Application of statistical weight matrix method in estimation of regional rainfall

In terms of the Doppler radar rainfall estimation algorithms, such as the Z-Rrelation, the average calibration, the Kalman filter, the optimum interpolation, the variation, the optimum Kalman filter and the variation Kalman, the regional rainfall are estimated and compared with those interpolated from the observations in the automatic precipitation observatories. The results suggest that the performance of each algorithms mentioned above is not very satisfied. The statistical analyses are applied to the estimated results, and the statistical weight matrix approach is employed to improve the accuracy of the regional rainfall estimations. The results reveal that the precision of the regional rainfall estimation from the average calibration, the Kalman filter, the optimum interpolation, the variation, the optimum Kalman filter and the variation Kalman are evidently superior to those from the Z-Rrelation, and the regional rainfall estimation from the Z-Rrelation shows the evident underestimation. What's more, the results further show that the accuracy of the estimated regional rainfall derived from the statistical weight matrix approach by integrating all individual algorithm mentioned above is evidently higher than that obtained by any individual ones. The quantitative rainfall estimations with the statistical weight matrix approach are very close to the automatic rain-gage network observed either in the spatial distribution or in the location of the intense precipitation centers. The regional precipitation estimation of the statistic weight matrix approach can truly reflect the precipitation status over the ground surface and might be served as a promising conventional method for estimation of the regional rainfall for the studied region.