To improve the quantitative estimation of solid precipitation in snowy mountainous regions of Japan, we evaluated and corrected the systematic underestimation in the APHRO_JP gridded daily precipitation dataset. Two major sources of bias were addressed: wind-induced gauge undercatch and the poor representation of climatological precipitation due to the scarcity of high-elevation observations. To mitigate the latter, we refined the climatology using output from a high-resolution non-hydrostatic regional climate model (NHRCM) developed by the Japan Meteorological Agency and applied point-preserving ratio interpolation, in which the ratio of daily precipitation to the corresponding daily climatology is interpolated while preserving observed station values.
We conducted four precipitation analyses, combining the presence or absence of wind-effect correction with model-based (NHRCM) climatology. These were evaluated using two independent validation methods: (1) snow weight observations at seven SW-Net sites (elevation range: 255–1310 m) and (2) water balance analysis in four dam catchments (mean elevation range: 727–1073 m). On average, wind-effect correction and climatology refinement reduced RMSE by 9.6% and 8.4%, respectively. In the water balance analysis, climatology refinement led to an average precipitation increase of approximately 18%, while wind-effect correction contributed an additional 7%.
Both validation methods demonstrated that climatology refinement had a greater impact than wind correction, particularly in high-elevation regions. These results emphasize that model-based climatology, combined with point-preserving ratio interpolation, can substantially improve precipitation estimates in data-sparse, snow-covered mountainous areas. The corrected gridded datasets developed in this study will be publicly available through the APHRODITE project website.
View full abstract