| It is well known that improved knowledge of the internal structure of typhoons can improve the accuracy of the parameterization of physical processes in numerical models and enhance the forecast of typhoon path and intensity.In the process of typhoon occurrence,development and movement,a considerable period of time is located in the middle and low latitude degree tropical ocean surface,so satellite observation plays a crucial role in filling the gap of basic typhoon data,and the Dropsonde over the ocean surface not only essential for detecting the internal structure of typhoons.The Dropsonde over the ocean surface not only provides an essential tool for detecting the internal structure of typhoons,but also provides a necessary tool for improving the accuracy of satellite microwave data retrieval.This paper takes the AMSU-A microwave sensor on board the European polar-orbiting satellite Metop-B as an example,evaluates and analyzes the accuracy of temperature and humidity profile retrieval of offshore typhoons by the current satellite microwave sensor and the causes of errors,and gives a preliminary revision method.The main research contents and conclusions are as follows.1.A space-time distance parameter is constructed to characterize the space-time match between the Dropsonde data and the satellite inversion data,and a Dropsonde-satellite inversion data set is established based on this parameter.A scheme is proposed to evaluate the accuracy of temperature and humidity profiles in the satellite inversion of typhoons using Dropsonde data,and the results are as follows:the average error of the whole layer inversion of temperature profiles is negative,the average error of the whole layer of temperature is-1?-2?,and the average error of humidity is about-5?-4*10-3 kg/kg;the root mean square error is 1.5?2.5?and1.5?3*10-3 kg/kg,respectively.The results of the three sets of control tests show that the closer the inversion position is to the typhoon center,the stronger the typhoon intensity is,and the whole layer of precipitable water reaches 60 mm or more,the lower the inversion accuracy is.It can be reasonably inferred that the thicker typhoon cloud system and larger typhoon precipitation will have a significant impact on the accuracy of the inversion of this product.2.Using BP neural network to directly revise the satellite inversion products,the average error of the revised temperature profile ranged from-1 to+0.5°C and the average error of the humidity profile ranged from-0.7 to 0.5*10-3 kg/kg,and the maximum average error of temperature was reduced from 2°C to 1°C and the maximum average error of humidity was reduced from 5*10-3kg/kg to 0.7*10-3 kg/kg.However,the lower layers showed larger error fluctuations.3.The physical-statistical inversion method of RTTOV forward and BP neural network inversion is used in combination,and the average error of the inversion temperature profile is in the range of-1 to+0.9°C,with a root mean square error of2-3°C.The inversion error is reduced by 0.53°C compared to the total error of the satellite inversion product.The vertical layer resolution of the satellite inversion profile below 700h Pa altitude is improved. |
PDF Full Text Request