Assessing Urbanization Impacts on Longterm Rainfall Trends in Houston Parastou Hooshialsadat1, S.J. Burian2, and J.M. Shepherd3 1University of Arkansas, 2University of Utah, 3NASA Goddard Space Flight Center Research Objective To determine the effect of urbanization of the Houston metropolitan area on precipitation variability within the city compared to regions seasonally upwind and downwind. Analysis Components: 1. Downscaling analysis using TRMM-PR and rain gauge data (Shepherd and Burian, 2003) 2. Quantification of alterations to storm event characteristics and diurnal rainfall pattern (Burian et al., 2004a, 2004b) 3. Trend analyses of long-term rainfall records 4. Linked meteorological-hydrological modeling Houston is the 4th largest city in the U.S. (1.6 million) and covers an area of 937 km2; 10th largest CMSA (more than 4 million) Houston sits on the 5,000 km2 Gulf Coastal Plain with a high elevation of 27 meters above sea level Mean Monthly Air Temperature (C) for Houston (Data from Bush International Airport) 35 Air Temperature (C) 30 25 20 15 10 5 0 Jan Feb Mar Apr May June July Aug Sept Oct Nov Houston’s climate is subtropical humid with very hot and humid summers and mild winters Dec Houston Urbanization Urban growth characterized using a combination of multi-temporal population, multi-spectral, and cadastral data 4.5E+06 4.0E+06 Population 3.5E+06 Average of 43% population increase in Houston Metro per decade since 1900 3.0E+06 2.5E+06 2.0E+06 1.5E+06 1.0E+06 5.0E+05 0.0E+00 1880 1900 1920 1940 1960 1980 2000 2020 Year Approximately 40% increase of urban surfaces in Houston Metro between 1978 and 2000 (current city limit is ~1000 km2) Theoretical Coordinate System Defining Upwind and Downwind Regions Based on Mean Annual 700 hPa Steering Flow from 1979 to 1998 (following Shepherd et al. 2002) Abcissa is aligned along the 230º (southsouthwest) 700 hPa mean vector The wind rose below indicates that the prevailing near-surface flow is predominately southeasterly (e.g. sea breeze driven), however, for steering flow and upwinddownwind delineation, the 700 hPa surface is most critical. UCR – Upwind Control Region Orange ellipse has a 100 km horizontal diameter and 50 km vertical diameter and is centered on 29.77,95.38. The westernmost boundary of the UCR is 125 km from the orange ellipse and the easternmost boundary of the UIR is 100 km from the orange ellipse UIR- Urban Impacted Region Diurnal Rainfall Pattern Annual Warm Season **(blue gages have both urban and pre-urban data) This component of the study focused on the analysis of rain gage records with the necessary temporal resolution (hourly or less increments) for a preurban time period (1940-1958) and an urban time period (1984-1999) 30 35 Urban (1984-1999) Urban (1984-1999) 30 Pre-Urban (1940-1958) Pre-Urban (1940-1958) Percent of Rainfall Percent of Rainfall 25 25 20 20 15 15 10 10 5 5 0 0 0 400 800 1200 1600 Four Hours Ending (CST) 2000 2400 Average annual diurnal rainfall distributions at gage 4311 (UA) for the urban (1984-1999) and preurban (1940-1958) time periods 0 400 800 1200 1600 Four Hours Ending (CST) 2000 2400 Average warm season diurnal rainfall distribution at gage 4311 for the urban (1984-1999) and preurban (1940-1958) time periods The peak fraction of daily rainfall is more pronounced for the 12-16 and 16-20 4-hr time increments for the urban time period compared to the pre-urban time period; The warm season experiences a greater diurnal modification 25 30 Percent of Rainfall 20 25 Percent of Rainfall 15 10 Urban (1984-1999) 5 Pre-Urban (1940-1958) 20 15 10 Urban (1984-1999) 0 5 0 400 800 1200 1600 2000 Four Hours Ending (CST) Pre-Urban (1940-1958) 2400 Average annual diurnal rainfall distribution for the average of UCR gages 1671, 5193, 569, and 9364 0 0 400 800 1200 1600 Four Hours Ending (CST) 2000 2400 Average warm season diurnal rainfall distribution for the average of UCR gages 9364, 1671, and 3430 The change in diurnal rainfall distribution is visibly less in the UCR compared to the UA; The warm season has also experiences a greater diurnal modification Average warm season rainfall amounts (mm) in each time increment UA* UCR* 19401958 19841999 % Change 19401958 19841999 % Change 0-4 26 15 -42 28 29 +4 4-8 44 28 -36 54 50 -7 8-12 53 47 -11 75 54 -28 12-16 62 94 +52 50 65 +30 16-20 54 79 +46 32 35 +9 20-24 24 22 -8 28 25 -11 Total 263 285 +8 267 258 -3 * UA is the average of 4311 and 4309; UIR gages had insufficient data for warm season analysis; UCR is the average of 1671, 3430, 9364 Storm Event Characteristics This component of the study focused on the analysis of rain gage records with the necessary temporal resolution (hourly or less increments) for a pre-urban time period (1940-1958) and an urban time period (1984-1999) Storm Event Characteristics Average Warm Season Maximum 1-hr Rainfall Intensity Data Series Comparison Two Sample Test Mean Statistically Different ( = 0.05)? Pre-Urban UAR > UCR t Test No Wilcoxon No t Test Yes Wilcoxon Yes t Test Yes Wilcoxon Yes t Test No Wilcoxon No Post-Urban UAR > UCR UAR Post-Urban > Pre-Urban UCR Post-Urban > Pre-Urban Average maximum 1-hr rainfall intensity during the warm season has increased by 16% in the UAR compared to 4% in UCR Storm Event Characteristics Average Warm Season Number of Heavy Rainfall Events (> 25 mm) Data Series Comparison Two Sample Test Mean Statistically Different ( = 0.05)? Pre-Urban UAR > UCR t Test No Wilcoxon No t Test Yes Wilcoxon Yes t Test Yes Wilcoxon Yes t Test No Wilcoxon No Post-Urban UAR > UCR UAR Post-Urban > Pre-Urban UCR Post-Urban > Pre-Urban Average number of “heavy” rainstorms (> 25mm) during the warm season increased by 35% in the UAR compared to a 3% decrease in the UCR Trend Analysis The trend analysis used 10 rain gauges from the UA, and 20 each from the UIR and UCR. The gauges selected had the longest record lengths and the highest data coverage for the 50-year study period (1950-2000) Annual Warm Season UA UIR UCR UA UIR UCR Mean (mm) 1201 1210 993 326 289 232 St. Dev. (mm) 249 212 218 124 95 95 Cv 0.21 0.18 0.22 0.38 0.33 0.41 Skew 0.09 0.25 -0.01 0.82 0.40 0.92 Kurtosis -0.64 -0.23 -0.45 -0.40 -0.70 0.71 Average annual rainfall amount is greater in the UA and UIR than the UCR at the 0.95 confidence level Average warm season rainfall amount is greater in the UA than the UCR and UIR at the 0.95 confidence level There is no statistical difference between average annual rainfall in UA and UIR at the 0.95 level Average warm season rainfall amount is greater in the UIR than the UCR at the 0.95 confidence level Average annual trends 1800 Rainfall (mm) Linear: no trend exhibited (slope not significantly different from 0) for UCR; increasing trend for UA and UIR at 0.95 level Mann-Kendall: Annual rainfall is significantly increasing with time (90% confident) in each region. For UA and UIR, results are significant even for =0.05. 1400 1200 1000 Urban Area (UA) 800 mean = 1205 mm 600 400 1950 1960 1970 1980 1990 2000 Year 1800 1800 y = 4.5627x - 7801.4 R2 = 0.1019 1600 1600 y = 3.5935x - 6104.2 2 R = 0.0601 1400 Rainfall (mm) 1400 Rainfall (mm) y = 5.4931x - 9647.3 2 R = 0.107 1600 1200 1000 1200 1000 800 800 Urban Impacted Region (UIR) mean = 1210 mm 600 Upwind Control Region (UCR) 600 mean = 993 mm 400 1950 1960 1970 1980 Year 1990 2000 400 1950 1960 1970 1980 Year 1990 2000 350 Avg warm season trends 250 Rainfall (mm) Linear: no trend exhibited (slope not significantly different from 0) at 0.95 level Mann-Kendall: there is no evidence to conclude that the amount of warm season rainfall is increasing with time. 200 150 100 Urban Area (UA) 50 mean = 163 mm 0 1950 1960 1980 1990 2000 300 250 y = 0.4351x - 714.73 R2 = 0.0184 250 Rainfall (mm) y = 0.2726x - 422.33 R2 = 0.0073 200 150 100 50 200 150 100 50 Urban Impacted Region (UIR) Upwind Control Region (UCR) mean = 116 mm 0 1950 1970 Year 300 Rainfall (mm) y = 0.4657x - 756.62 R2 = 0.0124 300 1960 mean = 145 mm 0 1970 1980 Year 1990 2000 1950 1960 1970 1980 Year 1990 2000 Trend Analysis (cont’d)... The same battery of trend assessment tests were conducted for a difference statistic that represents the difference in average rainfall amount in a given year or warm season between the UA and UCR (R UA-UCR), UIR and UCR (R UIR-UCR), and the UA and UIR (R UA-UIR) Objective: Isolate the trend of differences between the three regions 500 Annual 400 300 y = 0.0748x - 139.49 R (mm) 200 y = 0.0382x - 66.821 100 y = 0.0366x - 72.672 0 1940 -100 -200 1950 1960 1970 1980 1990 2000 2010 UA-UCR (mm) UIR-UCR (mm) UA-UIR (mm) -300 Year Linear: increasing trend (slope > 0 at the 0.95 level) for UAUCR only Mann-Kendall: no significant trends found down to the 0.90 confidence level for all combinations; UA-UIR and UA-UCR differences increasing at =0.20 100 UA-UCR (mm) Warm Season UIR-UCR (mm) 75 UA-UIR (mm) R (mm) 50 25 Linear (UA-UCR (mm)) Linear (UIR-UCR (mm)) Linear (UA-UIR (mm)) y = 0.1287x - 222.86 y = 0.1083x - 194.93 y = 0.0204x - 27.929 0 1940 1950 1960 1970 1980 1990 2000 2010 -25 -50 Year Linear: increasing trend (slope > 0 at the 0.95 level) for UAUCR only Mann-Kendall: no significant trends found down to the 0.80 confidence level for all combinations Conclusions • Comparison of pre-urban and urban time periods suggests the diurnal rainfall distribution has been modified in urban areas beyond that responsible from natural background climate variability • Urbanization in Houston may be responsible for increased rainfall amounts during the midafternoon to late evening time periods in the urban area Conclusions • For recent period: annual and warm season diurnal rainfall patterns in the Houston UA and UIR display greater late afternoon and early evening rainfall amounts and occurrences compared to the UCR • This corroborates findings by Balling and Brazel (1987) for Phoenix and Huff and Vogel (1978) for St. Louis Conclusions Statistical comparison of average storm event characteristics from a pre-urban period and an urban time period indicates: 1. Average maximum 1-hr rainfall intensity during the warm season has increased in the UAR, but not in the UCR 2. Average number of “heavy” rainstorms (> 25mm) during the warm season has increased in the UAR, but decreased in the UCR Conclusions Annual rainfall amounts have had a strong increasing trend from 1950-2000 in the UA and UIR; and a weak trend in the UCR Warm season rainfall amounts have had very weak increasing trends from 1950-2000 Conclusions An increasing trend of R UA-UCR versus time and population is observed for annual and warm season rainfall in Houston No trend is observed for R UIR-UCR and R UA-UIR versus time and population Acknowledgements • This work has been supported by a NASA/ASEE Summer Faculty Fellowship (Burian), a NASA New Investigator Program (NIP) Grant (Shepherd), and a NASA Precipitation Measurement Mission award (PMM-0022-0069) (Shepherd, Menglin, and Burian) Questions??? Steve Burian burian@eng.utah.edu