17:15 〜 18:45
[MGI26-P04] Improving global precipitation estimates from rain gauge observations using local ensemble data assimilation
キーワード：Precipitation, Data Assimilation, Hybrid Error Covariance
It is essential to improve global precipitation estimates for better understanding on water-related disasters and water resources. This study proposes a new methodology to interpolate global precipitation fields from ground rain gauge observations using advanced ensemble data assimilation techniques. Here, we use the algorithm of the local ensemble transform Kalman filter (LETKF) in which the first guess and its error covariance are developed based on reanalyzed precipitation from the European Center for Medium-Range Forecasts (ERA5). We apply the hybrid approach to obtain the error covariance of the first guess, which is the combination of the climatological and flow-dependent error covariances. The former is constructed by using 20 years of historical data of ERA5, and the latter using the same data during the 5 days surrounding the targeted date as ensembles. The global rain gauge dataset provided by the National Oceanic and Atmospheric Administration Climate Prediction Center (NOAA CPC) are used for observation inputs in the LETKF algorithm.
Our estimates have better agreements with the independent reference data than the existing estimates issued by the NOAA CPC, as demonstrated by the verification against independent rain gauge observations in Asian countries. Furthermore, validations against the monthly precipitation dataset provided by the Global Precipitation Climatology Centre also indicate that our estimates are superior to the CPC product. The improvement was significant especially during rainy seasons in gauge-sparse regions such as Africa and the Himalayas. This improvement is presumably achieved because the proposed method constructs physically guaranteed background error variance and covariance owing to the usage of reanalysis data. In addition, we found that using the hybrid error covariance for estimation is beneficial than using the climatological static error covariance without any flow-dependency.
The results of this study will lead to a better use of ground rain gauge observations to achieve accurate global precipitation estimates.
Our estimates have better agreements with the independent reference data than the existing estimates issued by the NOAA CPC, as demonstrated by the verification against independent rain gauge observations in Asian countries. Furthermore, validations against the monthly precipitation dataset provided by the Global Precipitation Climatology Centre also indicate that our estimates are superior to the CPC product. The improvement was significant especially during rainy seasons in gauge-sparse regions such as Africa and the Himalayas. This improvement is presumably achieved because the proposed method constructs physically guaranteed background error variance and covariance owing to the usage of reanalysis data. In addition, we found that using the hybrid error covariance for estimation is beneficial than using the climatological static error covariance without any flow-dependency.
The results of this study will lead to a better use of ground rain gauge observations to achieve accurate global precipitation estimates.