| Ground-based microwave radiometers are passive remote sensing instruments that receive thermal radiation power emitted by a target at a specific frequency.As such,their direct data consists of radiance data,while their secondary product is an atmospheric temperature and humidity profile derived through inversion methods.Currently,the assimilation of ground-based microwave radiometer data is performed indirectly through inversion assimilation.However,this approach can introduce significant errors and negatively impact assimilation results.Direct assimilation can avoid these errors.This paper addresses the issue of large inversion errors during indirect assimilation of ground-based microwave radiometers and the inability to achieve direct assimilation.It presents research on quality control for inversion data and direct assimilation applications for ground-based microwave radiometer radiance data.A quality control scheme for indirect assimilation inversion data is designed and its effectiveness verified.RTTOV-gb is selected as the observation operator for direct assimilation in WRFDA and a direct assimilation module for ground-based microwave radiometer radiance data is implemented in WRFDA.The capability of RTTOV-gb brightness temperature simulation was tested along with single-point tests and direct assimilation experiments.The results show that:(1)Before quality control was applied to ground-based microwave radiometer inversion temperature and humidity profile data,the overall error in inversion temperature profiles across all sites was small.The scatter plot of FNL interpolation to each site showed low dispersion throughout the layer.Average deviation and root mean square error indicated that only a few sites had larger errors in inversion temperature between 0.5-3 km.However,due to the presence of error deviation prior to quality control,observation errors did not conform to a standard Gaussian distribution.This suggests that temperature data quality was good before quality control was applied.In contrast,inve rsion relative humidity data showed large overall errors for each site.Root mean square error for all sites throughout the layer exceeded 20%,with some reaching as high as 50%.Additionally,error distribution did not conform to a Gaussian normal distribution.(2)After undergoing quality control,the scatter plot of the inverted temperature profile data displays a reduced degree of dispersion at higher levels.Both the average deviation and root mean square error have decreased at these levels.The distribution of errors after quality control conforms to a Gaussian normal distribution.Similarly,after quality control,there has been a significant improvement in the overall data quality of the inverted relative humidity profile.Points with high degrees of dispersion in the scatter plot have been removed through quality control measures,resulting in a notable decrease in dispersion and an error reduction to within 50% of its original value.Assimilating temperature and humidity profile data that has undergo ne quality control can effectively enhance a model’s ability to forecast temperature and humidity fields as well as precipitation.(3)The RTTOV-gb simulation of ground-based microwave radiometer brightness temperature has shown to perform well and can serve as an observation operator for the direct assimilation of ground-based microwave radiometer data.By simulating the brightness temperature of station 54399 from June to October 2019 using both RTTOV-gb and Mono RTM and comparing the results with observed brightness temperature data,it was determined that RTTOV-gb’s performance in simulating ground-based microwave radiometer brightness temperature is on par with that of Mono RTM.Furthermore,after constructing a direct assimilation module within WRFDA,a single-point test was carried out to confirm the accuracy of the module and the validity of its algorithm.(4)Direct assimilation can more effectively assimilate ground-based microwave radiometer data and improve the temperature and humidity initial fields of the model,thereby enhancing the model’s ability to forecast extreme precipitation.By comparing the impact of direct and indirect assimilation on the model’s temperature and humidity initial fields,it was found that direct assimilation can reduce the error in the model’s initial field and bring it closer to the actual observed field.In contrast,indirect assimilation can increase the error in the initial field due to large errors introduced by data inversion methods,resulting in a negative effect.Moreover,from the perspective of 6-hour accumulated precipitation TS scores,direct assimilation can effectively improve large-scale precipitation forecasts by the model,particularly for heavy rain events where improvements are significant. |
