Assimilation Of Lightning Data In Mesoscale Numerical Model

A large number of observational analyses have shown that lightning data can be used to indicate areas of deep convection.Therefore,it is important to assimilate observed lightning data into numerical models,so that more small-scale information can be incorporated to improve the quality of the initial conditions and the subsequent forecasts.In this study,we conduct an exploratory study of lightning data assimilation that is suitable for operational run.The lightning data was used to adjust the model relative humidity,and the adjusted relative humidity was output as pseudo sounding observation,which was assimilated into WRFDA(WRF Data Assimilation system).Based on the three-dimensional variational data assimilation(3DVAR)technique,which was currently widely used in many operational NWP(Numerical Weather Prediction)centers,experiments with different configurations was conducted to study the suitability of lightning data assimilation.And the results of lightning data assimilation was compared with that of radar reflectivity data assimilation.Moreover,exploratory study of tropical lightning data assimilation was conducted to improve the forecast of tropical cyclone intensity based on the most powerful landing typhoon “Haiyan”.Finally,the hybrid ensemble-variational data assimilation technique that developed in recent years was used to further improve the result of lightning data assimilation.The main conclusions are summarized as follows:(1)It is feasible and effective to assimilate the relative humidity retrieved from lightning data by utilizing the 3DVAR method,with significant improvement both in the initial condition and subsequent forecast.Lightning data should assimilated with cycyling mode rather than single mode.The experiment in which lightning data was assimilated only once showed little or no improvement in the initial condition.The other three experiments,in which lightning data were assimilated in the cycling mode,showed varying degrees of improvement.From the comparative analyses,10-min interval was appropriate for cycled lightning data assimilation;and 60 min was the appropriate assimilation window length.Longer assimilation window length was probably unnecessary,yet shorter assimilation window length would not sufficient to effectively improve the initial condition.The use of double-moment microphysics scheme could further improve the forecast compared with that of single-moment microphysics scheme.Based on subjective comparisons and objective statistical scores,the improvement from lightning data assimilation could be maintained for about 3 h.(2)Based on the fact that lightning data shares some common features with radar reflectivity data(both data can be used to indicate convection yet both are not conventional model variable),the result of lightning data assimilation was compared with that of radar reflectivity data assimilation.It was found that assimilation of radar reflectivity data could effectively improve the initial conditions in a relatively short time length(less than 30 min),while lightning data assimilation needs longer time(about 1 h).Nevertheless,the forecast based on lightning data assimilation was more consistent with observation than that of radar reflectivity data assimilation,indicating that lightning data assimilation has the potential to replace radar reflectivity data assimilation in those regions where radar reflectivity data are generally not available(oceans and terrain-blocked areas).(3)Exploratory study of tropical cyclone(TC)lightning data assimilation was conducted to improve the forecast of TC intensity based on the most powerful landing typhoon “Haiyan”.It was found that assimilation of the inner core(0-100 km from TC center)lightning data can improve the TC intensity forecast,while assimilation of the lightning data in rainband regions has no improvement(if no negative effect)on the model forecast.Meanwhile,the time to start lightning data assimilation was very important.It was found that the improvement of assimilation of lightning data before the rapid intensification(RI)period was not as remarkable as that in the early stage of RI,and assimilation of lightning data in the late stage of RI has barely improvement on forecast.Overall,the improvement from lightning data(inner-core lightning data in the early stage of RI)assimilation could be maintained for about 60 h.(4)Concerning the background error covariance is assumed to be stationary and isotropic in 3DVAR,which is obviously not the truth and may depreciate the result.The hybrid ensemblevariational data assimilation technique that developed in recent years was used to further improve the result of lightning data assimilation.The results showed that the assimilation of lightning data with hybrid ensemble-variational method could partially improve the forecast of the precipitation center postion compared with that of 3DVAR,indicating the ensemble-based data assimilation technique has its potential advantage on lightning data assimilation.