Analyses Of High-resolution Merged Precipitation Products Over China Based On Temporal And Spatial Structure Score Indices

Based on the merged precipitation data from the National Meteorological Information Center, comparing with two sets of precipitation respectively retrieved from the satellites of CMORPH and TRMM as well as the observed data, the investigation of the applicability of the merged precipitation in China is performed. The spatial-temporal distributions of these precipitation data’s resolution are also analyzed. And the main factor which can affect the resolution of the merged data is mainly discussed. In this paper, besides of the traditional statistical indices such as mean error and correlation coefficient, three newest indices of spatial and temporal skill score and structure function are adopted. The main findings are as follows.(1) As for the spatial distribution, compared with two satellite precipitations, the error of merged precipitation generally decreases across China, and the regional differences greatly reduces, with the improved spatial consistency. The improvement is most obvious in the Jianghuai region. Except to the Qinghai-Tibet Plateau and the Northwest China, higher spatial consistencies between the emerged precipitation and observed precipitation appear in all the regions of China. Relative to the conventional assessment methods of the average deviation and root mean square errors, space skill score (SS) is more comprehensive and objective in the quantitative analysis, directly reflecting the spatial differences. Considering the temporal variation, the SS values for each month of the merged products are significantly greater than the CMORPH satellite and TRMM satellite data. And the increment of SS values after the emerging in warm season is less than in cold season. The three-year average SS value is improved from0to0.66. So spatial mean errors is significantly reduced after the precipitation emerged, and the simulation capability of to simulating the time-varying rainfall field is better in summer than winter.(2) Considering the time-varying, there is higher correlation between the emerged precipitation (satellite precipitation) and the observed precipitation. With respect to time correlation coefficient of satellite precipitation data which shows a decreasing trend from south to north in China, the high values of correlation coefficient between the emerged precipitation and observation cover most parts of China with the correlation value stably above0.9. Both summer and winter, the time average error of emerged precipitation generally reduces in each region with small spatial difference, greatly eliminated the north-south error gradient before the precipitation emerged and reducing the error magnitude. After the emerging of precipitation data, the temporal skill score (TS) is improved obviously, with the magnitude of improvement in winter better than in summer. Comparing TS scores of different regions, it is found that the simulation capabilities of the merged and satellite precipitation in Southeast China are generally superior to the Northwest. The TS values of emerged precipitation are closer to1than satellite data, and thus the time-varying accuracy of the emerged precipitation is higher.(3) The analysis to Structure function shows that in the Jianghuai region, South China, North China and Northeast China, with the increase of the distance between any two points within the grid area, the structure function curves of the observed precipitation and the emerged data are always very close to each other. So relative to CMORPH and TRMM satellite data, the quality of emerged products is better. In Southwest China with complex terrain, satellite data can not accurately reflect the actual precipitation; therefore observations play a more important role in this region. In northwest China and Qinghai-Tibet Plateau, the deviations between the emerged products and observed precipitation structure function curve are relatively larger.(4) The effect of site-grid density distribution to emerged precipitation is significant. The denser the site-grid within the region is, the closer the emerged precipitation in magnitude and spatial distribution is to the ground-based observations of precipitation. So the accuracy of precipitation in site-intensive region is higher than that of satellite data. The importance of satellite data to the emerged products in Sparse and dense region is different. In the eastern region of China, there enough observation stations, and thus the satellite data in dense regions contribute little to the emerged products. In the western region, if there is no site in the grid-box, the emerged data depends mainly on satellite data; otherwise the emerged data of the sparse region depends primarily on ground-based observation. The emerged precipitation data which effectively use of respective advantages of ground-based observations of precipitation and satellite precipitation are more reasonable in simulating of the precipitation value and distribution.(5) Compared with the satellite-derived data, the emerged data are able to describe better the average precipitation diurnal variation in China. As for the whole region of China, the emerging effect is better in summer than in winter. And both in summer and winter, the ability of emerged data to simulate the precipitation spatial structure in China overall is better in night than during the daytime. There are regional differences in diurnal variations of precipitation across China, and in simulation effects of the emerged data on the actual precipitation. In four regions of the northeast China, north China, Jianghuai, and South China during the summer and in winter in Jianghuai region and South China, the emerged precipitation can reflect the diurnal variation of average precipitation over these regions. In seven regions over China, the qualities of the emerged data are generally higher in summer than in winter. Compared to satellite data, emerged precipitation can reflect more accurately the persistent heavy rainfall, including the range and scope of the precipitation large value area, as well as the intensity, position, and movement path of the precipitation center. In the rainfall area, the temporal skill scores (TS) of emerged data as for the time evolution of the heavy precipitation process is generally higher than0.5, much higher than satellite data. So the emerged precipitation data have relatively strong ability to reproduce the actual precipitation process.