Variational merged of hourly gauge-satellite precipitation in China: Preliminary results

Abstract

The article describes a variational scheme for the analysis of high-resolution hourly precipitation from China Meteorological Administration gauges and NOAA CMORPH satellite products in China and tests their impact on data-sparse regions and the heavy rainfall occurrences during the summer season (June–August 2009). In the variational scheme, a cost function is defined to measure the distance between analyzed precipitation field and observed rainfall quantity. A recursive filter is incorporated into the cost function which helps spread the observations to nearby grid points. Then a quasi-Newton method is used to solve the optimal estimation problem by minimizing the cost function. The adjoint technique is used to derive the gradient of cost function with respect to analysis precipitation. A series of experiments are performed to intercompare the variational analysis with the original CMORPH satellite products (CMP) and the bias-adjusted satellite products (Adj-CMP) against the observations. The best overall performance is from the variational analysis especially rainfall intensity by more than 10 mm h-1 with a prevailing mean relative spatial bias nearly reduction zero, and the correlation coefficient is almost around 0.5 in convection active areas. Ground cross-validation experiments in which each affected station is withdrawn at once indicated that the variational analysis can particularly be beneficial and subsequent investigation of heavy rainfall events. It also reveals that the precipitation analysis field has the ability to improve the accuracy of rainfall estimation and capture the spatial precipitation pattern agreements in relatively data-sparse regions.