Title:Rain Code : Forecasting Spatiotemporal Precipitation based on Multi-frames Feature using ConvLSTM

Abstract:Recently, flood damages has become social problem owing to unexperienced weather condition behind climate change. An initial response to heavy rain and high water condition are important for mitigating social loss and faster recovery. The spatiotemporal precipitation forecast may contribute the higher accuracy of dam inflow prediction forward more than six hours for flood damage mitigation. This paper proposes a rain code approach for spatiotemporal precipitation forecasting. We insights a novel rainy feature fusion that represents a temporal rainy process including several hourly sequences. We demonstrates various range of rain code studies based spatiotemporal precipitation forecasting using the ConvLSTM. We applied to a dam region within the Japanese rainy term hourly precipitation data, under 2006 to 2019 around 127 thousands hours, every year from May to October. We can use the raider analysis hourly data on the central broader region with 136x148 km square, based on new data fusion rain code with 3-hours frames range sequences. Finally we concludes knowledge and lessons such as remarkable capabilities and limitations for dam inflow prediction.