Research On The Estimation Method Of Tropical Cyclone Intensity In The Western North Pacific Using Deviation Angle Variance Technique

Based on the FY series infrared satellite data,we analyzed the characteristics and estimation methods of Western North Pacific?WNP?tropical cyclones from the perspective of the cloud axisymmetry distribution using deviation angle variance technique?DAV-T?.DAV-T is an objective observation technique that enables automated monitoring of tropical cyclones,reducing the reliance on human experience in tropical cyclone forecast and early warning processes.This method uses the deviation angle variance value?DAV?to quantify the degree of cloud axisymmetry.The smaller the DAV at a certain point,the higher the degree of the cloud organization relative to the nearby cloud cluster at that point.Studies have shown that the axisymmetry of the tropical cyclone system can be characterized by the minimum DAV,which is the minimum value in DAV map?MMV?.This paper examines the changes of tropical cyclone structure and intensity during its whole lifecycle.It is found that there is a significant difference in the axisymmetry of deep convective clouds at different intensity stages of tropical cyclones.The MMV,which can represent the system axisymmetry of tropical cyclones,will decrease with the increase of tropical cyclone intensity,and the edge of the low-DAV area will be more regular.The correlation analysis shows that there is a good correspondence between the MMV and tropical cyclone intensity.At the same time,the MMV positions are basically the same as the tropical cyclone centers that recorded by the CMA-BestTrack dataset.The relative distances between the two are small except in the extinction and dissipative periods.According to the analysis of Lionrock and Meranti,it is found that when the tropical cyclone system is closed,the MMV position can indicate the tropical cyclone center better.However,when the system has obvious non-closed features,the MMV position always tends to the most convective area and deviate from the actual tropical cyclone circulation center.Thus,the moments with large relative distances often occurs during the formation and dissipation of tropical cyclones,especially after landing.In general,there is a more important influence on the relative distance when the tropical cyclone intensity is weak and the decrease of relative distance will turn slow while the tropical cyclone shows a stable morphological structure.In particular,statistical analysis found that MMV decrease drastically during the tropical cyclone intensity transform from weaker than TD to TD and TD to TS,indicating that the tropical cyclone organization will strengthen apparently during its generation and extinction periods.This is convenient for the identification of tropical cyclone genesis by the MMV threshold.According to the study of tropical cyclone intensity estimation process in the Western North Pacific from 2015 to 2017,it is found that the traditional Sigmoid model that fitted by the MMV single factor has poor stability and low precision.At the same time,the sensitivity experiment of DAV-T calculation radius for Sigmoid model shows that a larger grid radius will produce a better effect on the fitted model,which makes the model less efficient in the application.To overcome this problem,the multi-factor model was established by introducing factors such as MMV,MDAV,RD,TBB_m,S-20,TBB_std,Lat and Lon in the intensity estimation process.The relationship between these factors and the tropical cyclone intensity is established by the generalized linear model?GLM?,and the effects of different factors on the estimation precision are investigated in turn.The results show that the factors MMV,S-20,TBB_std,RD and Lon have a significant effect on the estimation accuracy,especially the first three factors,the contribution of which are significantly higher than other factors among different verification sets.However,the factors MDAV,TBB_m and Lat contribute relatively little to the intensity estimation.The GLM estimation results by the five factors of MMV,S-20,TBB_std,RD and Lon are in good agreement with the CMA-BestTrack dataset in different tropical cyclone intensity stages.In addition,according to the sensitivity experiment of DAV-T calculation radius for GLM,the estimation results are good even with little radius.In summary,the introduction of multiple factors in DAV-T by GLM can significantly improve the accuracy of tropical cyclone intensity estimation.