We held the 61st Data Assimilation Seminar at RIKEN Center for Computational Science (R-CCS) on 21 August.
The talk was given by Dr. Shigenori Otsuka from the Data Assimilation team here at RIKEN. He presented his research using two optical-flow based nowcasting systems for improving short-range precipitation prediction, 1) Phased Array Weather Radar (PAWR) 3D nowcasting system and 2) Global Satellite Mapping of Precipitation (GSMaP) RIKEN Nowcast (RNC) system. Given the limitations of using optical-flow systems for weather prediction, one being that convective weather systems can evolve very rapidly and non-linearly, his research has steered towards the use of deep learning techniques, where he has applied the Convolutional Long Short-Term Memory (Conv-LSTM) algorithm to the PAWR data. Preliminary results of real-time experiments with Conv-LSTM showed improvements in short-range forecasts.
We would like to thank Dr Otsuka for his interesting talk and wish him well with his continuing research using nowcasting systems.
For more information about this and all our DA seminars at RIKEN R-CCS, please see the DA seminar page: