Improvement of Satellite-based Precipitation Estimates over China Based on Probability Density Function Matching Method

In the evaluation and adjustment of satellite-based precipitation estimates, the gauge-based precipitation data is usually taken as the objective criteria. Assuming gauge-based analysis at grid boxes with station reports as the true value, the satellite precipitation data are corrected after adjusting probability density function (PDF), which makes the PDF distribution of the corrected measurements the same as that of the station observation. One advantage of the PDF method is that it can remove the range dependent bias effectively, which is also the main cause for its recent popularity in correcting the error of satellite-based precipitation estimations.Aiming to investigate the applicability of the PDF method, and then to adjust the systematic bias of the high resolution satellite-based precipitation estimations over China, the daily satellite precipitation data with the resolution of 0.25° by 0.25° and hourly data of 0.1° by 0.1°from CMORPH (Climate Prediction Center Morphing Technique) are adjusted, based on grid precipitation data interpolating stations collected and quality controlled by NMIC (National Meteorological Information Center) of China Meteorological Administration. Although CMORPH data has good performance on the space structure of rainfall over China in summer time, it has obvious systematic errors. CMORPH data underestimate large precipitation values while overrate small ones.After analyzing the bias characteristic of CMORPH precipitation data, different matching schemes are designed by adjusting PDF matching samples of spatial and temporal scale separately. The generalized cross-validation (GCV) statistical tests are used in evaluating the quality of correcting data. The evaluation results suggest that the PDF distribution and precipitation values of corrected precipitation products are close to those of the gauge-based precipitation. Both adjusted daily and hourly satellite precipitation data over China get a less systematic bias than the original ones. Even in the area of sparse observations, such as the Western China, the improvement is also remarkable. The adjusted rainfall data over the Western China not only maintain the basic spatial construction of original satellite products, but also improves their quantity value through closing them to the gauge observations. So the research demonstrates that the PDF method is an effective way in correcting systematic bias of satellite-based precipitation estimates over China.