Evaluation of distributed soil moisture simulations through field observations during the North American monsoon in Redondo Creek, New Mexico

Soil moisture is a key ecohydrological variable connecting ecosystem functions with hydrological processes in forested mountain basins. Nevertheless, few studies have investigated the spatiotemporal variations of soil moisture in mountainous settings or attempted to understand these through a modelling framework. In this study, we evaluate distributed soil moisture simulations in the Redondo Creek, New Mexico, using a set of field observations consisting of hourly data at a weather station and daily samples of soil moisture obtained during an intensive field campaign. Results indicate that simulated soil moisture patterns during the North American monsoon (NAM) are in reasonable agreement with observations across a range of ecosystems (wetland, grassland and forest). Using a set of temporal and spatial statistics, we quantify the distributed model performance at the point, ecosystem and basin scales. We find improved performance at the grassland and forest sites as compared to the near-stream wetlands. As a strong test of the model, we also compare the observed and simulated soil moisture probability density functions across the different ecosystems, finding that the overall match for grassland and forest was superior to the wetland areas. On the basis of the model simulations, we quantify, through visual inspection and geostatistical analysis, the controlling factors (rainfall, topography, vegetation and soils) on the soil moisture spatial patterns at different stages during the summer monsoon. Confidence in the distributed model gained through this analysis can open avenues for process studies and basin-scale predictions of mountain ecohydrology. Copyright (C) 2008 John Wiley & Sons, Ltd.