Comparing two high-resolution gauge-adjusted multisatellite rainfall products over India for the southwest monsoon period

摘要

Reliable high-resolution rainfall estimates are vital for hydrological and weather/climate-related applications and evaluation of high-resolution numerical model outputs. Multisatellite rainfall products provide immense opportunities to analyse rainfall at regular spatial and temporal scales, but suffer from large region- and season-dependent biases. Direct calibration or merging of ground-based observations with multisatellite rainfall estimates essentially provides more accurate rainfall estimates as it benefits from the relative merits of both the sources. In this study, two popular gauge-adjusted multisatellite rainfall products, the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA) research monitoring 3B42 version 7 (3B42V7) and Climate Prediction Center (CPC) Rainfall Estimation Algorithm version 2 (RFE2.0) are compared with the recently released and improved gridded India Meteorological Department (IMD) gauge-based rainfall estimates over India. The comparison is done for a 13year southwest monsoon season ranging from 2001 to 2013 at 0.25 degrees latitude/longitude resolution. A number of skill metrics such as mean, bias, co-efficient of variation, correlation co-efficient, anomaly correlation, pattern correlation and root-mean-square error (RMSE) are computed to assess the accuracy of both the merged satellite-gauge rainfall products. The prominent Indian monsoon rainfall features are well captured by both 3B42V7 and RFE2.0 products, in general. However, they overestimate mean rainfall at the all-India scale and the overestimation is comparatively larger for RFE2.0 than for 3B42V7. Even though the interannual variability of the Indian monsoon rainfall from both the gauge-adjusted multisatellite data sets is comparable with the gauge-based estimates for the study period, RFE2.0 overestimates light rainfall and underestimates heavy rainfall. Moreover, the comparison at sub-regional scales shows that 3B42V7 overestimates rainfall over eastern India and the foothills of the Himalayas and underestimates along the west coast and over the northeast, whereas RFE2.0 underestimates monsoon rainfall over the three sub-regions except over eastern India where it overestimates rainfall by about 25%. Both the gauge-adjusted multisatellite rainfall products show larger RMSE of the order of 100% along the west coast of India, which is a cause of concern. Overall analysis suggests that 3B42V7 is superior to RFE2.0 at synoptic scale over the Indian monsoon region.