A Land Data Assimilation System for Soil Moisture and Temperature: An Information Content Study

Abstract

Abstract A Canadian Land Data Assimilation System (CaLDAS) for the analysis of land surface prognostic variables is designed and implemented at the Meteorological Service of Canada for the initialization of numerical weather prediction and climate models. The assimilation of different data sources for the production of daily soil moisture and temperature analyses is investigated in a set of observing system simulation experiments over North America. A simplified variational technique is adapted to accommodate different observation types at their appropriate time in a 24-h time window. The screen-level observations of temperature and relative humidity, from conventional synoptic surface observations (SYNOP)/aviation routine weather report (METAR)/surface aviation observation (SA) reports, are considered together with presently available satellite observations provided by the Aqua satellite (microwave C-band), Geostationary Operational Environmental Satellite (GOES) [infrared (IR)], and observations available in the future by the Soil Moisture and Ocean Salinity (SMOS) satellite mission (microwave L-band). The aim of these experiments is to assess the information content brought by each observation type in the land surface analysis. The observation systems are simulated according to their spatial coverage, temporal availability, and nominal or expected errors. The results show that the observable with the largest dynamical response to perturbations of the control variable carries the greatest information content into the analysis. The observational error and the observation frequency counterbalance this feature in the analysis. If one considers a single observation both for soil moisture and soil temperature analysis, then satellite measurements (L-band, C-band, and IR in decreasing order of importance) are the primary source of information. When observation availability is considered and the highest temporal frequency of screen-level observations is used (1 h), a large amount of information is extracted from SYNOP-like reports. The screen-level observations are shown to provide valuable soil moisture information mainly during the daytime, while during nighttime these observations (and particularly screen-level temperature) are mostly useful for the soil temperature analysis. The results are presented with perspectives for future operational developments and preliminary assimilation experiments are performed with hourly screen-level observations.