Experimental use of TRMM precipitation radar observations in 1D+4D-Var assimilation

This paper presents a new application of the Tropical Rainfall Measuring Mission (TRMM) precipitation radar (PR) observations for indirect assimilation into the European Centre for Medium-Range Weather Forecasts (ECMWF) model. The PR reflectivities are first processed using a one-dimensional variational (1D-Vars) method to adjust model temperature and specific humidity. The retrieved Total Column Water Vapour (TCWV) is then assimilated into the operational four-dimensional variational (4D-Vars) system. The applicability of the I D+4DVar approach to the radar observations is discussed in detail. Several case studies were run to assess the feasibility and the effectiveness of assimilating PR reflectivities with a 1D-Vars approach. Results show good behaviour of the 1D-Vars system in terms of convergence and stability. Its performance in terms of retrieved TCWV is comparable to that of other 1D-Vars which make use of TRMM Microwave Imager (TMI) observations. When the 1D-Vars TCWV pseudo-observations are input into the 4D-Vars system, a positive impact is shown in the analysis and the subsequent forecasts, both on moisture-related fields and also on winds and surface pressure. The quality of the forecast is verified using track observations for the tropical cyclones. The track forecasts from the experiments which include 1D-Vars TCWV are generally closer to the observed track than a control run. Despite their much smaller spatial coverage than TMI observations, it is found that the PR data have a comparable impact, provided the satellite samples a meaningful portion of the storm, possibly its centre. This is possibly due to the fact that TCWV increments from PR and from TMI brightness temperature have similar magnitudes. These results show that active sensor data can provide indirect yet useful information on the moisture field and that this information can effectively be assimilated to improve the analysis and the forecast of tropical disturbances. Although this is a sub-optimal use of PR observations, due to the underexploitation of the vertical information contained in the reflectivity profiles, it is still a step forward towards using active sensor data in preparation for future satellite missions which will deliver this type of data on a global scale and at higher temporal resolution.