Validation studies of precipitation estimates from different satellite sensors over hungary - Analysis of new satellite-derived rain rate products for hydrological purposes

摘要

Many methods have been worked out to estimate precipitation rainfall from meteorological satellite radiances sensed by microwave (on-board low-orbiting satellites) and infrared (on-board geostationary satellites) sensors. Validation of such estimated precipitation products have become more and more important.This paper describes the validation study limited to the geographical area of Hungary of three satellite-based rain estimates for hydrology purposes. These estimates were recently developed in the frame of the Hydrology SAF project, launched by EUMETSAT (European Organization for the Exploitation of Meteorological Satellites) in 2005. While two of these products were estimated using microwave-based method, one of them was estimated using a combination of infrared (IR) and microwave (MW) radiances.The validation studies carried out at the Hungarian Meteorological Service covered statistical analyses of instantaneous values over monthly periods and case-by-case analyses. Statistics were evaluated for every month, and focus was given to the differences between summer and winter to depict the seasonal features of the rainfall estimates. All three products gave best results in summer months (probability of detection was between 0.5 and 0.7 compared to 0.1 and 0.2 in winter; and false alarm rate was 0.3-0.6 in summer compared to 0.8-0.9 in winter), due to the much easier detection of deep convective clouds by satellites. Correlation was between 0.2 and 0.4 in summer months for every product, whereas in the winter it was under 0.1. In case studies, the location of convective cells formed in the summer was reflected well by MW observation; the combined products showed overestimated area of low precipitation.Significant underestimation of heavy rainfall (mean error of -15 to 20. mm/h) was found in the case of the infrared-microwave (IR-MW) mixed product. According to case studies, the MW-based retrievals overestimated high precipitation intensities of the convective cells. However, in the monthly statistics the mean error was negative (-5 to 10. mm/h) which demonstrated that the overestimation was not systematic.Winter results revealed that light rainfall had low probability of detection both by MW and IR-MW combined measurements. However, the case study presented for a winter day showed for all the three products well detection of liquid precipitation. Overall results exposed more reliable detection of convective than stratiform precipitation.