A Research On Optimization-based Rainfall Parameter Retrieval Using Polarimetric Radar Data

Polarimetric radars offer the information on the size,shape,number concentra-tion of the hydrometeors in weather systems.The rainfall parameter retrieval facilitate the surveillance of the short-time severe precipitation and the study of the precipitation microphysics.Improving the accuracy of rainfall parameter retrieval can improve the ability of disaster prevention and mitigation and help to understand the microphysical structure and evolution of the weather systems in our country.The accuracy of rain-fall parameter retrieval using polarimetric radar data is mainly limited by two aspects,namely,the uncertainty of the relation between polarimetric parameters and rainfall caused by the variability of raindrop size distribution(DSD)and how the observation errors are handled in the retrieval algorithms.The study is mainly focused on these two aspects.First,the long-term observations of 2D video disdrometers are used to study the characteristics of DSDs in Eastern and Southern China and establish the models of at-tenuation correction and rainfall parameter retrieval in these two regions.As for the radar data quality control,the calibration of differential reflectivity(ZDR),the estima-tion of system phase term in total differential phase shift(?DP),the attenuation correc-tion,and the identification and removal of non-meteorological echo are studied in this study.And the applications to the C and S-band polarimetric radar measurements show that the quality of polarimetric radar data can be improved effectively.As for quanti-tative precipitation estimation(QPE),the study is focused on two aspects,namely,the retrival with empirical estimators and the optimized retrival based on the variational theory.The accuray of the estimation of the specific differential propagation phase shift(KDP)is the key to the QPE with the empirical relation between rainfall rate(R)and KDP.The DSD information in the horizontal reflectivity factor(ZH)and ZDR mea-surements are utilized to construct a more accurate physical constraints based on the self-consistency of polarimetric radar variables.Based on which,a hybrid method is developed.When comparing the KDP estimation for the measurements of C and S-band polarimetric radars,the hybrid method shows advantages over the LP method and the LSF method.As compared to the linear programming method,the mean correla-tion coefficient between the C-band radar-based hourly rainfall estimates and the gauge measurement increased from 0.90 to 0.96 and the mean relative error decreases from 0.80 to 0.57.Furthermore,the error characteristics of R(ZH)due to the DSD variability and mea-surement errors are determined using the long-term DSD observations.An variaitonal rainfall retrieval method is proposed,in which the R-KDP relation and R(ZH)act as the observation operator and the a priori,repectively.As compared to the method proposed by Hogan(2007),the correlation coefficient between the S-band radar-based hourly rainfall estimates and the gauge measurements increases from 0.90 to 0.92 and root-mean-squared error decreases from 2.41 mm to 2.10 mm in Southern China.At last,the C-G models in Eastern and Southern China are estabilished.The one-dimentional variational analysis is utilized for simultaneous optimal DSD retrieval and attenuation correction from ZH,ZDR,and ?DP.The rainfall parameters are hereby derived from the optimal estimated DSD.In the observation operator,the specific(differential)attenu-ation(AH and ADP)are directly calculated from DSD.It helps to avoid the impact of DSD variability on the attenuation correction,which is the main weakness of the lin-ear corretion method and ZPHI method.The method is applied to the data collected by NJU CPOL in a Meiyu event in Jianghuai area and the data collected by Zhuhai SPOL in a heavy rain event in Southern China.The estimated microphysical parame-ters clearly indicate the difference of the DSD properities of the convections in these two kinds of rainfall events.In the heavy rain event in Southern China,the rainfall rates calculated from the optimally estimated DSD are also used for QPE.Both the total ac-cumulated rainfall and the hourly accumulated rainfall show good agreements with the gauge measurements.This study estabilishes the connections between the polarimetric radar variables and rainfall parameters using long-term DSD observations in Eastern and Southern China.Meanwhile,the DSD variability and measurement errors are quantitatively con-sidered in the optimization-based algorithms and the information in the polarimetric radar measurements is synthetically utilized.The quatlity of the polarimetric radar data and the accuracy of the retrieved rainfall parameters are improved using these algo-rithms.