H41A-0779:
The Potential of Remotely Sensed Evapotranspiration and Soil Moisture Retrievals in Calibrating Land Surface Models

Thursday, 18 December 2014
Aiswarya Kunnath Poovakka1, Dongryeol Ryu1, Luigi J Renzullo2 and Biju George3, (1)The University of Melbourne, Parkville, Australia, (2)CSIRO Land and Water, Canberra, Australia, (3)Integrated Water and Land Management Program, ICARDA, Cairo, Egypt
Abstract:
Model calibration is frequently limited by the availability, quality, quantity and the nature of ground observations. Remotely sensed soil moisture and evapotranspiration (ET) provide an alternative source of hydrological information to inform models. In this study, microwave soil moisture retrievals from the Advanced Microwave Scanning Radiometer-EOS (AMSR-E) and daily estimates of ET from CSIRO MODIS ReScaled potential ET (CMRSET) model are adopted to calibrate a land surface model using 15 different objective functions considering various combinations of bias and correlation of ET and soil moisture. The Shuffled Complex Evolution (SCE) calibration algorithm is used to calibrate a grid-based land surface model modified from the Australian Water Resource Assessment - Landscape (AWRA-L) model. The study catchments are located in south Australia with ground observations of ET and soil moisture for validation. Parameters for calibration are selected based on the results from variance-based Sobols’ sensitivity analysis. The efficacy of each calibration is assessed mainly based on the streamflow predictability of the calibrated model.

Calibration schemes with a greater emphasis on ET provide good estimates of ET but the streamflow predictions are comparatively poor. It is found that the streamflow predictions improve when higher importance is given to soil moisture derived objective functions. The optimized parameter values exhibit wide variations for different objective functions even though they provide similar values for streamflow, ET and soil moisture. Lastly, we discuss about the reasons for poor performance of the ET-based calibration and the impact of physical properties of the catchment on the calibration. This study has important implications to the optimal use of remotely sensed observations for hydrological modeling at large catchments with sparse or no gauging.