Assimilation Of Satellite Cloud Observations Based On Hydrometeor Control Variables

The distribution,pattern,and changes of clouds or cloud systems reflect the state of atmospheric and dynamic trends,and cloud information has an essential value for analysis and forecast.Cloudy satellite data contain a large amount of dynamic,thermal,and water mass information related to typhoons,rainstorms,and other significant weather.The rational assimilation of satellite observations in cloud areas is of great importance to improve the hydrometeor variables in the model's initial field.This paper constructs the background error covariance of hydrometeor to explore the assimilation method of cloudy satellite data.It achieves the direct analysis of hydrometeor variables by introducing the hydrometeor control variables.Then,the adaptive partition assimilation,cloud-dependent variational assimilation,and the ensemble-variational hybrid assimilation based on hydrometeor control variables are established.The paper presents a series of studies on the scientific problem of how to assimilate satellite data in the cloud area in a more reasonable way.Firstly,this paper constructs the hydrometeor background error covariance and realizes the direct analysis of the hydrometeor variables by introducing the hydrometeor control variables.The spatial correlation shows that the hydrometeor variable has a smaller scale and stronger localization than the conventional variables;the variable correlation indicates that the hydrometeor is mainly related to the temperature,water vapor,and potential velocity functions.A series of single observation experiments show that a reasonable characterization of the spatial correlation and variable correlation of hydrometeor allows the increment of hydrometeor to propagate reasonably in horizontal and vertical and allows its increment to be transferred to other conventional variables to achieve inter-variate balance.Secondly,this paper further constructs the background error covariance of hydrometeor for cloud area and clear sky area based on the cloud classification technique.Then,we realize the adaptive piecewise assimilation of satellite data in cloud area based on this technique.The characteristics of the cloudy and clear background error covariance show that the cloudy background error covariance has a larger background error,stronger variable correlation,and smaller horizontal length scale.The adaptive piecewise assimilation scheme based on the cloud classification technique can adaptively determine the cloudy areas and adjust the background error of cloud areas based on satellite cloud observations.Single observation tests show that the larger background error in the cloud area makes the analyses place more weight on the observation than the clear sky.The results of the assimilation and forecasting experiments show that the adaptive piecewise assimilation scheme makes the magnitude and distribution of hydrometeor and water vapor in the cloudy area more reasonable,thus improving the precipitation forecast.Thirdly,based on the "cloud dependence" approach,this paper introduces the clouddependent background error covariance of the hydrometeor background field in the adaptive piecewise assimilation scheme and realizes the analysis of cloud dependence in the variational framework.Based on the satellite cloud amount,a cloud-dependent adjustment index is constructed so that background error of each variables varies with the cloud amount.The larger the cloud amount,the larger the background error,and vice versa.A series of single observation experiments show that the cloud-dependent hydrometeor background field error covariance both alleviates the problem of boundary discontinuity in the adaptive partitioning assimilation scheme and enables the analysis of cloud-dependent and multivariate correlation in a variational framework with little increase in computational resources.The precipitation assessment experiments show that the application of cloud-dependent hydrometeor background error covariance can more effectively reduce the analysis and forecast errors in the dynamic,thermal,and water vapor fields using satellite observations of cloud areas and thus effectively improve precipitation scores.The diagnosis of individual cases of strong convection shows that the increment of cloud-dependent hydrometeor analysis can be transferred to the humidity and wind fields,and the improvement of the humidity and dynamical fields can further support the development of hydrometeor and facilitate the development of strong convection.Finally,this paper constructs a hydrometeor ensemble-variational hybrid assimilation scheme based on the extended control variable method to achieve hybrid assimilation of satellite data in cloud areas.The hydrometeor hybrid assimilation scheme can combine the background error covariance of hydrometeor in climate state and the flow-dependent and multivariate correlated background error covariance calculated by the ensemble forecasts,which can introduce the flow-dependent hydrometeor analysis as well as alleviate the spurious correlation caused by the sampling error.Single observation experiments show that the hydrometeor mixing assimilation scheme can produce a flow-dependent increment to the hydrometeor analysis.The multivariate correlation in the ensemble covariance allows this increment to be transferred to the conventional variables.The precipitation assessment experiments show that the En Var-Hydro scheme effectively reduces errors in the analysis and forecast fields of traditional variables and improves precipitation forecast skill scores.Detailed diagnostics show that the ensemble-variational hybrid assimilation of satellite data in cloud areas can effectively enhance the distribution of hydrometeor in the model field.The improvement of hydrometeor further enhances the cloud area moisture analysis and convective effective potential energy through inter-variate correlation,which makes the model development more balanced and effectively prolongs the role of satellite cloud observation information in the model.