Evaluation of satellite-retrieved extreme precipitation rates across the central United States

Abstract

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116, D02115, doi:10.1029/2010JD014741, 2011 Evaluation of satellite‐retrieved extreme precipitation rates across the central United States A. AghaKouchak, 1 A. Behrangi, 2 S. Sorooshian, 1 K. Hsu, 1 and E. Amitai 3,4 Received 10 July 2010; revised 10 September 2010; accepted 27 October 2010; published 26 January 2011. [ 1 ] Water resources management, forecasting, and decision making require reliable estimates of precipitation. Extreme precipitation events are of particular importance because of their severe impact on the economy, the environment, and the society. In recent years, the emergence of various satellite‐retrieved precipitation products with high spatial resolutions and global coverage have resulted in new sources of uninterrupted precipitation estimates. However, satellite‐based estimates are not well integrated into operational and decision‐making applications because of a lack of information regarding the associated uncertainties and reliability of these products. In this study, four satellite‐ derived precipitation products (CMORPH, PERSIANN, TMPA‐RT, and TMPA‐V6) are evaluated with respect to their performance in capturing precipitation extremes. The Stage IV (radar‐based, gauge‐adjusted) precipitation estimates are used as reference data. The results show that with respect to the probability of detecting extremes and the volume of correctly identified precipitation, CMORPH and PERSIANN data sets lead to better estimates. However, their false alarm ratio and volume are higher than those of TMPA‐RT and TMPA‐V6. Overall, no single precipitation product can be considered ideal for detecting extreme events. In fact, all precipitation products tend to miss a significant volume of rainfall. With respect to verification metrics used in this study, the performance of all satellite products tended to worsen as the choice of extreme precipitation threshold increased. The analyses suggest that extensive efforts are necessary to develop algorithms that can capture extremes more reliably. Citation: AghaKouchak, A., A. Behrangi, S. Sorooshian, K. Hsu, and E. Amitai (2011), Evaluation of satellite‐retrieved extreme precipitation rates across the central United States, J. Geophys. Res., 116, D02115, doi:10.1029/2010JD014741. 1. Introduction [ 2 ] Precipitation plays a significant role in weather research, monitoring, and predictions. Improving our understanding of weather and climate, along with the development of reliable and uninterrupted measurements, are essential for proper assessment of weather conditions. Currently, in situ and radar‐based precipitation data are the major input for streamflow forecasts, flash flood warnings, and weather watches across the United States. While some regions have long‐term historical in situ precipitation measurements, poor spatial sampling makes the data inadequate to support monitoring, detection, and forecast studies. On the other hand, in most parts of the globe (except in a few developed countries), radar installations for precipitation measurements Department of Civil and Environmental Engineering, University of California, Irvine, California, USA. Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA. NASA Goddard Space Flight Center, Greenbelt, Maryland, USA. School of Earth and Environmental Sciences, Chapman University, Orange, California, USA. Copyright 2011 by the American Geophysical Union. 0148‐0227/11/2010JD014741 are not available. In the United States, with one of the most sophisticated radar measurement networks in the world, regions with extensive topographic relief (e.g., the western and southwestern United States) suffer from poor or non- existent radar coverage [Maddox et al., 2002]. In fact, Maddox et al. [2002] showed that at lower levels (e.g., 1000 m above ground level), which are closer estimates to ground‐level precipitation, the radar coverage area is sub- stantially smaller than at higher levels (e.g., 3000 m above ground level). [ 3 ] Clearly, the lack or absence of ground‐based precipi- tation networks hampers the development and use of flood and drought warning models, hydrological models, and extreme weather monitoring and decision‐making systems. Therefore, there exists the need to achieve alternative esti- mates of precipitation with sufficient sampling density, reliability, and accuracy to enable utilization of data for operational applications. Satellite‐derived precipitation esti- mates have the potential to improve precipitation observation at a global scale. In recent years, the National Aeronautics and Space Administration (NASA), National Oceanic and Atmospheric Administration (NOAA), and many other inter- national sponsored satellite missions have led to an increase in available precipitation data. These remotely sensed data have several advantages over in situ measurements, including D02115 1 of 11