Answers are grounded in this paper and cite their sources.
Reanalysis precipitation products provide spatially complete and temporally continuous estimates but remain subject to biases arising from model deficiencies and physical parameterizations. This study develops a rain-gauge fusion framework that combines the Box–Cox transformation with the local ensemble transform Kalman filter (LETKF) to improve hourly precipitation estimates from the CMA Regional Re-Analysis (CMA-RRA) over China. The Box–Cox transformation is introduced to alleviate the non-Gaussianity of hourly precipitation, making precipitation innovations more consistent with the Gaussian error assumptions underlying LETKF and thereby improving the quality of the resulting analysis increments. Results for 2021 show that both LETKF-based schemes improve the background precipitation, with the Box–Cox-based LETKF performing best. Relative to the original CMA-RRA precipitation, it reduces the annual mean bias by 82% and RMSE by 23%, and produces more consistent improvements across time, space, and rainfall intensity. These results highlight the importance of explicitly accounting for precipitation non-Gaussianity in LETKF-based reanalysis correction.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: