First Evaluation of the Utility of GPM Precipitation in Global Flood Monitoring

The Global Flood Monitoring System (GFMS) has been developed and used to provide real-time flood detection and streamflow estimates over the last few years with significant success shown by validation against global flood event data sets and observed streamflow variations (Wu et al., 2014). It has become a tool for various national and international organizations to appraise flood conditions in various areas, including where rainfall and hydrology information is limited. The GFMS has been using the TRMM Multi-satellite Precipitation Analysis (TMPA) as its main rainfall input. Now, with the advent of the Global Precipitation Measurement (GPM) mission there is an opportunity to significantly improve global flood monitoring and forecasting. GPM’s Integrated Multi-satellitE Retrievals for GPM (IMERG) multi-satellite product is designed to take advantage of various technical advances in the field and combine that with an efficient processing system producing “early” (4 hrs) and “late” (12 hrs) products for operational use.

On the road to replacing TMPA with the IMERG in the operational global flood simulation, the DRIVE model based GFMS parallel systems were run for periods to understand the impact of the new type of data on the streamflow and flood estimates. The GFMS has initially run at its primary resolution of 1/8th degree latitude/longitude with both data sets to isolate the impact of the rain information change. Specifically, this study is focused on (1) understanding the difference between the new IMERG products and other existing satellite precipitation products, e.g., TMPA, CMORPH, and ground observations; (2) addressing the challenge in the usage of the IMERG for flood monitoring through hydrologic models, given that only a short period of precipitation data record has been accumulated since the lunch of GPM in 2014; and (3) comparing the statistics of flood simulation based on the DRIVE model with IMERG, TMPA, CMORPH etc. as precipitation inputs respectively. Derivation of a global threshold map is a necessary step to define flood events out of modelling results, which requires a relatively longer historic information. A set of sensitivity tests are conducted by adjusting IMERG’s light, moderate, heavy rain to existing precipitation products with long-term records separately, to optimize the strategy of PDF matching. Other aspects are also examined, including higher latitude events, where GPM precipitation algorithms should also provide improvements. This study provides a first evaluating the utility of the new IMERG products in flood monitoring through hydrologic modeling at a global scale. It helps focus full implementation of the IMERG into GFMS and the retrospective calculations to be done for the full TRMM/GPM era.