Precipitation fields interpolated from gauge stations versus a merged radar-gauge precipitation product: influence on modelled soil moisture at local scale and at SMOS scale

Abstract. For the validation of coarse resolution soil moisture products from missions such as the Soil Moisture and Ocean Salinity (SMOS) mission, hydrological modelling of soil moisture is an important tool. The spatial distribution of precipitation is among the most crucial input data for such models. Thus, reliable time series of precipitation fields are required, but these often need to be interpolated from data delivered by scarcely distributed gauge station networks. In this study, a commercial precipitation product derived by Meteomedia AG from merging radar and gauge data is introduced as a novel means of adding the promising area-distributed information given by a radar network to the more accurate, but point-like measurements from a gauge station network. This precipitation product is first validated against an independent gauge station network. Further, the novel precipitation product is assimilated into the hydrological land surface model PROMET for the Upper Danube Catchment in southern Germany, one of the major SMOS calibration and validation sites in Europe. The modelled soil moisture fields are compared to those obtained when the operational interpolation from gauge station data is used to force the model. The results suggest that the assimilation of the novel precipitation product can lead to deviations of modelled soil moisture in the order of 0.15 m³ m⁻³ on small spatial (∼1 km²) and short temporal resolutions (∼1 day). As expected, after spatial aggregation to the coarser grid on which SMOS data are delivered (~195 km²), these differences are reduced to the order of 0.04 m³ m⁻³, which is the accuracy benchmark for SMOS. The results of both model runs are compared to brightness temperatures measured by the airborne L-band radiometer EMIRAD during the SMOS Validation Campaign 2010. Both comparisons yield equally good correlations, confirming the model's ability to realistically model soil moisture fields in the test site. The fact that the two model runs perform similarly in the comparison is likely associated with the lack of substantial rain events before the days on which EMIRAD was flown.