Assessment of terrestrial water balance using remote sensing data in South America

Abstract

This study investigated the potential of assessing the water balance in South America, at multiple scales, using satellite remote sensing precipitation and evapotranspiration datasets, terrestrial water storage changes from Gravity Recovery and Climate Experiment (GRACE), and discharge measurements. The remote sensing precipitation datasets included those of the Tropical Rainfall Measuring Mission (TRMM) and the Multi-Source Weighted-Ensemble Precipitation (MSWEP), whilst the ET constituents included the MODIS Global Evapotranspiration Project (MOD16) and the Global Land Evaporation Amsterdam Model (GLEAM). Uncertainties in precipitation and ET were evaluated using in situ measurements of 307 rain-gauge stations and 16 eddy covariance sites, respectively. The water balance closure and uncertainties were evaluated in 50 basins for the period 2003–2014, at different spatial scales and under different climatic conditions. Overall, MSWEP precipitation and GLEAM ET provided fewer uncertainties, whilst TRMM and MOD16 yielded greater bias and more errors. Better agreements between the TWSC from GRACE and the water balance were found in the large and medium-sized basins, with a root mean squared error (RMSE) of about 42%, whilst small basins and those located in regions with subtropical humid climates, where precipitation did not present strong seasonality, had a RMSE around 130%, indicating that climate conditions can influence the water balance closure. The TWSC from GRACE and the remote sensing water balance estimations were in agreement for more than half of the evaluated basins, mainly those located in tropical climate regions. Greater bias and more errors were found in the estimations of discharge in the semi-arid basins, with low runoff coefficients. Despite the water balance based on remote sensing estimations remaining a challenge due to the large uncertainties in precipitation, ET, and TWSC, our results showed their great potential for assessing terrestrial water-cycle dynamics and understanding their spatial and temporal variability at multiple scales.