|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 IMPACT ANALYSIS OF CHANGING RIVERINE FLOOD FREQUENCIES CAUSED BY CLIMATE CHANGE ON TRANSPORTATION INFRASTRUCTURE AND LAND USE A CASE STUDY OF PENSACOLA, FLORIDA By SUWAN SHEN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF T HE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN URBAN AND REGIONAL PLANNING UNIVERSITY OF FLORIDA 2010
2 2010 Suwan Shen
3 To my loving grandparents
4 ACKNOWLEDGMENTS I would like to tha nk all my committee members, Dr. Peng, Dr. Zwick, and Dr. Waylen, for their mentoring, keen assistance, and generous support. I would also like to thank my parents for their constant support and loving encouragement, which motivated me to complete this mil estone. Finally, I would like to thank all my friends for helping me through the rough times and never giving up on me.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 7 ABSTRACT ................................ ................................ ................................ ..................... 8 CHAPTER 1 I NTRODUCTION ................................ ................................ ................................ .... 10 2 LITERATURE REVIEW ................................ ................................ .......................... 13 2.1 Traditional F lood P revention in T ransportation P lanning and E ngineering ....... 14 2.2 Climate Change I mpact A ssessment and R esearch G aps ............................... 17 2.3 Precipitation Prediction and Flood Estimation ................................ ................... 19 3 DATA AND METHODOLOGY ................................ ................................ ................ 23 3.1 Flood Prediction ................................ ................................ ................................ 24 3.1.1 Spatial Interpolation Regression ................................ .............................. 25 3.1.2 Establish the Relationship between Antecedent Precipitation and Floods ................................ ................................ ................................ ........... 26 3.1.3 Extract Future Rainfall Pattern ................................ ................................ 30 3.2 Terrain Data Processing (Flood Map Generation) ................................ ............ 32 3.3 Impact Assessment ................................ ................................ ........................... 33 4 RESULTS ................................ ................................ ................................ ............... 38 4.1 Impact Assessment ................................ ................................ ........................... 38 4.2 Implications to U rban P lanning and T ransportation E ngineering ...................... 39 5 CONCLUSIONS AND DISCUSSIONS ................................ ................................ ... 45 APPENDIX: PLOTS OF PEARSON CORRELATION BETWEEN PARTIAL WEIGHTED RAINFALL SUM AND DISCHARGE VERSUS LENGTH OF PARTIAL SUM FOR EACH OF THE BASIN AREA ................................ ................ 49 LIST OF REFERENCES ................................ ................................ ............................... 58 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 62
6 LIST OF TABLES Table page 3 1 Current d ischarge with d ifferent r eturn p eriods in the s tudy a rea ....................... 26 3 2 Location of the d ischa rge s tations used i n the s tudy ................................ .......... 27 3 3 Preceding p eriod with m aximum c orrelation between a ccumulated r ainfall a mount and d ischarge v alues ................................ ................................ ............. 27 3 4 Linear r egression m odel p arameters for e ach b asin ................................ ........... 29 3 5 Estimated r ainfall s ufficient to g enerate f loods with v arying r eturn p eriods ........ 30 3 6 Future r ainfall and d ischarge v alues with d ifferent r eturn p eriods ....................... 32 4 1 Floodplain i ncreases ................................ ................................ ........................... 38 4 2 Transportation i nfrastructures at r isk ................................ ................................ .. 39 4 3 Inundation l and b y d ifferent l and u se t ypes ( u rban and b uilt u p) ........................ 40 4 4 Projected f lood r eturn p eriods ................................ ................................ ............. 41 4 5 Increase of f lood d esign f requency ................................ ................................ ..... 42
7 LIST OF FIGURES Figure page 3 1 City of Pensacola ................................ ................................ ................................ 23 3 2 Overall research framework. ................................ ................................ ............... 24 3 3 Ungauged points within the study area. ................................ .............................. 26 3 4 Relationship between standard runoff and 10 days accumulated rainfall ........... 28 3 5 Annual m aximum t en d ay a c cumulated p recipitation Upper Shillong .............. 31 3 6 Generalized v alue e xtreme d istribution of t en d ay a ccumulated r ainfall in Upper Shillong ................................ ................................ ................................ .... 31 3 7 Change of discharge at study area ................................ ................................ ..... 32 3 8 Current and f uture 50 year return period f lood m aps ................................ .......... 34 3 9 Current an d f uture 100 year return period f lood m aps ................................ ........ 36 4 1 Change of f lood a rea ................................ ................................ .......................... 38 4 2 Inundation l and by d ifferent l and u se t ype (50 y ear f lood) ................................ .. 43 4 3 Inundation l and by d ifferent l and u se t ype (100 year f lood) ................................ 43 4 4 Probability d istribution of c urrent d isch arge at s ite A ................................ .......... 44 4 5 Probability d istribution of c urrent d ischarge at s ite B ................................ .......... 44
8 Abstract of Thesis Presented to the Graduate School of the University of F lorida in Partial Fulfillment of the Requirements for the Degree of Master of Arts in Urban and Regional Planning IMPACT ANALYSIS OF CHANGING RIVERINE FLOOD FREQUENCIES CAUSED BY CLIMATE CHANGE ON TRANSPORTATION INFRASTRUCTURE AND LAND US E A CASE STUDY OF PENSACOLA, FLORIDA By Suwan Shen December 2010 Chair: Zhong Ren Peng Coc hair: Paul Zwick Major : Urban and Regional Planning The National Research Council recently announced that climate change would greatly affect the performance of t ransportation and urban infrastructures. More frequent precipitations and storms will significantly increase the frequency and magnitude of flooding, which in turn threatens the reliabi lity of urban infrastructures. With i ntent to help decision makers bett er understand the impacts of increasing precipitation and consider adaptation strategies, this study develops a methodology using historical rainfall and discharge records to estimate the area and frequency of future riverine flood. Pensacola in Florida is used as a case study. Unlike previous impact studies, which are often general and qualitative, this paper explores the possibility to use downscaled global climate model output (in forms of daily rainfall realizations) to conduct quantitative impact analy sis that could assist transportation and land use planning practice at the local level. Areas and infrastructures within projected floodplain are identified, showing that in Pensacola single family developments are most
9 vulnerable to the increase of precip itation compared with other land use types. Future flood exceedance probability is calculated and compared, indicating that new design and planning standards should be developed, adapting to the future increase of precipitation. The estimated flood map cou ld be used as a guide for land use planning, and the calculated frequency increase could help to adjust the drainage design frequency standard. The multidisciplinary methodology in this study could be applied to other MPOs and region, providing a procedure for using hydrological data to estimate the impact s of changing precipitation on urban infrastructure.
10 CHAPTER 1 INTRODUCTION Recent studies and reports conclude that climate change is an undeniable challenge for the current transportation system (Natio nal Research Council, 2008; Meyer, 2008; Committee on Climate Change, 2008). According to the Transportation Research Board Special Report 290 (National Research Council, 2008), changes in precipitation, sea level, storms and heat waves will have significa nt impacts on the performance of the transportation and urban system. As a result, it is especially important that climate change should be taken into consideration in the long range transportation and land use planning process (National Research Council, 2008; ICF, 2008). Among the five types of climate extremes (i.e. very hot days and heat waves, increases in Arctic temperatures, rising sea levels, intense precipitation, and extreme hurricanes) through which climate change will have primary effects on tra nsportation (National Research Council, 2008), precipitation change will have much broader impacts than the others on transportation and other infrastructure systems. It is highly likely (with probability of greater than 90%) that most of the United States will suffer more intense and frequent precipitation events, which have an average annual increase rate of 6.1 percent per century with regional variations (National Research Council, 2008; IPCC, 2007; Environmental Protection Agency). Changing precipitati on levels could challenge the transportation system through increases in flooding levels and the moisture levels in soils, which will affect the stability of pavement subgrades, foundations, and drainage designs (National Research Council, 2008; Meyer, 200 8). The National Research Council (2008) uses the Great Flood of 1993, which occurred in
11 the Mississippi and Missouri River system, to illustrate the potential impacts of such increase. During this event, a catastrophic flood caused disruptions to the surf ace transportation system within 500 miles of the river system, affecting rail, truck, and marine traffic (National Research Council, 2008). However, regardless of the potential severity of this challenge, research questions such as the extent to what the changing precipitation would affect the magnitude and frequency of riverine flooding and, the surrounding development and transportation infrastructure remain unexplored at the moment, due to the lack of climate projections and future flooding predictions at the local level. In order to answer these research questions and help transportation planners better adapt to potential precipitation change, this thesis provides an impact assessment of the changing riverine flood frequencies caused by climate change on transportation infrastructure and land use, using Pensacola, Florida, as a case study. Unlike previous, qualitative assessment, this paper focuses on detailed, quantitative analysis at the local level with the intent to help local practitioners adapt to climate change. The paper proposes an innovative methodology through which the global climate change forecasting results could be converted to flood projection at the local level. In the research process, first, the current flood frequency information is extended regionally to estimate flood frequencies at any unmonitored locations of interest in the basin. Then, using historic daily rainfall and flood data, the antecedent precipitation conditions which would likely give rise to flood events are establishe d. Based on the extracted relationship and realization of daily rainfalls under future climate change scenario, and the impacts of future flooding on transportation and land use in the study
12 area is finally estimated. Due to the lack of daily rainfall real ization at the local level, historical rainfall records in other geographical area is used as the proxy of a future scenario. In this study t he historical rainfall pattern in Upper Shillong, India is selected as the proxy for future simulation in the study area as it s annual rainfall is similar to the projected future annual rainfall in the study area This is explained in detail later This paper includes five parts. The first part (Introduction) briefly explains the reasons for conducting this study. The second part (Literature Review) introduces background research, as well as summarizing what has been done and need to be done in regards to assessing the impact of changing precipitation caused by climate change on transportation and land use infrastructu re. The third part (Data and Methodology) introduces the data selection, analysis scenarios, methodologies and analysis procedures. Basic assumptions of the analysis are also encapsulated in this section. The fourth part (Results) summarizes the increase o f floodplain, inundation area by different land use type, and critical transportation infrastructures at risk. The final section (Conclusions and Discussions) illustrates how the methodology and results can help local decision makers make decisions to prot ect crucial infrastructures and development against changing precipitation and shed a light on what needs to be done in the future.
13 CHAPTER 2 LITERATURE REVIEW Debate regarding the inevitability of climate change has raised concerns in society as to the p otential impacts of climate change. Even if the greenhouse gas (GHG) emissions do not increase further, the accumulation of GHG emissions over past centuries is enough to cause climate changes that will quite likely increase both the frequency and intensit y of extreme weathers in the near future (National Research Council, 2008; Environmental Protection Agency, 2005). Consequently, it is important to understand how changing climate variables interact with the current transportation system, so as to minimize the potential negative impacts of climate change in the long range transportation planning process. Warm temperatures, temperature extremes, heavy precipitation, sea level rise, and more intense tropical storms have been identified as several important fa ctors through which climate change will challenge the transportation system (National Research Council, 2008). While warm temperatures and temperature extreme would challenge the material of our infrastructure, the sea level rise, precipitation and more in tense storms will impose more threats on how we plan and maintain our network with the increase of flood risk (Maantay and Maroko, 2009; Zimmerman, 2001). However, flood risk is not given proper attention in either the transportation or land use planning p rocess or previous climate change impact assessment studies, in part because of the difficulties of collecting sufficient, accurate data across disciplines to produce flood prediction and impact assessments at the local level. Considering the complexity of flood generation process, studies in this paper will be narrowed down to discuss the impacts of riverine flood generated by intense
14 precipitation on transportation and land use planning only. Even estimating the impacts of changing precipitation on trans portation infrastructure and land use requires intensive interdisciplinary cooperation among planners, transportation engineers, geographers, and climatologists. Specifically, in the cooperation process, the following research questions need to be answered : 1. How does traditional transportation land use planning incorporate flood information? Are the existing tools capable to provide guidance for future flood prediction/modeling? 2. What are the state of the art methods in estimating the impact of climate change? What are implications for practitioners? What are the existing research gaps? 3. Regarding precipitation, what is the state of the art methodology for future prediction? How could we use the results of precipitation prediction to estimate the chang e of riverine discharge? How could we predict the spatial (location and area) and temporal change (flood frequency) of flood plain based on the change of riverine discharge? 4. What are the implications of these changes to transportation and land use plann ing? The remaining of this chapter provides a literature review trying to answer questions one, two and three, which constitute the foundation on which the research method in chapter three is developed. Chapter three provides a detailed description of the methods and data used to overcome these research gaps, and the remaining chapters will throw some lights on question four specifically for the study area. 2.1 Traditional F lood P revention in T ransportation P lanning and E ngineering Flood protection was tra ditionally viewed as an engineering task through the design of flood defense systems with a specific exceedance probability (probability of at least one exceedance per year) (Apel et al 2004; Buchele et al 2006). For instance, the frequency, intensity, and duration of extreme precipitation of different return periods have been used by civil engineers to design transportation infrastructures (National
15 Research Council, 2008). The Project Development and Design Manual proposed by the United States Departm ent of Transportation (U S Department of Transportation, 2008) provides detailed standards, criteria and recommended methods of dealing with hydrology related issues in highway and bridge design. There are also specific requirements regarding roadway hydr aulics in respect to the culverts, ditches, pavement drainage, storm drains, energy dissipaters, and alternative pipe materials (U S Department of Transportation, 2008). However, these standards are based solely on a number of assumption, statistical stat ionarity and historical records of flood magnitude and runoff. If the meteorological (e.g. intensity and frequency of precipitation), hydrological (e.g. river profile), socio economic (e.g. agriculture activity) conditions change, the runoff and the magnit ude of the flood will change correspondingly, impairing the effectiveness of these standards. Consequently, more comprehensive design procedures have been proposed (Apel et al 2004). Extending the flood protection to planning, Flood Insurance Rate Maps ( FIRM), are used as the guideline for land use and transportation planning. Hazard zones are identified in FIRM, and developments in these areas are restricted. However, traditional planning does not place strong emphasis on flood protection (Zimmerman, 200 1). The FIRMs are widely criticized as being outdated, inaccurate, underestimating both the size and the depth of the flood (Burby, 2001). Additionally, many infrastructures are located within low lying areas that are vulnerable to flood even without clima te change to pursue operational benefits at the sacrifice of increasing flood risk (Zimmerman, 2001). This situation puts a large amount of infrastructure at high risk considering the increase of frequency and magnitude of flood caused by climate change (J acob et al 2001).
16 In the past decades, with increased attention in flood protection, risk based design has been identified as a complete approach to deal with flooding, taking both the probability of the flooding and the consequences of flooding into con sideration (Apel et al 2004). Usually, all flooding scenarios with associated probabilities and potential damages will be considered in the risk assessment, and a flood risk curve would be generated as the final product, demonstrating the distribution of flood damage in the study area within acceptable uncertainty boundaries (Apel et al 2004). The parameters and techniques considered in risk assessment have been categorized into four types, namely meteorological (e.g. rainfall, temperature), hydrologica l (e.g. river runoff), socio economic (e.g. land use change), and a combination of the three (Ologunorisa and Abawua, 2005). Using different parameters, researchers have explored different causality and consequences of flood risk. For example, Single et al (1990, cited by Ologunorisa and Abawua, 2005) use the total summer rainfall as an indicator to explore the relationship between flood/drought and the amount of regional rainfall. Apel et al (2004) develop a flood risk model to explore the probabilities of occurrence for flood events with different magnitudes and economic damages, using historical gauged hydrological data and stochastic modeling method. Their model take hydrological load, flood routing, levee failure and outflow, and damage estimation int o account in the process (Apel, 2004). Buchele et al (2006) incorporated risk assessment in their regionalization model to study flood with return period of 200 to 10000 years at small ungauged basin. However, these figures themselves are the indicators o f the low confidence of these estimations, especially considering their usefulness in planning area.
17 Both traditional planning and engineering approach and risk assessment suffer some limitations when applied to studies regarding climate change. Generally, most of these studies use the extreme value distribution of discharge as their start point, and only a few consider the effect of rainfall runoff. The lack of consideration for the causality relationship between rainfall and flood limits their ability to incorporate the change precipitation trend in the model process. Furthermore, considering it is difficult to get the probability of each climate change scenarios (especially at the downscaled regional level), it is hard to incorporate the risk assessment t echniques in climate change impact assessment. In recent years, many studies have been conducted to assess the impact of climate change on transportation infrastructures and land use, with the intent to provide decision makers with information to adapt to climate change. Most of these studies use scenario based methods rather than risk based methods, due to the aforementioned difficulties. 2.2 Climate C hange I mpact A ssessment and R esearch G aps Recently many studies estimated the impacts of climate change o n transportation infrastructure and land use, and draw valuable conclusions with respect to riverine 2005; ICF International, 2007; U.S. Climate Change Science Program, 2008; Peterson et al 2008; Jacob et al 2007; Burkett, 2002; Titus, 2002; James et al 2009). However, in transportation system, research in this area is still at an early stage. Mo st of these studies focus on large scale, qualitative level analysis (ICF International, 2007; U.S. Climate Change Science Program, 2008; Peterson et al 2008; Jacob et al 2007). In another word, those quantitative studies are not specific or accurate e nough to assist
18 local practitioners to develop local adaptation strategies to climate change (U.S. Climate Change Science Program, 2008; Suarez et al 2005). Detailed impact assessments of climate change on transportation and land use infrastructures are needed at the local level. The impact of precipitation varies at difference scales of study. From planning perspective, most of the research was focused on national scale and regional scale assessment, while from the engineering perspective the focus needs to be at local scale analysis. Koetse and Rietveld (2009) summarize existing transportation engineering research directions in estimating the impact of increased precipitation on transportation. Their review demonstrates that most of the existing studies focus on operation aspects, trying to analyze the impacts of precipitation on trip production, mode choice, and accident rate. There is a requirement for intermediate level research, which could provide local planners and project managers with estimation o f the direct impact of increasing precipitation on infrastructure disruption. The biggest challenge in estimating the impact of climate change on land use and transportation infrastructure, especially in term of changing precipitation, is the lack of accu rate flooding prediction at the local level. Local flood prediction requires both rainfall projections and the extraction of the regional relationship between flooding and precipitation at the local level. Both of these conditions are currently lacking. As a result, two research gaps must be overcome. First, the regional association between flooding and precipitation needs to be extracted from empirical data. Second, local rainfall simulations rather than global average projections should be used. Most prev ious studies use global average projections, rather than local projection to conduct impact
19 assessment (Jerry et al 2000; National Research Council, 2008; Peterson et al 2008). While the global average projection delivers a simple and general idea abou t climate change, it does not provide an accurate depiction of climate change at the local level, considering the differences in terrain, land use, coastal erosion, local subsidence and other factors. In order to bridge these two research gaps, the next se ction will provide a review of the existing data and methods regarding precipitation prediction and flood estimation. 2.3 Precipitation Prediction and Flood Estimation Flood generation is a complicated process, involving the interactions among meteorologi cal, hydrological, and socio economic factors (Ologunorisa and Abawua, 2005). Specifically, Durotoye (2000, cited by Ologunorisa and Abawua, 2005) elaborates that river peak discharge, heavy rainfall peaks, stream channel change, and tidal flooding are fou r major factors that cause inundation in deltaic plains. Furthermore, study shows that there is a positive coincidence between the changes of the frequency, magnitude, duration of heavy rainfall and the change of the frequency of extreme floods (McEwen, 19 99 cited by Ologunorisa and Abawua, 2005), which indicate that the change in precipitation will definitely influence the magnitude and recurrence probability of flood events. According to Fowler and Hennessy (1995), it is widely acknowledged that climate change will substantially change the frequency and magnitude of extreme daily precipitation. Milly et al (2002) further extend this conclusion to the substantial increase of the frequency of large floods in the twentieth century. All of these studies dem onstrate that precipitation is a factor that could not be overlooked in the flood estimation process.
20 There are a number of models and GIS techniques regarding flood estimation, but only a few take rainfall runoff simulations into account. Buchele et al (2006) constructs a model with consideration for the effect of precipitation and compared that with detailed rainfall runoff simulation. However, the limitation of their research is that their regression model uses mean annual precipitation depth, which is not a good indicator to reflect the change of the frequency of precipitation, while the rainfall runoff simulation uses intensive historical data and computation capabilities. Other models considering rainfall simulation use precipitation forecasting rath er than prediction with a focus on real time or near term forecasting (e.g. within 30 days), which is not suitable for long term planning activities (Todini, 1999; Frei et al 2000). Consequently, there is a demand for studies to bridge the gaps between t he large scale, qualitative analysis and real time simulations. Global climate models could provide a general simulation of precipitation frequency distribution changes (Simonovic et al 2003). General circulation models (GCMs) such as Australian Commonwe alth Scientific and Industrial Research Organization (CSIRO) four atmosphere layer GCM (CSIRO4), nine atmosphere layer GCM (CSIRO9), and the United kingdom Meteorological Office high resolution GCM (UKHI), could provide decades of daily precipitation data of each of the greenhouse gas emission conditions (Fowler and Hennessy, 1995). The simulation results are generally consistent, indicating that the change of precipitation is more significant in frequency and intensity than in total amount (Fowler and Henn essy, 1995). Researchers have interpreted the significance of these results in return periods to generate implications for flood planning (Fowler and Hennessy, 1995). However, because GCMs use coarse
21 grids, this limits the application of the model results to large scale flood protection system (Simonovic et al 2003). Currently, the production of downscaled simulation data is in progress. For example, California Reanalysis Downscaling at 10 km (CaRD10) has been produced to produce hourly, 10 km resolution downscaled analysis for California to support regional scale climate change applications (Kanamitsu and Kanamaru, 2007). In Southeastern United States, dynamical downscaling results at 20km resolution has been compared with the coarse resolution (2.5 lat itude/longitude) large scale atmospheric variables from the National Center for Environmental Prediction (NCEP)/DOE reanalysis (R2) for 16 summer seasons (1990 2005), indicating a better prediction over shorter time scales (Lim et al 2010). Although the downscaled modeling results for the study area are not available at present, it is said that downscaled 10km gridded rainfall for Southeast U S reanalysis will be available within several months. This study is designed with the intent to incorporate both the downscaled climate projection results and regional hydrological parameters (e.g. lag times, stream channel, etc.) in the flood impact assessment. To achieve this objective, the following steps are taken in the flood prediction process: 1. Estimation o f current discharge and rainfall data. Daily discharge and rainfall data are recorded at specific stations, therefore spatial interpolation is needed to estimate discharge and rainfall at other locations of potential interest. 2. Establish the relationship between rainfall pattern and riverine discharge values, which require an estimation of regional hydrologic lag time, 3. Based on precipitation simulation, estimate the magnitude of future extreme precipitation with a certain return period, and 4. Delinea te a hydrological data inventory (including e.g. river bank, stream line, land in GIS software.
22 Specifically, this study uses regional empirical data to 1)estimate flood freq uency at ungauged locations in the study area, 2)extracting the lag time between precipitation events and discharge fluctuation and 3)establishing antecedent precipitation conditions which would likely give rise to flood events. Estimates as to the change of future flood frequency based on this antecedent precipitation condition and daily rainfall realizations are made, and finally analyzed with respect to corresponding transportation and land use impacts.
23 CHAPTER 3 DATA AND METHODOLOGY Pensacola (Figure 3 1), the seat of Escambia County, Florida is selected as a case study, as it has experienced rapid population growth, intensive coastal development and exposure to heavy precipitation events. In addition, all major hydrological networks in Escambia County are natural flows without artificial intervention, minimizing the external control over riverine floods. Figure 3 1 City of Pensacola As mentioned above, current climate change projections can provide daily simulations for future rainfall patterns. I n order to estimate the flood frequency characteristics in Pensacola, the relationship between rainfall and riverine discharge must first be estimated. Then, realizations of future daily rainfalls may be passed through this relationship to empirically deri ve future flood frequencies. This paper uses rainfall data, discharge data, hydrological characteristics, and Geographical Information System (GIS) analysis to establish those relationships for the Escambia River System.
24 The results are put into hydrology models to delineate floodplain and to estimate future flood impact on transportation and land use. There are three major parts in the research process, namely flood prediction, terrain data processing (flood map generation), and transportation and land use impact analysis (Figure 3 2, below). The following paragraphs provide a detailed description of the data and methodology used in each part. Figure 3 2 Overall research framework 3.1 Flood Prediction This section uses the historical discharge data and observed rainfall data to establish the relationship between rainfall data and flood discharge records in the Escambia River Basin. Using the established relationship, the most likely future
25 discharge pattern is predicted based on future rainfall projectio ns. Daily discharge data and daily rainfall data are obtained from U.S. Geological Survey (USGS) and the Southeast Regional Climate Center. These include 10 discharge stations with more than 30 years records, and 55 rainfall stations within the basin area. 3.1.1 Spatial Interpolation Regression In Escambia County there are only six gauge stations with discharge records. The small sample size makes it difficult to interpolate discharge values at the points of interest in study area. This paper uses the regre ssion provided by National Streamflow Statistics (NSS), which is developed by USGS, to estimate the current discharge values at ungauged points in our study area (see Figure 3 2, below). The estimation is specifically for Northwest Florida region. (3 1) where QT is the discharge value for a recurrence interval of T years in cubic feet per second, DA is the contributing drainage area in square miles, and LK is the percent lakes and ponds (Bridges, 1982).
26 Using these equations, the current discharge values with different return periods are calculated for our study area, specifically Site A and Site B (Figure 3 3). The results are provided in Table 3 1. Figure 3 3 Ungauged points within the study area. Table 3 1 Current d ischarge with d i fferent r eturn p eriods in the s tudy a rea Site A Site B Return Period (Years) Discharge (cubic feet per second) Standard Error (%) Equivalent Years Discharge (cubic feet per second) Standard Error (%) Equivalent Years 2 502 44 3 813 44 3 5 998 46 4 1630 46 4 10 1420 49 5 2340 49 5 25 2050 55 6 3440 55 6 50 2630 59 6 4450 59 6 100 3230 65 6 5520 65 6 200 3890 70 6 6740 70 6 500 4890 77 6 8580 77 6 3.1.2 Establish the Relationship between Antecedent Precipitation and Floods Within the region, f lood sizes are strongly controlled by the amount of precipitation during certain preceding period. That period and the nature of the relationship need to
27 be determined. Historic daily rainfall records from Southeast Regional Climate Center and discharge va lues from USGS within the Escambia river system are first obtained to estimate the length of that significant preceding period. Table 3 2 provides the location information of the eight discharge stations with more than 30 years records and associated rainf all records (within Escambia River Basin). Dates and magnitudes of the ten largest annual floods observed in each of eight streamflow records in the region are then extracted. Next, partial sums of weighted basin inputs up to 14 days prior to the peak flow are computed and correlated to peak discharge. Plots of correlation versus length of partial sum indicate that ten days prior is sufficient regardless the size of basin area (Table 3 3, for detail please refer to Appendix B). The weight of each rainfall s tation is determined by the size of the Thiessen Polygon. Table 3 2 Location of the d ischarge s tations used in the s tudy Name Longitude Latitude E scambia Ri ver near C entury 87.2342 30.965 E scambia Ri ver n ear M olino 87.2667 30.66806 B ig C oldwater C re ek 86.9722 30.70833 M urder C reek near E vergreen AL 86.9867 31.41833 P erdido Ri ver at B arrineau P ark 87.4403 30.69028 S hoal R iver n ea r M ossy H ead 86.3069 30.79583 S hoal R iver near C restview, FL 86.5708 30.69722 Y ellow R iver at M illigan, FL 86.629 2 30.75278 Table 3 3 Preceding p eriod with m aximum c orrelation between a ccumulated r ainfall a mount and d ischarge v alues Name and s ize of the b asin a rea C ritical time (days) S hoal R iver n ea r M ossy H ead, FL (basin: 122 sq miles) 3 M urder C reek near E ver green AL (basin: 176 sq miles) 6 B ig C oldwater C reek (basin: 237 sq miles) 4 P erdido R iver at B arrineau park (basin: 378 sq miles) 3 S hoal R iver near C restview, FL (basin: 468 sq miles) 1 Y ellow R iver at M illigan, FL (basin: 635 sq miles) 4 E scambia R iver near C entury (basin: 3825 sq miles) 4 E scambia R iver n ea r M olino (basin: 4139 sq miles) 10
28 The discharge value is assumed to be primarily influenced by the magnitude of the rain within the influential period (10 days in Escambia Basin Area) and the size of the basin area. As a result, for each of the eight streamflow stations with discharge records in Escambia River Basin, the 10 largest floods at each station is converted to a standardized measure of runoff by dividing discharge values by basin are a to get units of specific discharge (cfs/sq miles) to discover the relationship among discharge, rainfall, and the size of basin area. The runoff is then plotted against the rainfall total of preceding 10 days (Figure 3 4). The plot reveals that the relat ionships between standard runoff and the amount of 10 days accumulated rainfall are generally linear. Therefore, by forcing the intercept to be zero to simplify the model, the slopes and R squared of the linear regression model for each of the basin area a re calculated (Table 3 4). Finally, logarithm model are used to regress the relationship among discharge values, 10 days accumulated rainfall and the size of the basin area. Figure 3 4 Relationship between standard runoff and 10 days accumulated rainfa ll
29 Table 3 4 Linear r egression m odel p arameters for e ach b asin Name of the Basin Linear Regression slope (intercept=0) R squared basin area (sq miles) log(Basin Area) E scambia R iver near C entury 2.11 0.94 3825 3.58 P erdido R iver at B arrineau P ark 4.19 0.93 378 2.58 E scambia R iver n ea r M olino 1.97 0.93 4139 3.62 M urder C reek near E vergreen AL 3.98 0.98 176 2.25 B ig C oldwater C reek 6.08 0.96 237 2.37 Y ellow Ri ver at M illigan 4.7 0.82 635 2.8 S hoal R iver near C restview 4.28 0.86 468 2.67 S hoal R iver n ea r M ossy H ead 4.43 0.84 122 2.09 Regression yields functions that permit the establishment of an anticipated peak discharge in response to antecedent precipitation totals. The regression result is as below. correspondingly ( 3 2) Where AR is the ten days accumulated rainfall in inches, discharge is the peak daily discharge in cubic feet per second, and area is the size of the basin area in square miles. The R square of the regression is 0.660, standard error is 0.852. Using this equation and the calculated current discharge values, the ten days accumulated rainfall that will give rise to 50 and 100 year flood estimates at Basins A and B are estimated (Table 3 5). According to the results generated thro ugh this process, 34.30 inches ten days accumulated rainfall will be enough to generate 50 year return period flood at Basin A, while 34.49 inches ten days accumulated rainfall will be enough to generate 50 year return period flood at Basin B.
30 3.1.3 Extra ct Future Rainfall Pattern The third step is to extract future extreme distribution of ten days accumulated rainfall using future daily rainfall simulation. Currently, the down scaling of global climate projection is still under the process, and no future daily rainfall simulation is available in our study area. To overcome this difficulty, historic precipitation series from other climate regions are used as the proxy for future rainfall simulation in study area. According to global projections, the precipi tation of Gulf area will increase by 20% 30% by 2060 (U.S. Climate Change Science Program, 2008; Louisiana Coastal Wetlands Planning, 2003). As a result, the selected region should have at least 20% 30% more precipitation than that of the current Pensacola The average annual rainfall values are calculated for stations in 2005 Global Historical Climatology Network (GHCN) database and compared. Finally, historic daily records in Upper Shillong in India, which has about 35% more annual rainfall than Pensacola are selected as the proxy of future simulation for Pensacola. Table 3 5 Estimated r ainfall s ufficient to g enerate f loods with v arying r eturn p eriods Site 50 year Flood 100 year Flood Discharge (cfs) Rainfall (inches) Discharge (cfs) Rainfall (inches) Site A 2630 34.30 3230 42.12 Site B 4450 34. 79 5520 4 3.16 To extract the rainfall pattern in Upper Shillong, historical daily rainfall record from the 2005 Global Historical Climatology Network (GHCN) database are used to generate the ten days accumul ated rainfall series. Then, after extracting the annual maximum ten days accumulated rainfall, generalized extreme value distribution is used to calculate the maximum annual ten days accumulated rainfall with different return period: ( 3 3)
31 Where and The Chi Squared statistic is 3.9063. The recorded historic annual maximum 10 days accumulated precipitation data can be seen in Figure 3 5. Figure 3 6 illus trates the generalized extreme value distribution of the annual maximum ten day accumulated precipitation in Upper Shillong. Figure 3 5 Annual m aximum t en d ay a ccumulated p recipitation Upper Shillong Figure 3 6 Generalized v alue e xtreme d istribution of t en d ay a ccumulated r ainfall in Upper Shillong
32 According to the General Extreme Value distribution, the accumulated ten days annual maximum rainfall with 50 year return period is 41.35 inch (1013.08 mm), and the accumulated ten days annual maximum rainf all with 100 year return period is 46.41 inch (1137.05mm). Using equation (2), the future discharge value at Site A and B c an then be calculated (Table 3 6 ). Figure 3 7 shows the change between current discharge values and future discharge values at Basins A and B, which indicating an increase of magnitude of floods with the same recurrent possibility, or in another word, the increase of recurrent possibility of floods with the same level of magnitude. Table 3 6 Future r ainfall and d ischarge v alues with d i fferent r eturn p eriods Site 50 year Flood 100 year Flood Discharge (cfs) Rainfall (inches) Discharge (cfs) Rainfall (inches) Site A 3171 41.35 3559 46.41 Site B 5288 41.35 5935 46.41 Figure 3 7 Change of discharge at study area 3.2 Terrain Data Pr ocessing (Flood Map Generation) In this step, the results from Step One, together with land use data, terrain data, river profiles, and road and bridge information are inputs to hydrology software, so that
33 flood maps under different scenarios can be deline ated. In the study, Digital Flood Insurance Rate Map (DFIRM) from Federal Emergency Management Agency (FEMA) Map Service Center is used to extract the river profile for the study area in ArcGIS and HEC GeoRAS. The profile includes river central line, strea m flow lines, and bank information. Digital Elevation Model data from USGS are used to extract 0.3m Triangular Irregular Network (TIN), which generates cross section information. Land use data (2004) from the Florida Geographical Data Library (FGDL) is use d to assign the Mannings N (roughness) values for each cross section. HEC RAS software is used to conduct steady flow analysis, which produces results that could be converted to flood map in ArcGIS (Figure 3 8 3 9 and Figure 3 10, 3 11 ), where the light b lue boundary represents the increase of future flood. 3.3 Impact Assessment Finally, using spatial analysis in ArcGIS, an assessment of the change in floodplain area identifies transportation infrastructure at risk, and estimates the vulnerability of diff erent land use types in respect to the increase of precipitation in the study area. The transit route data are obtained from Florida Transit Information System for the year 2008. The land Use data is 2004 land use and land cover data from FDGL, developed b Restoration. The highway data comes from Florida Department of Transportation Roads Characteristics inventory (RCI) dataset, including major roads in 2009 version. The impact assessment includes two perspectives, spatial and temporal. From the spatial perspective, the impact assessment focuses on the number of infrastructures and the amount of land will be inundated directly, while from the temporal perspective, the
34 impact assessment foc us on estimating the future flood frequency and comparing that with the existing design standards. The results are provided in the following chapter. Figure 3 8 Current and f uture 50 year return period f lood m aps
35 Figure 3 9 Current and f uture 50 year return period f lood m aps ( s mall s cale)
36 Figure 3 10 Current and f uture 100 year return period f lood m aps
37 Figure 3 11 Current and f uture 100 year return period f lood m aps ( s mall s cale)
38 CHAPTER 4 RESULTS 4.1 Impact Assessment The increase of precipita tion in the study area will increase the size of the floodplain, thereby putting increasing numbers of transportation and urban infrastructure at greater risk. This section provides a detailed impact assessment by quantifying the increase of floodplain, es timating different land use at risk, identifying transportation infrastructures that are most vulnerable to flood increases. First, an increase in the frequency and magnitude of precipitation will directly increase the size of the floodplain. Table 4 1 and Figure 4 1 show that in the study area, the size of 50 year floodplain will increase by five percent from 2.25 square miles to 2.37 square miles, and the size of 100 year floodplain will increase by 2.89 percent from 2.42 square miles to 2.49 square miles Table 4 1 Floodplain i ncreases Total Flood Area (Square Miles) Percentage of Increase Current Future 50 year flood 2. 25 2. 3 7 5 % 100 y ear flood 2. 42 2. 49 2. 89 % Figure 4 1 Change of f lood a rea Second, the increase of precipitation will put m ore urban land (Level 1 land use description, based on Florida Department of Transportation classification schema) at
39 risk. Figures 4 2 and 4 3 show the varying land use types vulnerable to 50 and 100 year floods, respectively. Results indicate that urban and built up environments are most susceptible to floods caused by increases in precipitation, as compared to other land use types. Third, the increase of precipitation will also affect transportation infrastructure (e.g. highway and transit), increasing the failure possibilities of transportation infrastructure. Table 4 2 summarizes the number of transportation infrastructures vulnerable (direct intersect with floodplain) to flooding in the study area. Because the location of most infrastructures happens not to be in the flood zone, the number of vulnerable transportation infrastructure is not as much affected by future 100 year flood as the land inundation. Table 4 2 Transportation i nfrastructures at r isk Major Highway Major Roads Affected Bus Routes Sum Interstate Other Urban Arterial Sum Roadway Urban Collect or Urban Local Urban Minor Arterial Current 50y flood 3 2 1 19 5 4 2 8 5 Future 50y flood 3 2 1 2 1 5 6 2 8 5 Current 100y flood 4 2 2 2 1 5 6 2 8 5 Future 100y flood 4 2 2 2 1 5 6 2 8 5 4. 2 Implications to U rban P lanning and T ransportation E ngineering The results of this study provide valuable implications for the development of climate change adaptation strategies in Pensacola. As Figure 4 2 and Figure 4 3 indicates, urban and built up env ironment will be affected most by the floods caused by
40 increasing precipitation. Within the urban and built up land use category, single family dwelling units would be the most vulnerable sector to an increase in flooding in terms of having the largest amo unt of area at risk (Table 4 3). As a result, in order to reduce the cost of maintenance and protection in future, adaptation strategies should focus on restricting single family residential development within the projected floodplain, and reallocating the current residential development in the following decades. Table 4 3 Inundation l and By d ifferent l and u se t ypes ( u rban and b uilt u p) Land Use Types (Urban and Built up) Area ( Acres ) Current 50y Flood Future 50y Flood Current 100y Flood Future 100y Floo d Commercial and Services 40.45 46.29 49.59 55.82 Extractive 22.68 23.00 23.00 23.00 Fixed Single Family Units 468.23 520.78 538.22 569.46 Inactive Land with Street Pattern without Structures 11.22 12.58 13.00 13.61 Institutional 28.99 31.81 31.94 33.27 Marinas and Fish Camps 2.23 2.23 2.23 2.23 Multiple Dwelling Units, Low Rise 25.04 26.27 27.02 28.26 Residential Mixed Units 7.00 7.08 7.25 7.25 Undeveloped Land within Urban Areas 0.07 0.07 0.25 0.25 Second, th e results of the study demonstrate that current design standards are not sufficient to prepare transportation infrastructures for future extreme events. Currently, according to the Project Development and Design Manual proposed by United States Department of Transportation (U.S. Department of Transportation, 2008), most transportation infrastructures within the study area are designed to convey runoff of existing 50 year return period or 25 year return period flood. However, for the same magnitude of flood, the return period will be become short in future. Table 4 4 shows that the current 50 year return period flood will become an 18 year return period flood
41 for Basin A and 20 year return period flood for Basin B, while the current 100 year return period flo od will become 56 return period flood for basin A and 64 return period flood for Basin B. Therefore, an update of design specifications should be developed in the near future. Table 4 4 Projected f lood r eturn p eriods Site 50 year Flood 100 year Flood Ra infall Future Cumulative Probability Future Return Period Rainfall Future Cumulative Probability Future Return Period inch mm inch m m Site A 34.3 0 840.35 0.9458 18 42.12 1031.94 0.9820 56 Site B 34. 79 8 52 36 0.94 95 20 4 3.16 105 7 42 0.98 4 4 64 The most important impact of flood caused by increased precipitation will be on cross section and drainage designs (Meyer, 2008). For example, in order to accommodate the increase of flooding, the slope of paved surface may need to be changed higher in or der to remove water to the side of the road as soon as possible (Meyer, 2008). In addition, drainage systems, open channels, pipes and culverts should be adjusted to provide larger capacity system for the increase of water discharge at the given design fre quency. The parameters of the design standards are determined by the selected design frequency, which are usually defined in drainage manual. For instance, cross drain h ydraulics, including culverts, bridge culverts, and bridges, for mainline interstate and high use culverts are 50 years return period (Florida Department of Transportation, 2010). The selection of the design frequency is determined based on the considerati on of damage to property, structure and roadway, traffic interruption, hazard to human life and damage to stream and floodplain environment (Virginia Department of Transportation, 2002). Therefore, in order to keep the risk of route interruptions at the cu rrent level, the design frequency should be increased to
42 accommodate the increase of precipitation. To determine how to increase the design frequency in our study area, probability distributions are used to fit the current discharges at Site A and Site B ( Figure 4 4, 4 5). Johnson Sb distribution (Equation 4 1 ) best fits the trend of current discharge at Site A, with The Kolmogorov Smirnov Test statistic is 0.3, and P value is 0.39012. The Inverse G aussian distribution (Equation 4 2 ) best fits the current discharge trend at site B, with The Kolmogorov Smirnov Test statistic is 0.292, and P value is 0.42008. Using th ese distributions, the equivalent current frequencies of future 50 year return period floods and 100 year return period floods are calculated (Table 4 5). The results indicate that to keep the damage the same as the current 50 year return period flood leve l, the design frequency in Basin A should be raised to 70 years, and to 85 years in Basin B. To keep the damage equivalent to the current 100 year return period flood level, the design frequency should be raised to 120 years in Basin A, and 145 years in Ba sin B. (4 1 ) ( 4 2 ) Table 4 5 Increase of f lood d esign f requency Site Future 50 year Flood Future 100 year Flood Discharge Current Cumulative Probability Current Return Period Di scharge Current Cumulative Probability Current Return Period A 3171 0.985752 70 3559 0.991766 120 B 5288 0.98 827 85 5935 0.99 31 1 1 45
43 Figure 4 2 Inundation l and b y d ifferent l and u se t ype ( 50 year f lood ) Figure 4 3 Inundation l and b y d ifferent l and u se t ype ( 100 year f lood )
44 Figure 4 4 Probability d istribution of c urrent d ischarge at s ite A Figure 4 5 Probability d istribution of c urrent d ischarge at s ite B
45 CHAPTER 5 CONCLUSIONS AND DISC USSIONS In conclusion, increase of precipitation will affect both transportation and land use development in City of Pensacola, Florida from two aspects. First, increase of precipitation will generate riverine flood that will inundate more urban and built up environment, especially single family dwelling units. Corresponding adaptation strategies could be developed based on this finding. For example, land use control policy could be implemented or flood proofing or residential house raising could be applied. The adaptation strategies could be either struc tural or non structural. The structural adaptations include building levees, dams, dikes, channel improvements and so on. The non structural strategies involve land use control, public participation, information sharing, open space and wetland preservation and community activities such as flood preparedness. The objectives of both structural and non structural adaptations are to reduce the amount of impervious surface, limit the development within floodplain, and preserve natural resources that reduce floo ding. However, compared with non structural strategies, structural adaptations are usually costly with certain environmental negative effects (e.g. disruption of hydrological cycles) and the potential to encourage development within the protected area. Fur thermore, considering the total increase of floodplain in the study area is not enormous, it is recommended that local planners and decision makers would better seek non structural mitigation strategies. The mitigation could be focused on three sides 1) na tural preservation 2) governmental regulation 3) public participation. First, through acquisition and preservation of the open space and wetland within the projected floodplain, the increase of impervious surface could be effectively controlled to reduce f urther increase of runoff.
46 Second, through zoning, comprehensive floodplain management plan, the total amount, density, and design standard (e.g. height of the lowest floor) of the housing units with in the projected floodplain could be regulated, reducing the total amount of potential risk. Finally, through public information sharing and education and modification of insurance regulations, homeowner involvement could be promoted, which plays a crucial and active role in flood mitigation. For undeveloped ar ea within the projected floodplain, mitigation strategy one is proposed, that is the government should acquire the open land and preserve them. For developed area, mitigation strategy two and three is recommended to increase public attention and reduce the flooding risk though change of standards. Furthermore, I recommend the local government consider the update of floodplain to change the flood insurance policy and the premium. Decision makers could also consider the use of incentives for relocation. Seco nd, increase of precipitation will increase the freque ncy of the flood, which require s an update of the drainage design frequency for transportation infrastructure. The analysis results could be used to update the flood zone map, or serve as a guideline fo r design standard update, which could strengthen the protection of human health and safety, reduce infrastructure disruption, and reduce stress. The innovative methodology used in the study successfully links climate change model outputs with the flood pro jection and detailed impact assessment at the local level. Although the inputs (daily realization) to the analysis remain a large amount of uncertainty, the study at least provides a way to describe the potential situation of climate change scenario at the local level. Furthermore, the study not only provides a quantification of the impacts of changing riverine flooding on transportation and urban
47 infrastructures, but also a foundation for further risk assessment and economic cost analysis. In the future, l and use models could be incorporated in this process to simulate the reallocation of development within the projected flood area, and the potential economic cost could be quantified. In addition, the study could be integrated with both land use model and t ransportation models to quantify the indirect cost on transportation system, taking the land use change and restriction of development in the flooding area into consideration. Finally, the multidisciplinary methodology in this study could be applied to oth er MPOs and region to help decision makers make more informed decisions about adaptation strategies for climate change. The limitation of the study is the requirement of large amount of precipitation and rainfall data and manual work. However, considering the hydrological pattern are similar within a relative large area (e.g. Escambia River Basin in this study), which often includes several states, researches could be achieved through cooperation between states. The establishment of such cooperation will p rovide a possibility for small cities with limited resources to prepare for climate change. Another limitation is the uncertainty of the data and modeling process. In addition to the uncertainty in climate change projection data and the variability in sta tistic model, incomplete knowledge of the complicated flood generation process produce additional epistemic uncertainties. For example, the model does not consider the probability of levee failure and other additional activity that could change the stream channel. Risk assessment may throw a light on these issues, but more accurate climate projection at the local level is the premise for such application.
48 Further improvements could be made towards the understanding of causality of riverine flooding, simula ting the effects of variable such as rainfall statistics with different duration classes (e.g. 0.5 h to 72 h). Impact assessment could be enriched by estimating the ground floor elevation of the building and stage damage functions by building types, usage, and structure.
49 APPENDIX PLOTS OF PEARSON C ORRELATION BETWEEN PARTIAL WEIG HTED RAINFALL SUM AND DISCHARGE VE RSUS LENGTH OF PARTI AL S UM FOR EACH OF THE BASIN AREA The following maps show the location of each discharge station and the valid rainfall stati on contributing to the streamflow in the basin area. The Digital Elevation Model data and discharge records come from U.S. Geological Survey (USGS) and the rainfall station comes from both USGS and Southeast regional climate center. The red rectangles show the location of the discharge stations, and dashed yellow rectangles show the location of rainfall stations. The diagrams demonstrate the Pearson product moment correlation coefficient and the square of the Pearson correlation coefficient Station 1 Esc ambia River near Century Size of the basin a rea: 3825sq miles P artial sum lengths with maximum correlation: 4 days L ocation of the station and valid rainfall record stations
50 P lots of P earson c orrelation versus length of parti al s um for each of the basin area Station 2 Perdido R iver at Barrineau P ark Size of the b asin a rea: 378 sq miles P artial sum lengths with maximum correlation: 3 days L ocation of the station and valid rainfall record stations
51 P lots of P earson c orrelation versus length of partial s um for each of the basin area Station 3 Escambia R iver n ea r Molino Size of the b asin a rea: 4139 sq miles P artial sum lengths with maximum correlation: 10 days L ocation of the station and valid r ainfall record stations
52 P lots of P earson c orrelation versus length of partial s um for each of the basin area Station 4 Murder Creek near Evergreen AL Size of the b asin a rea: 176 sq miles P artial sum lengths with maximum co rrelation: 6 days L ocation of the station and valid rainfall record stations
53 P lots of P earson c orrelation versus length of partial s um for each of the basin area Station 5 Big Coldwater Creek Size of the b asin a rea: 237 sq miles P artial sum lengths with maximum correlation: 4 days L ocation of the station and valid rainfall record stations
54 P lots of P earson c orrelation versus length of partial s um for each of the basin area Station 6 Yellow Rive r at Milligan, FLA Size of the b asin a rea: 635 sq miles P artial sum lengths with maximum correlation: 4 days L ocation of the station and valid rainfall record stations
55 P lots of P earson c orrelation versus length of partial s u m for each of the basin area Station 7 Shoal River near Crestview, FL Size of the b asin a rea: 468 sq miles P artial sum lengths with maximum correlation: 1 day L ocation of the station and valid rainfall record stations
56 P lots of P earson c orrelation versus length of partial s um for each of the basin area Station 8 Shoal River near Mossy Head Size of the b asin a rea: 122 sq miles P artial sum lengths with maximum correlation: 3 day L ocation of the station and valid rainfall reco rd stations
57 P lots of P earson c orrelation versus length of partial s um for each of the basin area
58 LIST OF REFERENCES Apel et al ( 2004 ). Flood risk assessment and associated uncertainty Natural Hazards and Earth System S ciences (2004). V ol. 4, pp. 295 308. Bridges ( 1982 ) Technique for estimating magnitude and frequency of floods on natural flow streams in Florid a. U.S. Geological Survey Water Resource s Investigations Report 82 4012. USGS, 1982 Buchele et al. ( 2006 ) F lood risk mapping: contributions towards an enhanced assessment of extreme events and associated risks. Natural Hazards and Earth System Sciences (2006) vol. 6, pp. 485 503. Burby R.J ( 2001 ) Insurance and Floodplain Management: The US Experience. Flood Environmental Hazards Vol. 3, pp. 111 122. Burkett ( 2002 ) Potential Impacts of Climate Change and Variability on Transportation in the Gulf Coast/Mississippi Delta Region I n The Potential Impacts of Climate Change on T ransportation Workshop Summary. U .S. Department of Transportation, Workshop, 2002 Drainage Manual. Flori da Department of Transportation, March 2010. H www.dot.state.fl.us/rddesign/dr/files/2010DrainageManu al.pdf H Accessed June 5th, 2010 Environ mental Protection Agency Precipitation and Storm Changes Environ mental Protection Agency. http://www.epa.gov/climatechange/science/recentpsc.h tml H Accessed June 5th, 2010 Environmental Protection Agency 2005. Emission Facts: Average Carbon Dioxide Emissions Resulting from Gasoline and Diesel Fuel EPA420 F 05 001, http://www.epa.gov/o taq/climate/420f05001.htm H Accessed June 1 st 2010. Fowler and Hennessy. ( 1995 ) Potential Impacts of Global Warming on the Frequency and Magnitude of Heavy Precipitation. Natural Hazards Vol. 11, pp 283 303. Frei et al. ( 2000 ) Climate dynamics and ext reme precipitation a nd flood events in Central Europe. Integrated Assessment. Vol. 1, pp. 281 299. ICF International ( 2007 ) The Potential Impacts of Global Sea Level Rise on Transportation Infrastructure. Phase 1 Final Report: the District of Columbia Maryland, North Carolina and Virginia H http://www.bv.transports.gouv.qc.ca/mono/0965210.pdf H Accessed June 20th, 2010.
59 ICF ( 2008 ) Integrating Climate Change into the Transportation P lanning Process Federal Highway Administration, 2008. H http://www.fhwa.dot.gov/hep/climatechange/climatechange.pdf H Accessed May 15th, 2010. IPCC ( 2007 ). Summary for Policymakers In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Mi ller (eds. )]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. H http://www.ipcc.ch/pdf/assessment report/ar4/wg1/ar4 wg1 spm.pdf H Accessed June 15t h, 2010. Jacob Edelblum, and Arnold ( 2001 ) Infrastructure. Chapter 4 in Climate Change and a Global City: The Potential Consequences of Climate Variability and Change Metro East Coast edited by C. Rosenzweig and W. D. Solecki. Columbia Earth Institute. New York, p p. 47 65. Jacob, Gornitz, and Rosenzweig ( 2007 ) Vulnerability of the New York City metropolitan area to coastal hazards, including sea level rise: Inferences for urban coastal risk management and adaptation policies. In Managing Coastal Vulne rability L. McFadden, R. Nicholls, and E. Penning Rowsell, Eds. Elsevier, pp. 139 156. James et al ( 2009 ) Coastal Sensitivity to Sea Level Rise: A F ocus on the Mid Atlantic Region. Synthesis and Assessment Product 4.1 Report by the U.S. Climate Change Science Program and the Subcommitte on Global Change Research. Jan, 2009. H http://www.climatescience.gov/Library/sap/sap4 1/final report/default.htm H Accessed May 1s t, 2010. Jerry et al ( 2000 ) U S Global Change Research Program: National assessments of the potential consequences of climate variability and chan ge. National Assessment Synthesis Team, U.S. Global Change Research Program, 2000. http://www.gcrio.org/NationalAssessment/overpdf/overview.html H Accessed June 30th, 2010. Kanamitsu and Kanamaru. ( 2007 ) Fifty Seven Year California Reanalysis Downscaling at 10 km (CaRD10). Part I: Sy stem Detail and Validation with Observations. Journal of Climate Vol. 20, Issue 22. pp.5553 5571. Koetse and Rietvel. ( 2009 ) The impact of climate change and weather on transport: an overview of empirical findings. Transportation Research Part D Vol. 1 4, pp 205 221. L im et al ( 2010 ) High resolution subtropical summer precipitation derived from dynamical downscaling of the NCEP/DOE reanalysis: how much small scale information is added by a regional model? Climate Dynamics A vailable at http://dx.doi.org/10.1007/s00382 010 0891 2
60 Louisiana Coastal Wetlands Planning ( 2003 ) Protection and Restoration News Feb, 2003. No. 22. H http://lacoast.gov/new/Data/WaterMarks/Watermarks 2003 02.pdf H Accessed June 1st, 2010. Maantay J and Maroko A ( 2009 ) Mapping Urban Risk: Flood Hazards, Race, & Environmental Justice in New York. Applied Geography Vol. 29, pp 111 124 Meyer, M.D ( 2008 ) Design Standards for U.S. Transportation Infrastructure: The Implications of Climate Change Transportation Research Board http://onlinepubs.trb.org/onlinepubs/sr/sr290Me yer.pdf H Accessed May 1st, 2010. Milly et al ( 2002 ) Increasing risk of great floods in a changing climate. Nature Vol. 415, pp. 514 517. National Research Council, 2008. Potential Impacts of Climate Change on U.S. Transportation Transportation Resear ch Board Special Report 290. Transportation Research Board, Washington, D.C. H http://onlinepubs.trb.org/onlinepubs/sr/sr290.pdf H Accessed May 5th, 2010. Ologunorisa and Abawua. ( 2005 ) Floo d Risk Assessment: A Review. Journal of Applied Sciences and Environmental Management. Vol. 9(1), pp 57 63 Peterson T. C., McGuirk M., Houston T. G., Horvitz A. H., and Wehner M. F. ( 2008 ) Climate Variability and Change with Implications for Transportatio n. http://onlinepubs.trb.org/onlinepubs/sr/sr290Many.pdf H Accessed June 15th, 2010. Project Development and Design Manual. U.S. Department of Transportation Federal Highway Administratio n, March 2008. http://flh.fhwa.dot.gov/resources/manuals/pddm/Chapter_07.pdf#7.1.1 H Accessed June 1 st 2010. Simonovic et al ( 2003 ) Methodology for Assessment of Climat e Change Impacts on Large Scale Flood Protection System. Journal of Water Resources Planning and Management. V ol. 129, No. 5, pp. 361 371. Suarez et al ( 2005 ) Impacts of Flooding and Climate Change on Urban Transportation: A Systemwide Performance Asses sment of the Boston Metro Area. In Transportation Research Part D Transport and Environment. Vol. 10, 2005. pp. 231 244. Titus ( 2002 ) Does Sea Level Rise Matter to Transportation Along the Atlantic Coast? In The Potential Impacts of Climate Change on Tra nsportation, Summary and Discussion Papers, Federal Research Partnership Workshop, Brookings Institution, Washington, D.C., Oct. 1 2, 2002. pp. 135 150. Todini. ( 1999 ) An operational decision support system for flood risk mapping, forecasting and manageme nt. Urban Water Vol.1, pp 131 143.
61 U.S. Climate Change Science Program ( 2008 ) Impacts of Climate Change and Variability on Transportation Systems and Infrastructure: Gulf Coast Study, Phase I U.S. Department of Transportation, Center for Climate Change and Environmental Forecasting Washington, D.C. H http://www.climatescience.gov/Library/sap/sap4 7/final report/ H Accessed June 1st, 2010. Virginia Department of Transportation Drainage Man ual Location and Design Division, Hydraulics Section, Virginia DOT, 2002 http://www.extranet.vdot.s tate.va.us/locdes/electronic%20pubs/2002%20Drai nage%20Manual/pdf/_Start%20VDOT%20Drainage%20Manual.pdf H Accessed June 10 th 2010. Zimmerman, R ( 2001 ) Global Climate Change and Transportation Infrastructure: Lessons from the New York Area, in The Potenti al Impacts of Climate Change on Transportation: Workshop Summary and Proceedings Washington, U.S. Department of Transportation pp. 91 101. 2002. H http://climate.dot.gov/documen ts/workshop1002/zimmermanrch.pdf H Accessed May 20th, 2010.
62 BIOGRAPHICAL SKETCH Suwan Shen is currently a student at the University of Florida, with a major in urban and regional planning and a special interest in transportation planning. She got her und ergraduate degree in Geographical Information System (GIS) from Southeast University, Nanjing, China. D am at the University of Florida s he presented her works in three international conferences: Transportation Research Board 2009 meeting, the Association of Collegiate S c hools of Planning 2010 meeting, and International Association for China Planning, and has won the Karen R. Polenske Best Student Award. She also volunteers herself as an intern working for the N orth Central Florida Regional Planning Council (NCFRPC) in spring 2010. Suwan is preparing herself for a career in transportation planning. Her future research plans include exploring travel behavior and GIS application in transportation fields.