Assimilating remotely sensed cloud optical thickness into a mesoscale model

Abstract. The Advanced Regional Prediction System, a mesoscale atmospheric model, is applied to simulate the month of June 2006 with a focus on the near surface air temperatures around Paris. To improve the simulated temperatures which show errors up to 10 K during a day on which a cold front passed Paris, a data assimilation procedure to calculate 3-D analysis fields of specific cloud liquid and ice water content is presented. The method is based on the assimilation of observed cloud optical thickness fields into the Advanced Regional Prediction System model and operates on 1-D vertical columns, assuming that the horizontal background error covariance is infinite, i.e. an independent pixel approximation. The rationale behind it is to find vertical profiles of cloud liquid and ice water content that yield the observed cloud optical thickness values and are consistent with the simulated profile. Afterwards, a latent heat adjustment is applied to the temperature in the vertical column. Data from several meteorological stations in the study area are used to verify the model simulations. The results show that the presented assimilation procedure is able to improve the simulated 2 m air temperatures and incoming shortwave radiation significantly during cloudy days. The scheme is able to alter the position of the cloud fields significantly and brings the simulated cloud pattern closer to the observations. As the scheme is rather simple and computationally inexpensive, it is a promising new technique to improve the surface fields of retrospective model simulations for variables that are affected by the position of the clouds.