Using globally available soil moisture indicators for flood modelling in Mediterranean catchments
Abstract. Floods are one of the most dangerous natural hazards in Mediterranean regions. Flood forecasting tools and early warning systems can be very beneficial to reducing flood risk. Event-based rainfall–runoff models are frequently employed for operational flood forecasting purposes because of their simplicity and the reduced number of parameters involved with respect to continuous models. However, the advantages related to the reduced parameterization oppose to the need of a correct initialization of the model, especially in areas characterized by strong climate seasonality. In this case, the use of continuous models could be desirable but it is very problematic in poorly gauged areas where continuous rainfall and temperature data are not available. This paper introduces a Simplified Continuous Rainfall–Runoff model (SCRRM), which uses globally available soil moisture retrievals to identify the initial wetness condition of the catchment, and, only event rainfall data to simulate discharge hydrographs. The model calibration involves only three parameters. For soil moisture, besides in situ data, satellite products from the Advanced SCATterometer (ASCAT) and the Advanced Microwave Scanning Radiometer for Earth observation (AMSR-E) sensors were employed. Additionally, the ERA-Land reanalysis soil moisture product of the European Centre for Medium-Range Weather Forecasting (ECMWF) was used.
SCRRM was tested in the small catchment of the Rafina River, 109 km2, located in the eastern Attica region, Greece. Specifically, sixteen recorded rainfall–runoff events were simulated by considering the different indicators for the estimation of the initial soil moisture conditions from in situ, satellite and reanalysis data. By comparing the performance of the different soil moisture products, we conclude that: (i) all global indicators allow for a fairly good reproduction of the selected flood events, providing much better results than those obtained from setting constant initial conditions; (ii) the use of all the indicators yields similar results when compared with a standard continuous simulation approach that, however, is more data demanding; (iii) SCRRM is robust since it shows good performances in validation for a significant flood event that occurred on February 2013 (after calibrating the model for small to medium flood events). Due to the wide diffusion of globally available soil moisture retrievals and the limited number of parameters used, the proposed modelling approach is very suitable for runoff prediction in poorly gauged areas.