Assimilating Remote Sensing-Based ET into SWAP Model for Improved Estimation of Hydrological Predictions

An agro-hydrological simulation model is useful for agriculture monitoring and Remote Sensing provides useful information over large area. Combining both information by data assimilation is used in agro-hydrological modeling and predictions, where multiple remotely sensed data, ground measurement data and model forecast routinely combined in operational mapping procedures. Remote sensing cannot observe input parameters of agro-hydrological models directly. A method to estimate input parameters of such model from Remote Sensing using data assimilation has been proposed by Ines [2002] using the SWAP (Soil, Water, Atmosphere and Plant) model. A Genetic Algorithm (GA) loaded stochastic physically based soil-water-atmosphere-plant model (SWAP) was extended for the discussed problem and used in the study. The objective of this study was to implement a data assimilation scheme to estimate hydrological parameters (e.g soil moisture) of SWAP model. For this study six Landsat TM/ETM satellite images were obtained for part of the Great Plains (Path 29, Row 32) in the states of Nebraska (NE) for the 2006 growing season (May-October). Then a land surface energy balance model (METRIC) was used to map spatiotemporal distribution of evapotranspiration. The ability of METRIC accuracy was compared with the measurements at several flux sites with Bowen Ratio Energy Balance System units. Remotely sensed ET data and ground measurement data from experiment fields were then combined in a data assimilation to estimate parameters of the SWAP model. The system is initialized with a population of random solutions and searches for optima by updating generations. The result shows that the reasonable parameters (sowing date and harvesting date, Ground water level) were successfully estimated. On the basis of estimated parameters, soil moisture is predicted by SWAP model. The agro-hydrological model driven by the observed ET produces reasonable water cycle states and fluxes, and the estimates are moderately improved by assimilating ET measurements that provides information on the surface soil moisture state, while it remains challenging to improve the results by assimilating regional ET estimated from satellite-based measurements.