Study On The Tropical Cyclone Rain Retrieval From FY-3B Microwave Imager(MWRI) Brightness Temperature

A tropical cyclone is a rapidly rotating vortex system,always accompanied with strong winds and heavy rains,which brings catastrophic impacts on human life and property during the landfall.The development of passive microwave sensors makes accurate monitoring of tropical cyclones become possible.Microwave Imager(MWRI),onboard the Fengyun-3B Satellite(FY-3B),provides all-weather and all-day measurements.To provide effective parameters for tropical cyclones,the microwave mechanism and surface precipitation retrieval algorithms for tropical cyclones are studied.In this study,two physical algorithms are used to retrieve the surface precipitation for tropical cyclones:1)the Wentz & Spencer(W/S)algorithm based on passive microwave radiation observations;2)the GPROF algorithm,which uses an a priori database based on active microwave radiation observations and forward radiation transfer model to retrieve rain rates.Firstly,the W/S algorithm is used to retrieve the precipitation for tropical cyclones.The W/S algorithm uses 18.7 GHz and 36.5 GHz vertically and horizontally polarized brightness temperatures from FY-3B MWRI to retrieve rain rate.This algorithm uses the direct and unique relationship between rain absorption coefficient and rain rate to retrieve the rain rate.To evaluate the performance of W/S algorithm,the FY-3B MWRI retrieved rain rates were compared with the National Snow and Ice Data Center(NSIDC)precipitation product,AMSR-E rainfall estimation,and active rainfall radar(Precipitation Rader,PR)precipitation product.The MWRI rainfall retrievals show good consistency with active and passive precipitation products,indicating that MWRI has a strong rainfall detection capability and can be used to detect the precipitation for tropical cyclones.To evaluate the spatial distribution characteristics of tropical cyclone precipitation,MWRI retrievals were compared with the TMI precipitation product provided by Remote Sensing Systems(RSS),which shows that MWRI W/S precipitation retrieval algorithm obtained reasonable precipitation intensity and spatial distribution for tropical cyclones.The W/S algorithm retrieves the rain rate based on the physical relationship between the rainwater absorption coefficient and rain rate,which,however,does not consider the scattering on atmospheric particles.W/S algorithm shoW/S obvious absorption saturation at high rainfall estimation and the limited retrieved rain rate is 27 mm/h.Therefore,the GPROF(Goddard Profiling algorithm)algorithm based on active/passive microwave observations is adapted to MWRI sensor to improve the accuracy of tropical cyclone rainfall retrievals.Secondly,the GPROF 2017 algorithm is a fully parametric approach based on the Bayesian scheme.FY-3B MWRI brightness temperature is applied to the GPROF 2017 algorithm to estimate rain rates and hydrometeor vertical profiles for tropical cyclones.An a priori database for MWRI GPROF algorithm is constructed from MWRI simulated brightness temperatures providing by the Eddington approximate radiative transfer model,the Microwave Imager(GMI)and dual-frequency precipitation radar(DPR)combined data,and ancillary data.The MWRI GPROF rainfall retrievals were compared with the DPR rainfall precipitation product,the GMI GPROF and GMI Hurricane GPROF(HGPROF)precipitation retrieval products.The performance of MWRI GPROF algorithm retrievals shoW/S good consistency with active and passive rainfall products on the surface rain rate and spatial distribution of tropical cyclone.In terms of global precipitation estimation,the mean rain rates at different distances from best-track locations are used to identify substantial differences between MWRI and GMI GPROF retrievals for 4 years(2014-2017).It indicates that the MWRI GPROF algorithm is applicable to global tropical cyclone rainfall retrieval studies.To verify the improvement of tropical cyclone rainfall retrieval accuracy by MWRI GPROF algorithm,the rainfall retrievals of MWRI GPROF algorithm and W/S algorithm were compared.The results show that the overall rainfall retrieval accuracy of the MWRI GPROF algorithm is better than the W/S algorithmFinally,the error analysis of MWRI GPROF algorithm on the tropical cyclone rainfall retrievals was considered,which is mainly divided into three parts.1)the error introduced by the inter-calibration of MWRI and GMI brightness temperatures.The MWRI GPROF rainfall retrievals considering inter-calibration have better statistical results than the results without inter-calibration;2)the error introduced by the size of the database.Comparing the statistics of different databases' retrievals,it shoW/S that the reduction of the database profiles has a greater impact on the high rain rate,however,less on the low rain rate;3)the error introduced by the Bayesian weight function.By controlling the channels' Equivalent Difference Temperature(NEDT)weights,the increase of the weight value optimizes the high rain rate retrievals,however,the improvement of the low rain rate is not obvious.Considering all the impacts,the method for solving the underestimation in MWRI GPROF high rain rate retrievals is to increase the number and proportion of high-rain rate profiles in the database entries.