Evaluation of hydrologic models for Texas Flash Flood Alley

摘要

The 'flash flood alley'is an urban corridor between Dallas and San Antonio in central Texas, and is considered to possess the greatest risk of flash flooding in the United States due to its steep terrain, shallow soil, and abruptly high precipitationintensities. These high precipitation intensities are further driven by the sea-surface temperature and pressure anomalies defined by the variations in climate oscillations. This study first determines the potential links between the corridor's extreme streamflow (modeled using probability distributions) and Atlantic and Pacific Ocean based climate oscillations: (i) Atlantic Multidecadal Oscillation (AMO), (ii) North Atlantic Oscillation (NAO), (Hi) Pacific Decadal Oscillation (PDO), (iv) Pacific NorthAmerican Pattern (PNA), and (v) Southern Oscillation Index (SOI), using the Pearson correlation approach incorporating Leave-One-Out-Test (LOOT). Then it evaluates the performance ofprocess—based Soil Water Assessment Tool (SWAT) and data-driven Artificial Neural Network (ANN) models in different phases of the most correlated climate oscillation(s), based on both subjective and statistical goodness-of-fit tests, such as Nash-Sutcliffe efficiency (NSE), ratio of the root mean square error to the standard deviation of observed data (RSR), and percent bias (PBIAS). Since forecasting of extreme streamflows is important, results of this study will aid regional water boards in planning, designing, and managing hydrologic systems under climate change.