generation physics mechanism 34

PAGE 43

35 to account for the low relief topography of the region. Overall, the experiment demonstrates potential for the application of both CLM and LSP in the southeastern US.

PAGE 44

LIST OF REFERENCES Acs, F. and M. Hantel, 1998: The land-surface flux model PROGSURF. Global and Planetary Change, 19, 19-34. Baldocchi, D. D., B. B. Hicks, and T. P. Meyers, 1988: Measuring Biosphere-Atmosphere Exchanges of Biologically Related Gases with Micrometeorological Methods. Ecology, 69, 1331-1340. Beven, K. J., R. Lamb, P. Quinn, R. Romanowicz, and J. Freer, 1995: Topmodel. Computer Models of Watershed Hydrology, Water Resource Publications, Highlands Ranch, CO, 627-668. Blackmon, M., B. Boville, F. Bryan, R. Dickinson, P. Gent, J. Kiehl, R. Moritz, D. Randall, J. Shukla, S. Solomon, G. Bonan, S. Doney, I. Fung, J. Hack, E. Hunke, J. Hurrell, J. Kutzbach, J. Meehl, B. Otto-Bliesner, R. Saravanan, E. K. Schneider, L. Sloan, M. Spall, K. Taylor, J. Tribbia, and W. Washington, 2001: The Community Climate System Model. Bulletin of the American Meteorological Society, 82, 2357-2376. Bonan, G. B., 1996: A land surface model (LSM version 1.0) for ecological, hydrological, and atmospheric studies: Technical Description and User's Guide. NCAR Technical Note, NCAR/TN-417+STR, 150. Bonan, G. B., K. W. Oleson, M. Vertenstein, S. Levis, X. B. Zeng, Y. J. Dai, R. E. Dickinson, and Z. L. Yang, 2002: The land surface climatology of the community land model coupled to the NCAR community climate model. Journal of Climate, 15, 3123-3149. Chang, S., D. Hahn, C. H. Yang, and D. Norquist, 1999: Validation study of the CAPS model land surface scheme using the 1987 Cabauw/PILPS dataset. Journal of Applied Meteorology, 38, 405-422. 36

PAGE 45

37 Chen, T. H., A. HendersonSellers, P. C. D. Milly, A. J. Pitman, A. C. M. Beljaars, J. Polcher, F. Abramopoulos, A. Boone, S. Chang, F. Chen, Y. Dai, C. E. Desborough, R. E. Dickinson, L. Dumenil, M. Ek, J. R. Garratt, N. Gedney, Y. M. Gusev, J. Kim, R. Koster, E. A. Kowalczyk, K. Laval, J. Lean, D. Lettenmaier, X. Liang, J. F. Mahfouf, H. T. Mengelkamp, K. Mitchell, O. N. Nasonova, J. Noilhan, A. Robock, C. Rosenzweig, J. Schaake, C. A. Schlosser, J. P. Schulz, Y. Shao, A. B. Shmakin, D. L. Verseghy, P. Wetzel, E. F. Wood, Y. Xue, Z. L. Yang, and Q. Zeng, 1997: Cabauw experimental results from the project for intercomparison of land-surface parameterization schemes. Journal of Climate, 10, 1194-1215. Clapp, R. B. and G. M. Hornberger, 1978: Empirical Equations for Some Soil Hydraulic-Properties. Water Resources Research, 14, 601-604. Cosby, B. J., G. M. Hornberger, R. B. Clapp, and T. R. Ginn, 1984: A Statistical Exploration of the Relationships of Soil-Moisture Characteristics to the Physical-Properties of Soils. Water Resources Research, 20, 682-690. Dai, Y., X. Zeng, and R. Dickinson, 2001: Common Land Model (CLM): Technical Documentation and User's Guide. (climate.eas.gatech.edu/dai/clmdoc.pdf), 69. Diak, G. R., W. L. Bland, J. R. Mecikalski, and M. C. Anderson, 2000: Satellite-based estimates of longwave radiation for agricultural applications. Agricultural and Forest Meteorology, 103, 349-355. Dickinson, R., A. Henderson-Sellers, P. J. Kennedy, and M. F. Wilson, 1993: Biosphere atmosphere transfer scheme (BATS) version 1e as coupled to the NCAR Community Climate Model. NCAR Technical Note, NCAR/TN-378+STR, 72. Ghildyal, B. P. and R. Tripathi, 1987: Soil Physics. New York City, NY, John Wiley. Gonzalez-Sosa, E., I. Braud, J. L. Thony, M. Vauclin, and J. C. Calvet, 2001: Heat and water exchanges of fallow land covered with a plantresidue mulch layer: a modelling study using the three year MUREX data set. Journal of Hydrology, 244, 119-136. Green, W. and G. Ampt, 1911: Studies on soil physics. Journal of Agricultural Science, 4, 1-24. Jacobs, J. M., D. A. Myers, M. C. Anderson, and G. R. Diak, 2002a: GOES surface insolation to estimate wetlands evapotranspiration. Journal of Hydrology, 266, 53-65. Jacobs, J. M., S. L. Mergelsberg, A. F. Lopera, and D. A. Myers, 2002b: Evapotranspiration from a wet prairie wetland under drought conditions: Paynes Prairie Preserve, Florida, USA. Wetlands, 22, 374-385.

PAGE 46

38 Judge, J., L. M. Abriola, and A. W. England, 2003a: A summertime Land Surface Process and Radiobrightness model Part I: Development and Numerical Validation. Advances in Water Resources, in review. Judge, J., J. Metcalfe, D. McNichol, B. E. Goodison, A. E. Walker, and A. W. England, 2003b: A summertime Land Surface Process and Radiobrightness model Part II: Model Calibration. Water Resources Research, in review. Kim, E. J., 1999: Remote sensing of land surface conditions in arctic tundra regions for climatological applications using microwave radiometry, PhD thesis, University of Michigan. Koster, R. D., M. J. Suarez, A. Ducharne, M. Stieglitz, and P. Kumar, 2000: A catchment-based approach to modeling land surface processes in a general circulation model 1. Model structure. Journal of Geophysical Research-Atmospheres, 105, 24809-24822. Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges, 1994: A Simple Hydrologically Based Model of Land-Surface Water and Energy Fluxes for General-Circulation Models. Journal of Geophysical Research-Atmospheres, 99, 14415-14428. Liang, X., E. F. Wood, and D. P. Lettenmaier, 1996: Surface soil moisture parameterization of the VIC-2L model: Evaluation and modification. Global and Planetary Change, 13, 195-206. Liang, X., E. F. Wood, D. P. Lettenmaier, D. Lohmann, A. Boone, S. Chang, F. Chen, Y. J. Dai, C. Desborough, R. E. Dickinson, Q. Y. Duan, M. Ek, Y. M. Gusev, F. Habets, P. Irannejad, R. Koster, K. E. Mitchell, O. N. Nasonova, J. Noilhan, J. Schaake, A. Schlosser, Y. P. Shao, A. B. Shmakin, D. Verseghy, K. Warrach, P. Wetzel, Y. K. Xue, Z. L. Yang, and Q. C. Zeng, 1998: The Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) phase 2(c) Red-Arkansas River basin experiment: 2. Spatial and temporal analysis of energy fluxes. Global and Planetary Change, 19, 137-159. Liou, Y. A., J. F. Galantowicz, and A. W. England, 1999: A land surface process radiobrightness model with coupled heat and moisture transport for prairie grassland. Ieee Transactions on Geoscience and Remote Sensing, 37, 1848-1859. Loveland, T. R., B. C. Reed, J. F. Brown, D. O. Ohlen, Z. Zhu, L. Yang, and J. W. Merchant, 2000: Development of a global land cover characteristics database and IGBP DISCover from 1 km AVHRR data. International Journal of Remote Sensing, 21, 1303-1330. Mitsch, W. J. and J. G. Gosselink, 2000: Wetlands. New York City, NY, John Wiley.

PAGE 47

39 Mohr, K. I., J. S. Famiglietti, A. Boone, and P. J. Starks, 2000: Modeling soil moisture and surface flux variability with an untuned land surface scheme: A case study from the Southern Great Plains 1997 Hydrology Experiment. Journal of Hydrometeorology, 1, 154-169. Nijssen, B., I. Haddeland, and D. P. Lettenmaier, 1997: Point evaluation of a surface hydrology model for BOREAS. Journal of Geophysical Research-Atmospheres, 102, 29367-29378. Philip, J. R. and D. De Vries, 1957: Moisture movement in porous materials under temperature gradients. Transactions, American Geophysical Union, 38, 222-232. Philip, J. R., 1957: Theory of infiltration, 1. The infiltration equation and its solution. Soil Science, 83, 345-357. , 1987a: Inverse solution for one-dimensional infiltration and ratio A/K1. Water Resources Research, 26, 2023-1017. , 1987b: The Infiltration Joining Problem. Water Resources Research, 23, 2239-2245. Pitman, A. J. and A. Henderson-Sellers, 1998: Recent progress and results from the project for the intercomparison of landsurface parameterization schemes. Journal of Hydrology, 213, 128-135. Pitman, A. J., A. Henderson-Sellers, C. E. Desborough, Z. L. Yang, F. Abramopoulos, A. Boone, R. E. Dickinson, N. Gedney, R. Koster, E. Kowalczyk, D. Lettenmaier, X. Liang, J. F. Mahfouf, J. Noilhan, J. Polcher, W. Qu, A. Robock, C. Rosenzweig, C. A. Schlosser, A. B. Shmakin, J. Smith, M. Suarez, D. Verseghy, P. Wetzel, E. Wood, and Y. Xue, 1999: Key results and implications from phase 1(c) of the Project for Intercomparison of Land-Surface Parametrization Schemes. Climate Dynamics, 15, 673-684. Rossi, C. and J. R. Nimmo, 1994: Modeling of Soil-Water Retention from Saturation to Oven Dryness. Water Resources Research, 30, 701-708. Schaake, J., Q. Duan, V. Koren, and A. Hall, 2001: Toward improved parameter estimation of land surface hydrology models through the Model Parameter Estimation Experiment (MOPEX). Soil-Vegetation-Atmosphere Transfer Schemes and Large-Scale Hydrological Models, Maastricht, The Netherlands, IAHS Pblishing, 91-97. Schotanus, P., F. T. M. Nieuwstadt, and H. A. R. Debruin, 1983: Temperature-Measurement with a Sonic Anemometer and Its Application to Heat and Moisture Fluxes. Boundary-Layer Meteorology, 26, 81-93.

PAGE 48

40 Shao, Y. P. and A. Henderson-Sellers, 1996: Validation of soil moisture simulation in landsurface parameterisation schemes with HAPEX data. Global and Planetary Change, 13, 11-46. Slater, A. G., C. A. Schlosser, C. E. Desborough, A. J. Pitman, A. Henderson-Sellers, A. Robock, K. Y. Vinnikov, K. Mitchell, A. Boone, H. Braden, F. Chen, P. M. Cox, P. de Rosnay, R. E. Dickinson, Y. J. Dai, Q. Duan, J. Entin, P. Etchevers, N. Gedney, Y. M. Gusev, F. Habets, J. Kim, V. Koren, E. A. Kowalczyk, O. N. Nasonova, J. Noilhan, S. Schaake, A. B. Shmakin, T. G. Smirnova, D. Verseghy, P. Wetzel, X. Yue, Z. L. Yang, and Q. Zeng, 2001: The representation of snow in land surface schemes: Results from PILPS 2(d). Journal of Hydrometeorology, 2, 7-25. Stieglitz, M., D. Rind, J. Famiglietti, and C. Rosenzweig, 1997: An efficient approach to modeling the topographic control of surface hydrology for regional and global climate modeling. Journal of Climate, 10, 118-137. Tanner, B. D. and J. P. Greene, 1989: Measurements of sensible heat flux and water vapor fluxes using eddy correlation methods. DAAD 09-87, 94 pp. Twine, T. E., W. P. Kustas, J. M. Norman, D. R. Cook, P. R. Houser, T. P. Meyers, J. H. Prueger, P. J. Starks, and M. L. Wesely, 2000: Correcting eddy-covariance flux underestimates over a grassland. Agricultural and Forest Meteorology, 103, 279-300. University of Florida-Institute of Food Agricultural Sciences. 1985. Characterization data for selected soils. Soil Science research report No. 85-1 van der Keur, P., S. Hansen, K. Schelde, and A. Thomsen, 2001: Modification of DAISY SVAT model for potential use of remotely sensed data. Agricultural and Forest Meteorology, 106, 215-231. Verseghy, D. L., N. A. McFarlane, and M. Lazare, 1993: Class a Canadian Land-Surface Scheme for Gcms .2. Vegetation Model and Coupled Runs. International Journal of Climatology, 13, 347-370. Webb, E. K., G. I. Pearman, and R. Leuning, 1980: Correction of Flux Measurements for Density Effects Due to Heat and Water-Vapor Transfer. Quarterly Journal of the Royal Meteorological Society, 106, 85-100. Wood, E. F., D. P. Lettenmaier, X. Liang, D. Lohmann, A. Boone, S. Chang, F. Chen, Y. J. Dai, R. E. Dickinson, Q. Y. Duan, M. Ek, Y. M. Gusev, F. Habets, P. Irannejad, R. Koster, K. E. Mitchel, O. N. Nasonova, J. Noilhan, J. Schaake, A. Schlosser, Y. P. Shao, A. B. Shmakin, D. Verseghy, K. Warrach, P. Wetzel, Y. K. Xue, Z. L. Yang, and Q. C. Zeng, 1998: The Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) phase 2(c) Red-Arkansas River basin experiment: 1. Experiment description and summary intercomparisons. Global and Planetary Change, 19, 115-135.

PAGE 49

41 Zeng, X. B., M. Shaikh, Y. J. Dai, R. E. Dickinson, and R. Myneni, 2002: Coupling of the common land model to the NCAR community climate model. Journal of Climate, 15, 1832-1854.

PAGE 50

BIOGRAPHICAL SKETCH Brent Whitfield was born in 1978 in West Palm Beach, Florida. He received an International Baccalaureate diploma from Suncoast Community High School in 1996, after which he began his undergraduate studies at the University of Florida. In 2001, Brent graduated with a Bachelor of Science in civil engineering. In the summer of 2001 he joined the Summer Institute for Atmospheric and Hydrospheric Sciences at NASAs Goddard Space Flight Center in Greenbelt, Maryland. In the fall of 2001, Brent began his studies as a candidate for a Master of Engineering in the Department of Civil and Coastal Engineering under the tutelage of his major professor, Dr. Jennifer Jacobs. 42


Permanent Link: http://ufdc.ufl.edu/UFE0000824/00001

Material Information

Title: Svat Calibration of Point and Regional Scale Water and Energy Dynamics
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000824:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000824/00001

Material Information

Title: Svat Calibration of Point and Regional Scale Water and Energy Dynamics
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000824:00001


This item has the following downloads:


Full Text












SVAT CALIBRATION OF POINT AND REGIONAL SCALE
WATER AND ENERGY DYNAMICS


















By

BRENT WHITFIELD


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Brent Whitfield















ACKNOWLEDGMENTS

I would sincerely like to thank Dr. Jennifer Jacobs, chairperson of my supervisory

committee, for all of her guidance and support throughout the project. I would also like

to thank my committee members, Dr. Jasmeet Judge and Dr. Kirk Hatfield, for all of their

valuable suggestions and encouragement. I am very appreciative for the help of Dr. Paul

Houser and Dr. Matthew Rodell from NASA's Goddard Space Flight Center in

Greenbelt, Maryland. I am grateful to Gerard Ripo, Sudheer Reddy Satti, Shirish Bhat,

Aniruddha Guha, Enching Hsu, Siqing Liu and Jeremy Andrews for all of their help in

data collection, analysis and peer review.

I am indebted to my girlfriend, Erica Zingone, for all of her encouragement and

for providing me with a justification for taking monthly vacations to Atlanta, Georgia. I

am grateful for my parents, Randy and Martha Whitfield, for the emotional and financial

support afforded me throughout my education at the University of Florida. I would also

like to thank my dog Jake Whitfield for his consistent allegiance and watchful eye over

all of my work.

Research assistantship was provided by NASA NIP Grant NAG5-10567.
















TABLE OF CONTENTS

Page

A C K N O W L E D G M E N T S ................................................................................................. iii

L IST O F T A B L E S .............. ................................................... ................ .. ..v... ..v

LIST OF FIGURES .................................................. ............................ vi

ABSTRACT ...................................................................... vii

CHAPTER

1. IN T R O D U C T IO N ................................................................................................... 1

2. M O D EL C O M PA R ISO N S ...................................................................... ...............6...

3. STU D Y A R E A ................................................................. ................. ........ ...... 11

4. SIM U LA TION D ESIGN .. .................................................................. ............... 15

P aram eteriz atio n ......................................................................................................... 15
Initialization ................................................................................. ...................... 16
F o rc in g s ......................................................................................................................1 7

5. RESULTS AND D ISCU SSION ................................... ...................... ............... 18

S o il M o istu re .............................................................................................................. 1 8
S o il T em p e ratu re ........................................................................................................2 2
Surface Heat Fluxes ........................... ........... ........................ 25

6 C O N C L U SIO N ...................................................... ................................................ 34

L IST O F R E FE R E N C E S ... ........................................................................ ................ 36

BIO GR APH ICAL SK ETCH .................................................................... ................ 42















LIST OF TABLES


Table page

1. Methodology for parameterization for CLM and LSP. ........................ ..................... 7

2. Characterization of land surface processes for CLM and LSP.............................................8...

3. Instrum entation used in this study .................................................................................14

4. Required atmospheric forcing for CLM and LSP ..................................................... 17















LIST OF FIGURES


Figure page

1. Log scale plot of variation of parameterized saturated hydraulic conductivity with
depth below the surface for CLM and LSP .............................................. ............... 9

2. Water table depth with time from April 10th the June 20th. Experiment period from
May 4th to May 17th .. ..................... ........ ...................12

3. Comparison of measured soil water characteristic curve with three soil water
characteristic curves defined by modeled soil matric potential (SMP) and seven
hydraulic conductivity (K) curves based on the Clapp and Hornberger relationship
w within the CLM param eterization schem e ............................................. ............... 16

4. Comparison of measured soil moisture with CLM and LSP modeled soil moisture at a)
7.6 cm b) 12.7 cm and c) 17.8 cm ...................................................... ................ 20

5. Comparison of measured subsurface soil temperature and CLM and LSP modeled soil
temperature at a) 2.5 cm, b) 7.6 cm, c) 12.7 cm and d) 17.8 cm...........................24

6. Comparison of modeled and measured surface fluxes for a) CLM net radiation and b)
C L M and L SP latent heat flux ................................................................ ............... 30

7. Comparison of modeled and measured surface fluxes for a) CLM and LSP sensible
heat flux and b) CLM and LSP ground heat flux ................................... ............... 31

8. Comparison of measured and modeled surface energy fluxes for a typical cloudy day,
May 7th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat fluxes. .... 32

9. Comparison of measured and modeled surface energy fluxes for a typical clear day,
May 13th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat fluxes..33















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Engineering

SVAT CALIBRATION OF POINT AND REGIONAL SCALE
WATER AND ENERGY DYNAMICS

By

Brent Whitfield

May 2003

Chair: Jennifer Jacobs
Major Department: Civil and Coastal Engineering

The Community Land Model (CLM) and the Land Surface Process model (LSP)

are typical soil vegetation atmosphere transfer models (SVAT) that were used to simulate

the land surface processes for a wet prairie community in the southeastern US for May 4-

17, 2001. The research objective is to provide a basis for calibration of SVAT modeling

processes for the highly variable convective atmospheric conditions and the atypical

hydrogeology of the southeastern US. Since CLM is a watershed or regional scale model

and LSP is a field scale model, each model has distinctive parameterization schemes and

methodologies for simulating land surface processes. The intercomparison of CLM and

LSP provides perspective for the strengths and weaknesses of the generalized

parameterization schemes of a regional scale model as compared to a more biophysically

adept field scale model. The study area is Paynes Prairie State Preseve in north-central

Florida, which can be characterized by the biological communities of freshwater marsh,

wet prairie and pasture.









Validation data were provided using standard meteorological instrumentation, an

eddy correlation system for measurement of surface heat fluxes and subsurface

measurements of moisture, temperature and matric potential. The CLM and LSP

simulations of soil moisture at 7.6, 12.7 and 17.8 cm compared well with the measured

results showing a RMSE less than 3.5% volumetric water content for both models at all

three depths. Soil temperature simulations corresponding to 2.5, 7.6, 12.7 and 17.8 cm

show good agreement with measured temperatures with a RMSE less than 1.5 K for all

four depths in both simulations. Simulation of the components of the energy balance

showed different trends for CLM than LSP. CLM's simulation showed the strongest

agreement with the measured latent heat fluxes (RMSE of 62 Wm-2) as compared with

the simulation of sensible (RMSE of 61 Wm-2) and ground heat fluxes (RMSE of 27

Wm-2). The LSP simulation of latent heat flux showed the poorest agreement (RMSE of

122 Wm-2) as compared with the simulation of sensible (RMSE of 29 Wm-2) and ground

heat fluxes (RMSE of 5 Wm-2). An investigation of model performance on a typical

cloudy day (May 7th) and a typical clear day (May 13th) demonstrated potential for the

application of both CLM and LSP in the highly variable climatic conditions of the

southeastern US.















CHAPTER 1
INTRODUCTION

Improved quantification of the energy, moisture, and momentum fluxes between

the atmosphere and the Earth's surface remains a critical component for studies in

hydrology, meteorology and biogeoscience. Scientific endeavors, such as climate

modeling, are often interested in quantifying surface fluxes at large spatial scales over a

variety of land cover types that are not feasibly monitored using field based techniques.

In the stead of complex direct measurement, soil-vegetation-atmosphere transfer (SVAT)

models are often used to quantify these fluxes and stores. The most suitable

characteristics for SVAT models are often determined by their desired application. The

divergence in applications of SVAT models has lead to a variety of models that have

structural differences for the characterization of land surface processes. Only by

comparing the results of separate SVAT models with a known set of validation data can

the utility of different modeling schemes be determined.

The applicability of a given SVAT model is significantly dependent on its required

parameterizations and model structure (Schaake et al. 2001). Increased parameterization

allows for a greater description of the physical environment and theoretically creates a

more robust simulation of surface fluxes. However, this increased robustness often leads

to a decreased flexibility in usage. Often a less sophisticated model parameterization can

prove advantageous, due to the characteristic heterogeneity of the land surface, as well as

a general scarcity of insitu measurement of physical parameters. An effective SVAT









model should incorporate adequate sophistication to accurately simulate the environment,

but not be so complex as to reduce usability.

In the Project for Intercomparison of Land surface Parameterization Schemes or

PILPS, several scientific teams explored the value of different methods of model

parameterization (Pitman and Henderson-Sellers 1998). The multiphase PILPS

compared results from as many as 23 land surface models including BATS (Dickinson et

al. 1993) and VIC-2L (Liang et al. 1996). The PILPS focus was a community-wide

intercomparison of modeling schemes. Thus, none of the studies provided a detailed

analysis of individual models. However, the intercomparisons did present results that can

be used to guide more detailed intercomparisons (Chen et al. 1997). PILPS Phase 1

comparisons of annually-averaged latent heat flux, sensible heat flux, and radiative

temperature, as well as monthly-averaged surface runoff and soil moisture with surface

fluxes generated from NCAR climate data showed poor agreement in tropical forest and

grassland settings. The revised experimentation framework demonstrated the importance

of congruent forcing and parameterization schemes in model comparison studies

(Pitman et al. 1999). Phase 2 provided a similar comparison using observed climate data

from four different sites, the Cabauw experimental station in the Netherlands (Chen et al.

1997), the HAPEX-MOBILHY site in southwest France (Shao and Henderson-Sellers

1996), the Red-Arkansas River Basin in the Midwestern US (Liang et al. 1998), and a

boreal site in Valdai, Russia (Slater et al. 2001). PILPS Phase 2 identified the differences

in land surface model mechanisms and characterized the variability in the modeled

results. Shao and Henderson-Sellers (1996) describe the difficulty in identifying the link

between the model's performance and the responsible mechanism. Because land surface









models are non-linear systems with interacting components, the range of results in each

of the Phase 2 intercomparisons is larger than acceptable measurement error. All of the

Phase 2 intercomparisons, except those examining snow-covered terrain, exhibited a

smaller disparity in latent heat fluxes with respect to the average annual measured latent

heat flux than both the sensible heat flux and the ground heat flux (Shao and Henderson-

Sellers 1996, Chen et al. 1997, Liang et al. 1998). PILPS demonstrated modeling

schemes' strengths in simulating general patterns of land surface processes. The overall

scatter in simulation results, which existed between any set of models, may be

categorized based on the structure of modeling schemes. However, PILPS's

investigation determined that within categories of like modeling schemes there exists

varying levels of sensitivity to parameterization (Slater et al. 2001).

While PILPS characterized the overall status of SVATs, model studies that include

fewer schemes allow more detailed examination of the models' strengths and weaknesses

for a variety of landscapes and climatic conditions. The Mohr et al. (2000) study of the

untuned PLACE model in the Southern Great Plains experiment effectively simulated the

near surface temperature, moisture and land surface fluxes in the first 48 hours following

a heavy rainfall. However, they diverged from observations during the drying period due

to simplifications in the parameterization of subsurface moisture, vegetation cover and

soil texture. For the prairie grassland in the Radiobrightness Energy Balance Experiment

(REBEX) near Sioux Falls, SD, the LSP/R model exhibited small errors when comparing

predicted and observed soil and canopy temperatures in a northern prairie over a 14-day

period in October (Liou et al. 1999). Energy fluxes were not examined. The van der

Keur et al. (2001) study of the modified DAISY SVAT model had good agreement









between the modeled and observed land surface fluxes in non-water limiting conditions

for winter wheat in Jutland, Denmark. However, its application during water limiting

conditions revealed that the simulated latent heat fluxes underestimated observed values

and required a more complete parameterization of canopy resistance during stressed

conditions. Nijssen et al. (1997) demonstrated good seasonal agreement with observed

fluxes using the hydrology-soil-vegetation model DHSVM in forested sites in the Boreal

region of Canada. However, the Nij ssen simulation had a time lag for the simulated

sensible heat fluxes that suggests the need for a more complete parameterization scheme

for the soil thermal model. The importance of SVAT simulation efforts to provide

validation in an array of climactic and vegetative conditions is exhibited in the broad

range of locations found in existing research: mid-latitude grasslands and croplands

(Wood et al. 1998, Liou et al. 1999, Mohr et al. 2000), boreal forests and arctic tundra

(Njissen et al. 1997, Kim 1999, Slater et al. 2001), and tropical forests (Pitman et al.

1999). Although SVAT validation research conducted thus far has been representative of

several regions and climate types, global extrapolation requires additional validation

research across a greater number of climates and biomes.

It is the purpose of this paper to provide a basis for calibration of SVAT modeling

processes for the highly variable convective atmospheric conditions and the atypical

hydrogeology in the southeastern US. The southeastern US is characterized by

subtropical climatic conditions, that consist of high humidity, convective heating and

high annual rainfall. The low topographic relief and high water tables create a unique

environment for vegetative communities. These communities are distinguished by

shallow root zones and variable stomatal resistances and reflectance properties (Mitsch






5


and Gosselink 2000). A comparison of two SVAT models of contrasting design is used

to determine the strengths and weaknesses of each methodology. The Common Land

Model (CLM) (Dai et al. 2001) is a regional or watershed scale model, while the Land

Surface Process (LSP) model (Liou et al. 1999) is a field scale model. This

intercomparison examines the utility of increased sophistication in parameterization as it

relates to differences in scale and its practical application in the southeastern US














CHAPTER 2
MODEL COMPARISONS

The Common Land Model (CLM) was developed as part of a multi-disciplinary

and multi-institutional project designed to provide land surface forcing for the lower

boundary of the Community Climate System Model (Blackmon et al. 2001). Dai et al.

(2001) describe the technical elements of CLM. Zeng et al. (2002) and Bonan et al.

(2002) demonstrate the coupling of CLM with CCM3. The Land Surface Process (LSP)

model is designed to link traditional land surface models and satellite microwave

observations to allow for data assimilation (Liou et al. 1999). Judge et al. (2003a and

2003b) describe recent modifications and further developments to the LSP model.

CLM and LSP can be distinguished primarily by their model processes and

required parameters that reflect their designed application scale. Table 1 summarizes the

differences between their parameterization schemes. As CLM was designed to represent

single columns with spatial extents as large as 1/4 by 1/4, it incorporates a minimum of

insitu measurement of model parameters and initialization data. The only parameters that

are user-provided are land cover class, soil texture, longitude, latitude, and soil color

index. The remaining environmental parameters are derived from an internal

parameterization scheme based on land cover classification. Land cover class is specified

based on 18 different International Geosphere-Biosphere Programme (IGBP)

classifications (Loveland et al. 2000) from which the remaining vegetation parameters are

derived. Estimation of surface fluxes from multiple land cover classifications uses a tile-









mosaic approach similar to Koster et al. (2000). Soil parameters, such as hydraulic

conductivity and water retention curves, are generated from empirically based derivations

using soil texture (Clapp and Hornberger 1978, Cosby et al. 1984).

Table 1. Methodology for parameterization for CLM and LSP.

Parameters CLM LSP
Soil Texture User Defined User Defined
q Empirical Calculation User Defined
Ksat Clapp and Hornberger User Defined
k Empirical Calculation User Defined
Wilting Point User Defined User Defined
Water Retention Curve Clapp and Hornberger User Defined
Root Depth Empirical Calculation / IGBP User Defined
LAI Empirical Calculation / IGBP User Defined
Canopy Height IGBP User Defined
Roughness Length IGBP User Defined

In contrast, as LSP was designed as a research tool for field scale applications, it

uses a more detailed parameterization scheme and requires insitu measurement of many

environmental parameters. Field specific parameters such as soil texture, longitude,

latitude, LAI, canopy height, and canopy biomass are user specified. Two physical soil

parameters, air entry pressure (',o) and pore size index (X), are required to generate the

soil water characteristic curve. LSP uses the Rossi and Nimmo (1994) relationship based

on the Brooks and Corey water retention model.

The mechanics of the hydrological processes: evapotranspiration, infiltration, and

runoff vary significantly between the two models. Table 2 describes these differences for

CLM and LSP. Both models determine bare soil evaporation using the Philip and De

Vries (1957) diffusion model. CLM's transpiration process is an aerodynamic approach

based on the BATS model (Dickinson et al. 1993) with a stomatal resistance component









from the LSM model (Bonan, 1996). LSP's transpiration module uses an aerodynamic

approach derived from the CLASS model (Verseghy et al. 1993). LSP's infiltration rate

requires soil hydraulic and physical properties, soil temperature, vegetation cover, and

surface characteristics like slope and roughness (Ghildyal and Tripathi, 1987). The

maximum infiltration rate is estimated using a quasi-analytic solution to Richard's

equation for vertical infiltration in a homogeneous soil with a constant initial moisture

profile (Green and Ampt, 1911, Philip 1957, Philip 1987a, Philip 1987b).

Table 2. Characterization of land surface processes for CLM and LSP.

Processes CLM LSP
Soil Evaporation Diffusion Diffusion
Evapotranspiration Aerodynamic (BATS and LSM based) Aerodynamic (CLASS based)
Infiltration Richard's Equation Richard's Equation (modified)
Subsurface Heat Transport Fourier's Equation Fourier's Equation
Runoff Surface runoff and baseflow (TOPMODEL based) Hortonian Flow

One of the distinguishing characteristics of CLM is the surface runoff generation

process. CLM uses assumptions from the watershed scale model TOPMODEL (Beven et

al. 1995) as adapted for land surface modeling (Stieglitz et al. 1997). These assumptions

include: a baseflow component defined by the saturated hydraulic conductivity of the

lower layers of the profile, an exponential decrease in the saturated hydraulic

conductivity with depth and surface runoff based on the parameterization of saturated and

unsaturated spatial extents. The exponential decrease of saturated hydraulic conductivity

with depth is illustrated in Figure 1 using the conductivities defined by the soil profile

used in this study. Figure 1 also illustrates the saturated conductivity used by LSP as

parameterized by the soil profile of this study. The parameterization of the fraction of

saturated and unsaturated landscape is based on an empirical relationship defined by the

relative water table. Since LSP is designed for field scale use only, runoff is only










generated from saturation excess and that depth of water not infiltrated at the end of each

timestep.




00


05 -


110


15


20 **- CLM


25 I


30 -
00000 00000 00001 00010 00100 01000 1 0000
Saturated Hydraulic Conductivity (mm/s)


Figure 1. Log scale plot of variation of parameterized saturated hydraulic conductivity
with depth below the surface for CLM and LSP

Two modifications were made to CLM. For the low gradient topography and

hydrogeology characteristic of much of the southeastern US, the TOPMODEL baseflow

generation mechanism is not appropriate. The TOPMODEL baseflow generation

mechanism was eliminated allowing the lower layers to remain saturated and eliminating

excessive drainage. The parameterization of the relationship between soil moisture and

root resistance was also modified for the CLM evapotranspiration mechanism based on

measured stresses (Jacobs et al. 2002a).

Both CLM and LSP discretize the soil profile into layers. Each allow soil

properties to differ by layer. CLM has a 10-layer profile with the thickness of each layer

determined by a unitless scaling factor and an exponential function that increases with

depth. Given a typical vertical scaling factor of 0.025, the thickness of the uppermost









layer of CLM is 1.75 cm, while the thickness of the lowest layer is 113.7 cm. LSP

layers' thicknesses also increase with depth. However, the LSP profile has 60 layers.

The top layers are very fine, yet a deep soil layer may be simulated. Each layer's

thickness may be defined as appropriate for the application.

CLM and LSP also have fundamental computational differences. CLM determines

the water movement across layer interfaces using a first-order Taylor expansion and

solving the resulting equation using a tridiagonal matrix solution. The flux of heat across

CLM layer interfaces is solved using the Crank-Nicholson numerical scheme and a

tridiagonal matrix solution (Dai et al. 2001). These methods generate soil moisture and

temperature profiles at the same temporal resolution as the timestep of the simulation.

LSP uses a block-centered finite difference approach that determines the temporal

resolution based on the parameterized convergence criteria. The temporal resolution of

the resulting LSP profile also differs from the timestep of the climate forcing.














CHAPTER 3
STUDY AREA

The Paynes Prairie State Preserve is a regional basin that contains biological

communities such as freshwater marsh, wet prairie and pasture. This 5600 ha system in

north-central Florida, USA is 13 km long (east-west) and ranges in the north-south

dimension from 1.5 km to 7 km. The summer climate conditions are strongly influenced

by surface heating with significant cumulus cloud cover and few cloud-free days. The

basin is a large irregularly shaped bowl resulting from solution of the underlying

limestone. The basin is either seasonally or perennially flooded based on a surficial

aquifer that is separated from the highly transmissive Floridan aquifer by the underlying

Hawthorne formation. This geologic formation acts as a semi-confining layer. The

primary drain for the Preserve is a sinkhole breach (Alachua Sink) in the semi-confining

layer. Based on rainfall patterns, two water-control structures are used to simulate

natural basin water levels annually and inter-annually. Generally, water levels are

highest during the summer months and lowest during the months of April, September,

October, and November. However, no diversion of water into the Preserve through the

control structures occurred during the study period due to a multi-year drought.

The study was conducted in a wet prairie community located in the Paynes Prairie

Preserve (290 34' 14 N, 82 16' 46 W). The study period is a 13-day rainless span

from May 4th until May 17th of 2001. The study period occurs within a 40-day dry down

from saturated conditions from April 17th until May 27th. There is no precipitation during

the 40-day dry down period with the exception of three storm events, of which only one











is notable in depth. The first event is a 2.3 cm rainfall on April 27th. The second and


third precipitation events are both 0.25 mm in depth and occur on May 2nd and May 17th.

As is seen in Figure 2, the water table depth reflects these rainfall patterns. The


experiment period is during a steady dry down condition in which the depth to the water


table increased from 62 cm to 86 cm.





00


-02


-04

-I06


0-
C 08


-10


-1 2 Experiment
Period

-14
100 110 120 130 140 150 160 170 180
Time (Day)

Figure 2. Water table depth with time from April 10th the June 20th. Experiment period
from May 4th to May 17th.

The prairie is a relatively flat, treeless plain with moderately dense ground cover.

The mean canopy height during the study period is approximately 1.0 m. The site's soils


include Emeralda fine sandy loam, Wauberg sand, and Ledwith Muck. The soils consist


of sands with an organic surface layer that are underlain by clay. Field observations

showed that the majority of the root zone was contained in the upper 10-cm soil layer


with approximately 95% of the root zone contained in the upper 25-cm soil layer.









Typically, the wet prairie is inundated for 50 to 100 days each year, burns every two to

four years, and has low organic matter acculumulation. This prairie community is

routinely burned in accordance with simulated natural periodicity in order to maintain the

communities' natural conditions and ecological diversity. Prior to the experiment, the

study area was most recently burned in December 1999.

The site is instrumented with a tower-based meteorological station installed onsite

to provide several meteorological and micrometeorological measurements.

Instrumentation onsite also provided subsurface measurements of water content, soil

temperature and soil matric potential. Table 3 describes the instrumentation used for data

collection. Jacobs (2002b) also provides a complete description of the instrumentation

and study area. All atmospheric and flux instrumentation was mounted on a 6.1-m tower,

and logged using a CR10X datalogger (CSI, Inc.). Land surface fluxes were measured

using an eddy-flux approach. Evapotranspiration was measured directly using an energy-

budget variant of the eddy correlation approach (Tanner and Greene 1989, Twine et al.

2000). Fluctuations in wind speed, virtual air temperature, and vapor density were

sampled at 6 Hz, and 30-minute average covariances were calculated to estimate the

fluxes. The latent heat fluxes were corrected for temperature-induced fluctuations in air

density (Webb et al. 1980) and for the hygrometer sensitivity to oxygen (Tanner and

Greene 1989). Sensible heat fluxes were corrected for differences between the sonic

temperature and the actual air temperature (Schotanus et al. 1983). Both the sensible and

latent heat fluxes were corrected for misalignment with respect to the natural wind

coordinate system (Baldocchi et al. 1988). The Bowen-ratio method was used to close the

surface energy balance relationship (Twine et al. 2000). Ground heat flux was measured










approximately 10 cm below the surface. The measured soil heat flux was corrected for

the energy stored in the upper 10 cm of soil using measurements of soil temperature.

Near-surface volumetric soil water content was recorded at three depths: 7.6 cm,

12.7 cm and 17.8 cm. Soil water potential and soil temperature were also measured

within the soil profile at points coinciding with the moisture probe depths, as well as an

additional temperature measurements at 2.5 cm. Subsurface measurements were sampled

every minute, and 30-minute averages were logged using an AM 416 multiplexer coupled

with a CR10X datalogger (CSI, Inc.).

Table 3. Instrumentation used in this study

Variable Instrumentation Height (m)
Net Radiation REBS Q*7.1 6.5
Sensible Heat Flux CSI 3-D Sonic Anemometer CSAT3 5.75
Latent Heat Flux CSI Krypton Hygrometer KH20 5.75
Ground Heat Flux REBS RFT 3.1 -0.1
Wind Speed and Direction RM Young CS 800-L Anemometer 6.5
Precipitation Texas Electronincs TE525 6.5
Relative Humidity Vaisala HMP45c 5.8
Temperature Vaisala HMP45c 5.8
Pressure Vaisala PTB 100 1.0
Subsurface Temperature CSI 107 -0.025,-0.076, -0.127, -0.178
Subsurface Moisture CSI 615L -0.076, -0.127, -0.179
Subsurface Matric Potential CSI 257 -0.025,-0.076, -0.127, -0.178














CHAPTER 4
SIMULATION DESIGN

Parameterization

CLM and LSP soil parameters are characterized by soil texture composition from

percentages of sand, clay, and loam. The study area's soil is predominantly Wauberg

Sand. Parameters for CLM and LSP soil layers were determined by a weighted average

based on the soil composition from the Soil Characterization Laboratory soil composition

profile (UF-IFAS 1985). Figure 2 illustrates a comparison of the soil water characteristic

curves generated from the Clapp and Hornberger relationship used by CLM with soil

water characteristic curves derived from measured values. Figure 2 also includes the

hydraulic conductivity as it varies with volumetric water content for each of the five

CLM layers contained in the top 23 cm of the profile.

Vegetation parameterization for CLM is based on the IGBP characterization of

landscapes. The Land Process DAAC (edcdaac.usgs.gov/main.html) defines the study

area as a cropland using the IGBP land classification. This characterization, derived from

remote sensing data, may be indicative of agricultural pastures bordering the study area,

but does not provide an accurate description of the vegetative communities in the region.

Due to the hydrogeology and climatic conditions of Paynes Prairie, the plant species

within the flooded prairie community are predominately grassland and wetland species.

Therefore, the IGBP land cover classification chosen for CLM parameterization was

wetland, while within the IGBP parameterization scheme the grassland parameters for

leaf and stem reflectance and transmittance were used. CLM's leaf area index (LAI)







16


parameters are generated from annual maximum and minimum values based on IGBP


wetland properties. The temperatures of the lowest soil layer define the seasonal


relationship of LAI. LSP's vegetation parameters are user defined. Here, LSP's LAI was


proportionally increased during the experiment from 2.0 to 2.5,




400 -0 030
l Measured
3 SMP 0-9 cm depth
350 -K-SMP 9-16 5cm depth 0 025
-*--SMP 16 5-23 cm depth
300 --- --K 1 76-4 51 cm depth
-n- K 0-1 75 cm depth
K 4 52-9 05 cm depth 0 020
250 K 9 06-16 56 cm depth
K 16 57-23 cm depth

3 200 0015 E

150
0010

100

0 005
50

0 MM 0000
005 010 015 020 025 030 035 040
e

Figure 3. Comparison of measured soil water characteristic curve with three soil water
characteristic curves defined by modeled soil matric potential (SMP) and
seven hydraulic conductivity (K) curves based on the Clapp and Hornberger
relationship within the CLM parameterization scheme.

Initialization

The CLM and LSP initialization schemes require the initial soil temperature profile


and the initial soil moisture content profile. The soil temperature profile is intialized


using measured values for the top 23 cm. For the lower profile, the initial temperature is


set equal to the temperature measured at 23 cm. The soil moisture profile is initialized


using measurements in the top 17 cm of the profile. The profile is saturated below 62


cm. The initial soil moisture is assumed to increase linearly with depth between 17 cm


and the water table.









Forcings

Table 4 summarizes the forcing required by CLM and LSP. All forcing

excluding radiation, are from measured values. Estimates of incoming shortwave and

longwave radiation were derived from measured net radiation. During the nighttime

(generally 6:30 PM to 6:30 AM EDT), shortwave radiation is zero and net longwave

radiation is equivalent to measured net radiation. Daytime values of the longwave

radiation balance were estimated using Diak et al.'s (2000) remotely-sensed radiation

estimation scheme based on GOES data. Jacobs et al. (2002b) demonstrated the utility of

GOES data in the study area. The diurnal radiation surface albedo as a function of time

was determined for a cloud-free day using measured incoming shortwave radiation,

measured net radiation and GOES estimated net longwave. The solar insolation is

calculated from measured net radiation, GOES derived net longwave radiation and

surface albedo.

Table 4. Required atmospheric forcing for CLM and LSP

Model Forcings
CLM LSP
Incoming Shortwave Radiation (Wm-2) Incoming Shortwave Radiation (Wm-2)
Incoming Longwave Radiation (Wm-2) Incoming Longwave Radiation (Wm-2)
Air Temperature (K) Air Temperature (K)
Specific Humidity (-) Relative Humidity (%)
Precipitation (mm 0.5hr-1) Precipitation (mm 0.5hr-1)
Wind Speed (ms-1) Wind Speed (ms-1)
Atmospheric Pressure (hPa)














CHAPTER 5
RESULTS AND DISCUSSION

The CLM and LSP simulation results are evaluated using three categories. The

modeled soil moisture, soil temperature and surface heat fluxes are compared with

measured validation data. The two main statistical quantities used for simulation

assessment are mean absolute error (MAE) and root mean squared error (RMSE). The

mean absolute error is defined as the average of the absolute differences between

modeled and measured results:

n
modi Xmeasi
MAE = 1- -
n


Equation 1. Mean absolute error.

The root mean squared error is the square root of the average of the differences between

modeled and measured results squared, as follows:




R S (mod,i Xmeas,i
RMSE = i=-
n


Equation 2. Root mean squared error.

Soil Moisture

The observed point measurements of soil moisture were compared to the simulated

moisture for the layer that contains that measurement point. Each CLM and LSP soil









layer has homogeneous moisture and temperature. The soil moisture measurements at

7.6, 12.7 and 17.8 cm correspond to CLM layers 3, 4 and 5, respectively and to LSP

layers 11, 15 and 19. Figure 3 shows the evolution of modeled and measured soil

moisture during the study period in terms of volumetric water content (VWC).

CLM's layer 3 exhibits a more rapid dry down than the observed dry down. This

layer has a higher modeled water content than observed during the first two days of the

experiment. The model dry down rate results in the simulated layer having a lower

moisture content than the observed for all periods after four days with an increasing bias

for the duration. The MAE and RMSE for the CLM simulated results are 0.013 m3m-3

and 0.016 m3m3. LSP matches the dry down rate of the measurement at 7.6 cm,

however, there is an initial offset in the modeled results of approximately 0.03 m3m3.

This initial offset results in a high bias of simulated moisture contents for the entire

experimental period. The LSP simulated results demonstrate a MAE and RMSE of 0.034

m3m- and 0.035 m3m-3.

The CLM simulation results corresponding to the 12.7 cm measurement

demonstrate good agreement with the measured dry down rate. A small bias results in

simulated moisture content values that are lower than measured by approximately 0.03

m3m-3 throughout the study period. The LSP model shows very good agreement with the

measured values over the period. The MAE and RMSE for the LSP modeled results are

0.009 m3m-3 and 0.011 m3m-3. Both CLM and LSP show a similar rate of dry down for

the measurements at 12.7 and 17.8 cm and provide good moisture content estimates at the

deepest measurement depth. CLM's moisture contents are slightly drier (0.01 m3m-3)

than measured, while LSP's modeled water contents are slightly wetter (0.02 m3m3).






















a)
MAE RMSE

0 25 CLM 0013 0016
LSP 0 034 0 035


E 02



015



01


Measured CLM Modeled LSP Modeled
005 -
124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)


03



0 25-



E 02



0 15



01


MAE RMSE
CLM 0 032 0 033
LSP 0009 0011


- Measured -- CLM Modeled LSP Modeled


124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)





C)
MAE RMSE

5 CLM 0011 0013
LSP 0 020 0 022






5 -



1


02



E 0
0O

001




0


-Measured --CLM Modeled -LSP Modeled
0 05
124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)



Figure 4. Comparison of measured soil moisture with CLM and LSP modeled soil

moisture at a) 7.6 cm, b) 12.7 cm and c) 17.8 cm.









The measurement error associated with a CS615 moisture probe is 2.5%

volumetric water content. CLM simulations at 7.6 and 17.8 cm depths have errors less

than the instrument error. The CLM simulation at 12.7 cm is less than 1.5 times the

instrument error. LSP's 7.6 cm depth exhibits error of less than 1.5 times the instrument

error. LSP's 12.7 cm and 17.8 cm depths show better agreement with an error less than

the instrument error. The volumetric soil moisture errors identified in this study are

consistent with the results of other studies. In PILPS Phase 2(b), Shao and Henderson-

Sellers (1996) recognized 3% volumetric water content error margins as reasonable,

while Mohr et al. (2000) found error margins in the near surface soil simulation of 5%

volumetric soil moisture.

Comparisons of the modeled results to the observations suggest different strengths

and weaknesses for each model. For all three depths, both models capture the diurnal

fluctuations in moisture. The observations showed increasing soil moisture as the soil

water profile reestablished during the evening. LSP did an excellent job of modeling the

phase and amplitude of the measured soil moisture fluctuations throughout the

experiment period. Both are able to capture the magnitude and location of soil water

extraction due to evapotranspiration. While, CLM's daytime dry down appears to

function appropriately, CLM was unable to replicate the recharge from the lower layers.

The water retention curves used by the models appear to result in two distinct

errors. For LSP, the soil water profile rapidly equilibriates to a biased profile. The LSP

soil water retention curve appears shifted such that the soil matric potential for each depth

corresponds to a higher moisture content than the moisture content that is measured.

CLM overestimates the daily dry down rate, because it does not redistribute water from









lower layers to upper layers during the nighttime. This pattern can be explained by

examining the CLM soil water characteristic curves in Figure 2. For soil moisture

contents between 0.15 m3m-3 and 0.17 m3m3, the slope of the measured retention curve is

very steep, however, the CLM water retention curve is relatively flat in this range.

Daytime root water extraction results in an observed soil water profile that increases in

depth from the surface. This profile results in a large gradient of matric potentials in

response to the moisture content profile that drives soil water upward in the soil profile

during the evening. However, in CLM's modeled soil column, comparable moisture

profiles exhibit a much smaller gradient of matric potential. Thus, CLM has a reduced

upward movement of moisture and increasingly dryer surface conditions.

Soil Temperature

The comparison of the modeled soil temperature profile with observed soil

temperatures follows that methodology for the soil moisture comparison. Soil

temperature and moisture were measured at the same depths with an additional

temperature measurement at 2.5 cm depth. The time evolution of modeled CLM and LSP

soil temperatures compared with measured values for the observation period are seen in

Figure 4. Overall, both models show good agreement with the timing and magnitude of

the actual diurnal fluctuations. Simulated soil temperatures of both CLM and LSP for all

depths show a warming trend over the study period. For the 2.5 cm depth, the maximum

daily soil temperature increases by 2 K over the two-week period. However, the

simulated maxima are somewhat warmer than the measured maximum during the latter

parts of the experiment. For the second depth, the maxima of both simulations are within

1 K of the measured maximum early in the experiment and greater than 3.0 K and 2.0 K

for CLM and LSP, respectively, at the end of the experiment. At the 7.6 cm









measurement depth, both simulations lead the measured diurnal fluctuations by 2.0 to 3.0

hours throughout the study period. While the temperatures are quite similar, the shift

increases in errors to 1.2 K and 1.4 K for CLM and 0.8 K and 0.98 K for LSP. The 12.7

and 17.8 cm comparisons do not have a temporal discrepancy. Both comparisons exhibit

the same enhanced warming trend that is evident at both 2.5 and 7.6 cm, as shown in

Figures 4 c and d. At 12.7 cm, the early diurnal cycles show slightly lower modeled

temperature maxima than the measured temperature maximum. The agreement improves

during the experimental period with the final diurnal cycle exhibiting good agreement as

compared to the measurements. The MAE and RMSE are 0.45 K and 0.54 K for CLM

and 0.52 K and 0.67 K for LSP. The 17.8 cm measurements compare well with both

simulated time series for the first 10 days with the final three cycles showing a small

warm bias of 1.5 K between the measured and modeled temperatures at all points during

the cycle. The lowest layer had the smallest error for both CLM with a MAE and RMSE

of 0.35 K and 0.44 K and LSP with errors of 0.33 K and 0.41 K.

For both models, the modeled error decreases with depth, as the diurnal

fluctuations are damped by the increasing soil thickness. The measurement error

associated with a CS107 temperature probe is + 0.5 K. Analysis of the CLM modeled

soil temperatures shows an agreement with the 2.5 cm and 7.6 cm measured values

within 2.5 times the instrument error. The comparison of the CLM modeled temperatures

with the 12.7 cm and the 17.8 cm measured values show errors less than the instrument

error. The LSP simulation shows a similar trend. From the 2.5 to 17.8 cm depths, the

ratio of LSP's MAE's compared to the instrument errors are less than 2.5, 2.0, 1.5 and

1.0, respectively. The errors of the simulated soil temperatures compare reasonably











































Time (days)


E ...
S297

S295

E 293
291

E 299
289
287









2985






E 299
297
T295

E- 2 9 3

291



287
285





305

303

301

E 299

297

295
' 293
E
12 291


c)
MAE RMSE
CLM 0 5 0 5
LSP 0 51 0 67











Measured- -CLMModeled-LSPModeled



Time (days)



d)
MAE RMSE
CLM 0 35 0 44
LSP 0 33 0 40


Time (days)


Figure 5. Comparison of measured subsurface soil temperature and CLM and LSP

modeled soil temperature at a) 2.5 cm, b) 7.6 cm, c) 12.7 cm and d) 17.8 cm.


MAE RMSE


Time (days)









well with similar modeling studies. Both Chen et al. (1997) and Mohr et al. (2000) found

errors within +2 K for subsurface temperatures. The offline simulations of PILPS Phase

1 found ranges between models of 1.4 K for tropical forest and 2.2 K for grassland in a

multi-year study. Judge et al. (2003b) found a 1.8 K average difference between modeled

and measured temperature for bare soil and a 1.0 K mean difference for brome grass.

In general, the modeled temperatures for both models show a warming trend over

the duration of the period at all depths. This trend reflects the observed dry down and

warming conditions. CLM's simulated diurnal temperature fluctuations are greater in

magnitude than LSP's. This likely reflects the decreased thermal heat capacity of CLM's

relatively drier soil. Despite the higher moisture content, the LSP layers' temperatures

generally show higher fluctuations in temperature than the measurements. This could be

due to the soil texture parameterized thermal capacity of dry soil within LSP being lower

than the actual thermal capacity of the soil solids.

Surface Heat Fluxes

The comparison of modeled surface fluxes with observations from field data

demonstrates some disparity in the modeled energy balance for both models. Figure 5a

demonstrates the agreement of CLM modeled net radiation and measured net radiation.

As both downwelling longwave and shortwave radiation force CLM, the agreement of

the net radiation demonstrates the accuracy of the simulated reflected radiation or albedo

and upwelling longwave radiation. Figures 5b, 6a and 6b illustrate the differences

between the observed fluxes and the CLM and LSP modeled fluxes for three components

of the energy balance: latent heat flux, sensible heat flux and ground heat flux. The

ground heat flux of CLM is compared to the measured ground heat flux at the surface (10

cm depth measurement corrected for soil heat storage), while the LSP ground heat flux









for model layer 13 is comparable to the actual 10 cm flux measurement (Figure 6b). The

MAE and RMSE were calculated using only daytime values (6:30 am until 6:30 pm) due

to missing nighttime measurements resultant from dew on the lens of the krypton

hygrometer.

The CLM results for latent heat fluxes show the best agreement, while the sensible

heat and ground heat fluxes have larger errors. The LSP results have the largest

discrepancy between simulated and measured fluxes for latent heat fluxes, whereas the

sensible heat and ground heat fluxes demonstrate better agreement. The diurnal variation

in all energy fluxes is well captured by both models, with the exception of the LSP latent

heat flux. The diurnal variation of the simulated latent heat flux increasingly lags the

measured diurnal variation as the experimental period progresses. At the beginning of

the experiment, the diurnal variation of the simulated latent heat fluxes lags the measured

by 2 hrs. The lag is approximately 3 hrs during the last day of the experiment. The MAE

and RMSE derived from comparisons of the measured and modeled latent heat for CLM

is 44.79 Wm-2 and 61.69 Wm-2. The corresponding statistics for the LSP simulation are

99.43 Wm-2 and 122.32 Wm-2. A MAE of 47.09 Wm-2 and a RMSE of 60.92 Wm-2 is

derived from comparisons of the measured and CLM simulated sensible heat flux. LSP

has much better agreement with measured sensible heat flux with respective MAE and

RMSE values of 23.50 Wm-2 and 29.08 Wm-2. CLM has relatively high errors of 20.83

Wm-2 and 26.19 Wm-2 for the MAE and RMSE of the simulated ground heat flux as

compared to LSP's errors of 4.27 Wm-2 and 5.31 Wm-2.

The MAE associated with the simulated daytime CLM latent heat flux represents

19% of the average 232.11 Wm-2 measured latent heat flux. The MAE of the LSP









simulation is 43% of the average measured latent heat. The magnitude of the CLM

daytime sensible heat MAE is approximately 48% of the magnitude of the average

measured sensible heat flux, 98.7 Wm-2. A similar examination of LSP shows the

magnitude equal to approximately 24% of the average measured sensible heat flux. The

MAE resulting from comparisons of the measured and CLM daytime simulated ground

heat flux is nearly equal to the average magnitude of the measured ground heat flux of

22.55 Wm-2 representing approximately 92% of the magnitude. The MAE of LSP

daytime simulated ground heat flux is only 19% of the average measurement magnitude.

Simulation of surface fluxes often provides the largest discrepancies from

measurements. Chen et al. (1997) found reasonable agreement with ranges across

modeling schemes of 30 Wm-2 and 25 Wm-2 for sensible and latent heat flux,

respectively. Chang et al. (1999) also found good agreement using the CAPS model,

with a monthly discrepancy of 7.1 Wm-2 in latent heat flux and 7.7 Wm-2 for sensible heat

flux. However the same study found discrepancies in diurnal amplitude of ground heat

flux around 20 Wm-2 with a phase difference of 2.5 hours. Both Chen et al. (1997) and

Chang et al. (1999) calculated error margins based on the full diurnal cycle. A trend

evident in several SVAT model validation studies (Acs and Hantel 1998, Chang et al.

1999, Gonzalez-Sosa et al. 2001) is that latent heat fluxes are more accurately simulated

than sensible heat fluxes and ground heat fluxes. This trend suggests that the model

physics representing the processes of latent heat flux are more comprehensive than the

model physics describing the sensible or ground heat flux. The results of the CLM

simulation provide another example of this tendency. In contrast, the LSP model has a

tendency to underpredict the latent heat fluxes as the soil column dries out and









evapotranspiration conditions drop below potential, while accurately simulating the

sensible and ground heat fluxes.

As the climatic conditions of the southeastern US may significantly influence

model performance, a comparison of modeled fluxes were examined for two days that

exhibit large differences in cloud cover. Within the experimental period, May 7th

represents a typical cloudy day, whereas May 13th is a typical clear-sky day. The average

measured net radiation is 210.4 Wm-2 on May 7th and 394.2 Wm-2 on May 13th. Figures 7

a, b and c illustrate the latent heat flux, sensible heat flux and ground heat fluxes for May

7th. Both CLM and LSP underpredict the latent heat fluxes during the first 5.0 hours of

the cloudy diurnal cycle. In order to maintain the energy balance, both CLM and LSP

overpredict the measured sensible heat during the same period. CLM also overpredicts

ground heat fluxes for the first half of the day, while LSP shows very good agreement

throughout the day. During midday, the net radiation shows high variability that is

indicative of cumulus atmospheric conditions. CLM closely tracks the variation in net

radiation, but overpredicts the effect of that variability on the measured latent and

sensible heat fluxes. While LSP demonstrates similar patterns, LSP better captures the

effect of the variability. This may be a function of LSP's higher temporal resolution of

the simulation.

Figures 8 a, b and c demonstrate the latent heat, sensible heat and ground heat

fluxes for the relatively clear skies observed on May 13th. CLM's simulated latent heat

fluxes underpredict measured latent heat fluxes for the first three hours of the diurnal

cycle. After the initial period, the measured values are overpredicted until the peak.

CLM's latent heat and sensible heat fluxes show similar agreement with the









measurements. Whereas the modeled latent heat fluxes overpredict midday values, the

modeled sensible heat fluxes underpredict during the same period. LSP's simulated

latent heat fluxes underpredict the measured latent heat fluxes for the first 10 hours of the

day. The diurnal variation of LSP's simulated latent heat fluxes has a 3 hr phase shift.

LSP's simulated sensible heat and ground heat fluxes show very good agreement with the

measured fluxes over the course of the day. Overall, both models seem to simulate the

surface energy fluxes well in both clear sky and convective climatic conditions. These

results support the validity of the CLM and LSP model physics for determination of high

temporal resolution applications for climates typical of the southeastern US.

















800

700 a)

600

CT1 500-

400










-100-
Measured --CL MAE RMSE
200- CLM 1920 2390
S200 -------
4 100 -







-100 -










-Measured- -CLM Modeled LSP Modeled
-200
124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)

600and b) CLM and LSP latent heat flux.
b) MAE RMSE
CLM 44 79 61 69
50 -LSP 99 44 122 32


400


x 300


200





0-

[ Measured -CLM Modeled LSP Modeled
-100
124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)


Figure 6. Comparison of modeled and measured surface fluxes for a) CLM net radiation
and b) CLM and LSP latent heat flux.






















Measured
- -CLM Modeled
- LSP Modeled


MAE RMSE
CLM 47 09 60 92
LSP 23 50 29 08


-100 -
124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)


600



500


E 400



z 300
F-

a)
200



0 100

0
0


100 2


x
0 -3
IL-


-100 "

o
-200


-100 -4C
124 125 126 127 128 129 130 131 132 133 134 135 136 137

Time (days)


Figure 7. Comparison of modeled and measured surface fluxes for a) CLM and LSP
sensible heat flux and b) CLM and LSP ground heat flux.


600


500


E 400

X
.2 300
IL-


200


S 100


0











1000 --
a)
900 -a)

800 -

700
E 600

500

400 -

S300 -

200
-J


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
EDST (Hours)


100 ----
100 Net Radiation -- Measured Sensible Heat -*-CLM Modeled --LSP Modeled

200
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
EDST (Hours)


--- Measured Ground Heat at Surface
-*-CLM Modeled
-.-LSP Modeled at 10 cm
-t--Measured Ground Heat at 10 cm


400

300

200

100 r-,
E
0

-100 X
u-
-200

-300 "

-400

-500 0

-600

-700

-800


0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

EDST (Hours)


18 19 20 21 22 23 24


Figure 8. Comparison of measured and modeled surface energy fluxes for a typical

cloudy day, May 7th for a) Latent heat flux, b) Sensible heat flux and c)

Ground heat fluxes.


--- mdodod&M.NMN











1000 -
900 a)

800

700
E 600

500

400

X 300

200


100

-100 ----
-- Net Radiation ---Measured Latent Heat -*-CLM Modeled -.-LSP Modeled
-200
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
EDST (Hours)

1000

900 b)

800
700

S600

500

400

300
'V 200

100

0 -

-100 Net Radiation -.-Measured Sensible Heat -*-CLM Modeled -*-LSP Modeled

-200
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
EDST (Hours)

1000 400

900 c) ~-- Measured Ground Heat at Surface 300

800 -*-CLM Modeled 200
-8-LSP Modeled at 10 cm
700 -*-Measured Ground Heat at 10 cm 100 -


X 500 -100 X

400 -200

300 -300 -

200 -400
o o
9 100 -500 (9

0 .600

-100 -700

-200 -800
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
EDST (Hours)



Figure 9. Comparison of measured and modeled surface energy fluxes for a typical clear

day, May 13th for a) Latent heat flux, b) Sensible heat flux and c) Ground heat

fluxes.














CHAPTER 6
CONCLUSION

The purpose of this research was to compare two SVAT models' applicability in a

high water table environment characteristic of the southeastern US. As all SVAT models

rely on similar mathematical approximations of biophysical reality, the main difference

between a field scale model and a watershed scale single-column model is

parameterization. LSP's strength is its meticulous reproduction of biophysical

mechanisms at the land surface. In order to provide an exhaustive simulation of land

surface processes, the LSP model can use all available data to parameterize the canopy

and soil profile. Alternatively, CLM has an inherent database that contains default

parameterization schemes for each IGBP land cover.

The results indicate that both CLM and LSP can be used with success in the

convective climate and high water table environment of the southeastern US. The two

models were used to simulate a 13-day dry drown. The simulations show that subsurface

moisture content and surface energy fluxes provide good agreement with measured

values. Both models' simulated soil temperatures show very good agreement as

compared to measured temperatures. The models were able to appropriately replicate the

soil warming that occurred over the experimental period as well as the diurnal

fluctuations. Both models simulated energy fluxes within expected error ranges. When

considering the field scale of the experiment, CLM performs quite well in comparison

with the biophysically more sophisticated LSP model. However, for use in the

southeastern US, it is important to alter CLM's baseflow generation physics mechanism






35


to account for the low relief topography of the region. Overall, the experiment

demonstrates potential for the application of both CLM and LSP in the southeastern US.
















LIST OF REFERENCES


Acs, F. and M. Hantel, 1998: The land-surface flux model PROGSURF. Global and
Planetary Change, 19, 19-34.

Baldocchi, D. D., B. B. Hicks, and T. P. Meyers, 1988: Measuring Biosphere-
Atmosphere Exchanges of Biologically Related Gases with Micrometeorological
Methods. Ecology, 69, 1331-1340.

Beven, K. J., R. Lamb, P. Quinn, R. Romanowicz, and J. Freer, 1995: Topmodel.
Computer Models of Watershed Hydrology, Water Resource Publications,
Highlands Ranch, CO, 627-668.

Blackmon, M., B. Boville, F. Bryan, R. Dickinson, P. Gent, J. Kiehl, R. Moritz, D.
Randall, J. Shukla, S. Solomon, G. Bonan, S. Doney, I. Fung, J. Hack, E. Hunke, J.
Hurrell, J. Kutzbach, J. Meehl, B. Otto-Bliesner, R. Saravanan, E. K. Schneider, L.
Sloan, M. Spall, K. Taylor, J. Tribbia, and W. Washington, 2001: The Community
Climate System Model. Bulletin of the American Meteorological Society, 82, 2357-
2376.

Bonan, G. B., 1996: A land surface model (LSM version 1.0) for ecological,
hydrological, and atmospheric studies: Technical Description and User's Guide.
NCAR Technical Note, NCAR/TN-417+STR, 150.

Bonan, G. B., K. W. Oleson, M. Vertenstein, S. Levis, X. B. Zeng, Y. J. Dai, R. E.
Dickinson, and Z. L. Yang, 2002: The land surface climatology of the community
land model coupled to the NCAR community climate model. Journal of Climate,
15, 3123-3149.

Chang, S., D. Hahn, C. H. Yang, and D. Norquist, 1999: Validation study of the CAPS
model land surface scheme using the 1987 Cabauw/PILPS dataset. Journal of
AppliedMeteorology, 38, 405-422.









Chen, T. H., A. HendersonSellers, P. C. D. Milly, A. J. Pitman, A. C. M. Beljaars, J.
Polcher, F. Abramopoulos, A. Boone, S. Chang, F. Chen, Y. Dai, C. E.
Desborough, R. E. Dickinson, L. Dumenil, M. Ek, J. R. Garratt, N. Gedney, Y. M.
Gusev, J. Kim, R. Koster, E. A. Kowalczyk, K. Laval, J. Lean, D. Lettenmaier, X.
Liang, J. F. Mahfouf, H. T. Mengelkamp, K. Mitchell, 0. N. Nasonova, J. Noilhan,
A. Robock, C. Rosenzweig, J. Schaake, C. A. Schlosser, J. P. Schulz, Y. Shao, A.
B. Shmakin, D. L. Verseghy, P. Wetzel, E. F. Wood, Y. Xue, Z. L. Yang, and Q.
Zeng, 1997: Cabauw experimental results from the project for intercomparison of
land-surface parameterization schemes. Journal of Climate, 10, 1194-1215.

Clapp, R. B. and G. M. Hornberger, 1978: Empirical Equations for Some Soil Hydraulic-
Properties. Water Resources Research, 14, 601-604.

Cosby, B. J., G. M. Hornberger, R. B. Clapp, and T. R. Ginn, 1984: A Statistical
Exploration of the Relationships of Soil-Moisture Characteristics to the Physical-
Properties of Soils. Water Resources Research, 20, 682-690.

Dai, Y., X. Zeng, and R. Dickinson, 2001: Common Land Model (CLM): Technical
Documentation and User's Guide. (climate.eas.gatech.edu/dai/clmdoc.pdf), 69.

Diak, G. R., W. L. Bland, J. R. Mecikalski, and M. C. Anderson, 2000: Satellite-based
estimates of longwave radiation for agricultural applications. Agricultural and
Forest Meteorology, 103, 349-355.

Dickinson, R., A. Henderson-Sellers, P. J. Kennedy, and M. F. Wilson, 1993: Biosphere
atmosphere transfer scheme (BATS) version le as coupled to the NCAR
Community Climate Model. NCAR Technical Note, NCAR/TN-378+STR, 72.

Ghildyal, B. P. and R. Tripathi, 1987: Soil Physics. New York City, NY, John Wiley.

Gonzalez-Sosa, E., I. Braud, J. L. Thony, M. Vauclin, and J. C. Calvet, 2001: Heat and
water exchanges of fallow land covered with a plant- residue mulch layer: a
modelling study using the three year MUREX data set. Journal of Hydrology, 244,
119-136.

Green, W. and G. Ampt, 1911: Studies on soil physics. Journal ofAgricultural Science,
4, 1-24.

Jacobs, J. M., D. A. Myers, M. C. Anderson, and G. R. Diak, 2002a: GOES surface
insolation to estimate wetlands evapotranspiration. Journal of Hydrology, 266, 53-
65.

Jacobs, J. M., S. L. Mergelsberg, A. F. Lopera, and D. A. Myers, 2002b:
Evapotranspiration from a wet prairie wetland under drought conditions: Paynes
Prairie Preserve, Florida, USA. Wetlands, 22, 374-385.









Judge, J., L. M. Abriola, and A. W. England, 2003a: A summertime Land Surface
Process and Radiobrightness model Part I: Development and Numerical
Validation. Advances in Water Resources, in review.

Judge, J., J. Metcalfe, D. McNichol, B. E. Goodison, A. E. Walker, and A. W. England,
2003b: A summertime Land Surface Process and Radiobrightness model Part II:
Model Calibration. Water Resources Research, in review.

Kim, E. J., 1999: Remote sensing of land surface conditions in arctic tundra regions for
climatological applications using microwave radiometry, PhD thesis, University of
Michigan.

Koster, R. D., M. J. Suarez, A. Ducharne, M. Stieglitz, and P. Kumar, 2000: A
catchment-based approach to modeling land surface processes in a general
circulation model 1. Model structure. Journal of Geophysical Research-
Atmospheres, 105, 24809-24822.

Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges, 1994: A Simple
Hydrologically Based Model of Land-Surface Water and Energy Fluxes for
General-Circulation Models. Journal of Geophysical Research-Atmospheres, 99,
14415-14428.

Liang, X., E. F. Wood, and D. P. Lettenmaier, 1996: Surface soil moisture
parameterization of the VIC-2L model: Evaluation and modification. Global and
Planetary Change, 13, 195-206.

Liang, X., E. F. Wood, D. P. Lettenmaier, D. Lohmann, A. Boone, S. Chang, F. Chen, Y.
J. Dai, C. Desborough, R. E. Dickinson, Q. Y. Duan, M. Ek, Y. M. Gusev, F.
Habets, P. Irannejad, R. Koster, K. E. Mitchell, 0. N. Nasonova, J. Noilhan, J.
Schaake, A. Schlosser, Y. P. Shao, A. B. Shmakin, D. Verseghy, K. Warrach, P.
Wetzel, Y. K. Xue, Z. L. Yang, and Q. C. Zeng, 1998: The Project for
Intercomparison of Land-surface Parameterization Schemes (PILPS) phase 2(c)
Red-Arkansas River basin experiment: 2. Spatial and temporal analysis of energy
fluxes. Global and Planetary Change, 19, 137-159.

Liou, Y. A., J. F. Galantowicz, and A. W. England, 1999: A land surface process
radiobrightness model with coupled heat and moisture transport for prairie
grassland. Ieee Transactions on Geoscience and Remote Sensing, 37, 1848-1859.

Loveland, T. R., B. C. Reed, J. F. Brown, D. 0. Ohlen, Z. Zhu, L. Yang, and J. W.
Merchant, 2000: Development of a global land cover characteristics database and
IGBP DISCover from 1 km AVHRR data. International Journal of Remote
Sensing, 21, 1303-1330.

Mitsch, W. J. and J. G. Gosselink, 2000: Wetlands. New York City, NY, John Wiley.









Mohr, K. I., J. S. Famiglietti, A. Boone, and P. J. Starks, 2000: Modeling soil moisture
and surface flux variability with an untuned land surface scheme: A case study
from the Southern Great Plains 1997 Hydrology Experiment. Journal of
Hydrometeorology, 1, 154-169.

Nijssen, B., I. Haddeland, and D. P. Lettenmaier, 1997: Point evaluation of a surface
hydrology model for BOREAS. Journal of Geophysical Research-Atmospheres,
102, 29367-29378.

Philip, J. R. and D. De Vries, 1957: Moisture movement in porous materials under
temperature gradients. Transactions, American Geophysical Union, 38, 222-232.

Philip, J. R., 1957: Theory of infiltration, 1. The infiltration equation and its solution. Soil
Science, 83, 345-357.

-, 1987a: Inverse solution for one-dimensional infiltration and ratio A/K1. Water
Resources Research, 26, 2023-1017.

-, 1987b: The Infiltration Joining Problem. Water Resources Research, 23, 2239-
2245.

Pitman, A. J. and A. Henderson-Sellers, 1998: Recent progress and results from the
project for the intercomparison of landsurface parameterization schemes. Journal of
Hydrology, 213, 128-135.

Pitman, A. J., A. Henderson-Sellers, C. E. Desborough, Z. L. Yang, F. Abramopoulos, A.
Boone, R. E. Dickinson, N. Gedney, R. Koster, E. Kowalczyk, D. Lettenmaier, X.
Liang, J. F. Mahfouf, J. Noilhan, J. Polcher, W. Qu, A. Robock, C. Rosenzweig, C.
A. Schlosser, A. B. Shmakin, J. Smith, M. Suarez, D. Verseghy, P. Wetzel, E.
Wood, and Y. Xue, 1999: Key results and implications from phase 1(c) of the
Project for Intercomparison of Land-Surface Parametrization Schemes. Climate
Dynamics, 15, 673-684.

Rossi, C. and J. R. Nimmo, 1994: Modeling of Soil-Water Retention from Saturation to
Oven Dryness. Water Resources Research, 30, 701-708.

Schaake, J., Q. Duan, V. Koren, and A. Hall, 2001: Toward improved parameter
estimation of land surface hydrology models through the Model Parameter
Estimation Experiment (MOPEX). Soil-Vegetation-Atmosphere Transfer Schemes
andLarge-Scale Hydrological Models, Maastricht, The Netherlands, IAHS
Pblishing, 91-97.

Schotanus, P., F. T. M. Nieuwstadt, and H. A. R. Debruin, 1983: Temperature-
Measurement with a Sonic Anemometer and Its Application to Heat and Moisture
Fluxes. Boundary-Layer Meteorology, 26, 81-93.









Shao, Y. P. and A. Henderson-Sellers, 1996: Validation of soil moisture simulation in
landsurface parameterisation schemes with HAPEX data. Global and Planetary
Change, 13, 11-46.

Slater, A. G., C. A. Schlosser, C. E. Desborough, A. J. Pitman, A. Henderson-Sellers, A.
Robock, K. Y. Vinnikov, K. Mitchell, A. Boone, H. Braden, F. Chen, P. M. Cox, P.
de Rosnay, R. E. Dickinson, Y. J. Dai, Q. Duan, J. Entin, P. Etchevers, N. Gedney,
Y. M. Gusev, F. Habets, J. Kim, V. Koren, E. A. Kowalczyk, 0. N. Nasonova, J.
Noilhan, S. Schaake, A. B. Shmakin, T. G. Smirnova, D. Verseghy, P. Wetzel, X.
Yue, Z. L. Yang, and Q. Zeng, 2001: The representation of snow in land surface
schemes: Results from PILPS 2(d). Journal ofHydrometeorology, 2, 7-25.

Stieglitz, M., D. Rind, J. Famiglietti, and C. Rosenzweig, 1997: An efficient approach to
modeling the topographic control of surface hydrology for regional and global
climate modeling. Journal of Climate, 10, 118-137.

Tanner, B. D. and J. P. Greene, 1989: Measurements of sensible heat flux and water
vapor fluxes using eddy correlation methods. DAAD 09-87, 94 pp.

Twine, T. E., W. P. Kustas, J. M. Norman, D. R. Cook, P. R. Houser, T. P. Meyers, J. H.
Prueger, P. J. Starks, and M. L. Wesely, 2000: Correcting eddy-covariance flux
underestimates over a grassland. Agricultural and Forest Meteorology, 103, 279-
300.

University of Florida-Institute of Food Agricultural Sciences. 1985. Characterization data
for selected soils. Soil Science research report No. 85-1

van der Keur, P., S. Hansen, K. Schelde, and A. Thomsen, 2001: Modification of DAISY
SVAT model for potential use of remotely sensed data. Agricultural andForest
Meteorology, 106, 215-231.

Verseghy, D. L., N. A. McFarlane, and M. Lazare, 1993: Class a Canadian Land-
Surface Scheme for Gcms .2. Vegetation Model and Coupled Runs. International
Journal of Climatology, 13, 347-370.

Webb, E. K., G. I. Pearman, and R. Leuning, 1980: Correction of Flux Measurements for
Density Effects Due to Heat and Water-Vapor Transfer. Quarterly Journal of the
Royal Meteorological Society, 106, 85-100.

Wood, E. F., D. P. Lettenmaier, X. Liang, D. Lohmann, A. Boone, S. Chang, F. Chen, Y.
J. Dai, R. E. Dickinson, Q. Y. Duan, M. Ek, Y. M. Gusev, F. Habets, P. Irannejad,
R. Koster, K. E. Mitchel, 0. N. Nasonova, J. Noilhan, J. Schaake, A. Schlosser, Y.
P. Shao, A. B. Shmakin, D. Verseghy, K. Warrach, P. Wetzel, Y. K. Xue, Z. L.
Yang, and Q. C. Zeng, 1998: The Project for Intercomparison of Land-surface
Parameterization Schemes (PILPS) phase 2(c) Red-Arkansas River basin
experiment: 1. Experiment description and summary intercomparisons. Global and
Planetary Change, 19, 115-135.






41


Zeng, X. B., M. Shaikh, Y. J. Dai, R. E. Dickinson, and R. Myneni, 2002: Coupling of
the common land model to the NCAR community climate model. Journal of
Climate, 15, 1832-1854.















BIOGRAPHICAL SKETCH

Brent Whitfield was born in 1978 in West Palm Beach, Florida. He received an

International Baccalaureate diploma from Suncoast Community High School in 1996,

after which he began his undergraduate studies at the University of Florida. In 2001,

Brent graduated with a Bachelor of Science in civil engineering. In the summer of 2001

he joined the Summer Institute for Atmospheric and Hydrospheric Sciences at NASA's

Goddard Space Flight Center in Greenbelt, Maryland. In the fall of 2001, Brent began

his studies as a candidate for a Master of Engineering in the Department of Civil and

Coastal Engineering under the tutelage of his major professor, Dr. Jennifer Jacobs.