A simple assimilation method to ingest satellite soil moisture into a limited-area NWP model

Recently several studies discussed the potential and operational use of satellite soil moisture measurements in new land surface analysis feeding global Numerical Weather Prediction (NWP) models. This work seeks to establish whether a limited-area NWP model might benefit from the assimilation of remotely sensed soil moisture data. The question is important because it is well known that even small errors in the initial conditions could amplify in the future states and lead to erroneous predictions. On the other hand, remotely sensed soil moisture observations are attractive because they offer a synoptic point of view and their reliability with respect to in-situ measurements is demonstrated.The authors used a simple nudging scheme in order to assimilate remotely sensed soil moisture data into a limited-area NWP model. This assimilation method is computationally cheap and simple to implement. Its impact on numerical outputs is evaluated with respect to a control simulation performed without assimilation. Results obtained in both simulations are validated with in-situ soil moisture measurements and with 2-meter temperature observations. Results demonstrate the benefits of the assimilation especially in those remote areas where the coverage of observational instruments is poor and where the irregularity of observations implies interpolation errors when reporting data on a regular grid. As found in bibliography, in well monitored areas the impact of the assimilation is almost neutral.