An approach to reliability characterization of gsmap near-real-time precipitation product

Abstract

Reliability information of satellite precipitation products is required for various applications. This study de-scribes and evaluates a reliability flag of the Global Satellite apping of Precipitation Near-Real-Time precipitation product (GSaP_NRT). This flag was developed to characterize the reliability of GSaP_NRT data simply and qualitatively by considering its algorithm characteristics. The reliability at each pixel is represented by any one of ten levels (10 being the best and 1 the worst) by considering three major factors: 1) “surface type reli-ability”—which takes into account that estimation of rainfall using passive microwave sensors is better over the oceans than over land and coastal areas; 2) “low-temperature reliability”—which takes into account the lower reliability due to surface snow cover in low-temperature conditions; and 3) “oving Vector with Kalman Filter (VK) propagation reliability”—which means that the reliability gets worse with the increase in time since the last overpass of the passive microwave sensor. To evaluate the utility of the reliability flag, statistical indices are calculated for each reliability level using gauge-calibrated ground radar data around apan. t is found that the reliability flag represents the differences in GSaP accuracy: the accuracy worsens as the reliability decreases. The GSaP errors exhibit seasonal changes that are well represented by the ten levels of the reliability flag, indicating that the reliability flag can be used to catch seasonal variations in GSaP accuracy due to changes in environmental factors. This study also raises the possibility of improving the reliability flag by using information related to heavy orographic rainfall. t is shown how the error features of heavy orographic rainfall differ from those of the total rainfall, and it is suggested that heavy orographic rainfall information can be utilized to further improve the reliability flag.