A Pseudo-Ensemble/Variational Hybrid Assimilation Method Based On Historical Forecasts

To design a convenient data assimilation method with anisotropic and inhomogeneous background error covariance without increasing the computation amount, a pseudo-ensemble/variational hybrid data assimilation method is proposed based on historical forecasts. The new hybrid data assimilation method combines the pseudo-ensemble forecast error covariance represented by a set of historical forecast errors with the static background error covariance. The historical forecast error covariance was calculated from the forecasts of difference between the different forecasts respectively valid at the same time.Based on pseudo-ensemble-variational hybrid data assimilation system, a series of single observation experiments and cycling assimilation and forecasting experiments for two heavy rainfall event and typhoon Fanapi were performed compared to three dimensional variational method(3DVar). The main conclusions as follows:(1)The results of the single observation experiments indicate that the pseudo-ensemble/variational hybrid data assimilation method can introduce anisotropic and inhomogeneous covariance information into the data assimilation system without increasing the computation amount.In addition, correlations between moisture and other control variables are introduced by the pseudo-ensemble background error covariance, the moisture correlations are difficult to calculate in 3DVar and are not take into account in WRFDA-3DVar default control variables option. It is via these additional moisture correlations denoted in pseudo-ensemble background error covariance that the observations of wind, temperature and surface pressure can impact moisture field.(2)Two groups of cycling assimilation and forecasting experiments of two heavy rainfall events indicate that the hybrid experiments provide better forecast skill than the 3DVar for the rainfall, since the anisotropic and inhomogeneous background error covariance and correlations between moisture and other control variables are introduced into the pseudo-ensemble/variational hybrid data assimilation system, producing a more reasonable distribution of analysis and forecasts of the wind field and the precipitable water.(3) A series of assimilation and simulation experiments for typhoon Fanapi show that the track, minimum sea level pressure and wind speed using the pseudo-ensemble/variational hybrid data assimilation method were better than that of 3DVar. At the same time, the weight coefficient, the horizontal localization length scale and the number of historical forecast errors will influence the performance of assimilation and forecasting.(4) Furthermore, the new hybrid method using historical forecast errors to calculate forecast error covariance rather than using ensemble forecast error, thus, in addition to effectively improving the analysis and forecasts, this hybrid method also has the merits of simple calculation and high calculation efficiency.