|Subject||Development of Early Warning System for Flood using GPM Satellite and GIS System (Korean)|
|Author||Dr. Kyungwon Park||Date||2016.01.01|
This study develops a method for precipitation retrieval algorithm and a rain/no rain cloud classification system using Korea geostationary satellite images as well as GPM (Global Precipitation Mission), DPR (Dual Precipitation Radar), and GMI (GPM Microwave Imager) sensors. This method can be applied to rainfall observations in the Korea region. Co-locaed GPM CMB Level 2 precipitation with rain flags, precipitation type flags, land/ocean flags, convective/stratiform flags, cloud top height flags, and COMS geostationary satellite 5 channels are employed to develop real-time flood forecasting and monitoring systems for self-governing bodies. Shortwave infrared, water vapor, and infrared 1 channels are used to classify rainfall clouds from clear sky conditions. The new algorithm was used to perform validation when compared with old algorithm based on Tropical Rainfall Measuring Mission (TRMM) satellite for Busan city flood case at August 25, 2014. The results show that the new algorithm based on the GPM satellite is more accurate in estimating satellite rainfall. Rainfall cloud classification, which distinguishes rainfall clouds from non-rainfall clouds, is also improved. Integrated Multi-satellite Retrievals for GMP (IMERG) compared the early-run, late–run, and final–run products for the Busan City flood case using radar and ground station data. The early-run produced an overestimation when compared with the late-run product. The late-run is more accurate than the early–run when compared with the finalrun. The converted Python module was successful in improving the computing power for the parallel computing method.