Evaluation And Quality Control Of All-sky IR Radiance Simulation For Typhoon Based On Himawari-8

The assimilation of cloudy infrared(IR)radiance is significant for improving the initial value of Numerical Weather Prediction(NWP)systems,which is more complicated compared to the assimilation of clear-sky satellite radiance data.There are many reasons for that namely,the strong nonlinearity of cloud-related process,non-normality and large variances of O-B(observation-minus-background),the poor performance of cloud process in NWP and radiative transfer(RT)models.Based on AHI observations of six intense typhoon landed in China(named as NEPARTAK,NIDA,MERANTI,MEGI,SARIKA,and HAIMA)in 2016,the statistical characteristics of WRF and RTTOV models have been investigated in this research.To assimilate the all-sky IR radiance data effectively,samples that are poorly reproduced should be carefully eliminated.The results shows:(1)The WRF model was used to simulate the typhoons'temperature,humidity,near-surface wind field and cloud profile,and the simulated track and intensity were compared to Tropical Cyclone Best Track Dataset.The average deviation between the simulated and the observed tracks is less than 75km and the intensity difference of low pressure center is less than 35hPa.It indicates that the tracks and the typhoon center pressure are simulated effectively by WRF except some samples affected by the terrain.(2)BTs(Brightness Temperature)of Himawari-8/AHI were simulated using RTTOV model.The main research channels were water vapor channels(Band 8:6.2?m and Band10:6.9?m),atmospheric window area(Band 13:10.4?m)and the troposphere lower temperature channel(Band 16:13.3?m).The observation-minus-background(O-B)difference between simulated and observed temperatures under clear sky and cloud conditions was compared,and the mean bias(MB),the root mean squared error(RMSE),the correlation coefficient(COR),and the index of agreement(IOA).The results show that the clear-sky simulation cannot reproduce the location,shape and cloud system of the typhoon eye area,whereas the cloudy simulation can better reproduce the position of the typhoon eye and the dense spiral cloud system.however,underestimated BT depression in simulations(i.e.simulated BT is warmer than observed BT are found in some regions,especially for developed cloud systems such as the center region of a typhoon)In the sensitivity test of the cloud parameters,channel13 is the most sensitive to clouds,followed by the channels 16,10,and 8.Correspondingly,under the cloud conditions,the simulation effect of channel 8 is the best,followed by channels 10 and 16,whereas channel 13 has the worst simulation effect.This indicates that the more sensitive the channel is to the cloud,the greater the difference is between the observation and the simulation.(3)simulated BT compared to the observed one has lower density,a higher temperature in the center,poor performance on low observed temperature,which may be related to the lack of high cloud prediction and calculation of cloud scattering in the model.Two QCs(Quality Control)were developed in this paper:one excluded extremely low BTs(low-BT QC)and the other removed split samples which were affected by clouds extremly.For the first QS 200k sample points were used as the threshold value of low-BT QC,whereas for the second standard,we introduced cloud effect named C_A and the boundary C_A>3.0,|O-B|>1.8C_A was applied.Comparing the O-B probability density distributions of different typhoons prior to and following the quality control,the results show that after quality control,the O-B differences of the water vapor channels(channel 8,10)are significantly improved(the average deviation is reduced by about 1K),and the probability density function comes to normal distribution.Channel 13 and 16 did not improve much after the quality control It is concluded that this type of quality control is more applicable to the water vapor channels and can be deployed in similar channels or similar instruments.