Skip to main content
ARS Home » Research » Publications at this Location » Publication #237812
Title: Multi-Objective Calibrationo of Hydrologic Model Using Satellite Data

Author
item CHEN, FAN - SSAI
item Crow, Wade

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 3/10/2009
Publication Date: 4/15/2009
Citation: Chen, F., Crow, W.T. 2009. Multi-objective calibration of hydrologic model using satellite data [abstract]. Abs. 10, BARC Poster Day.

Interpretive Summary:

Technical Abstract: Hydrologic modeling often involves a large number of parameters, some of which cannot be measured directly and may vary with land cover, soil or even seasons. Therefore parameter estimation is a critical step in applying a hydrologic model to any study area. Parameter estimation is typically done by a manual or automated calibration of the model in order to identify the parameter values that produce the best fit between model predicts and observations. While observed stream flow records are often the only available data for calibration, the emergence of remote sensing provides an independent data source for the calibration task, which implies tremendous potential for data-sparse or ungaged basins. We used thermal remote sensing-based surface evapotranspiration (ET) data as well as observed stream flow data to carry out a series of single- or multi- objective calibrations of the USDA-ARS developed watershed model Soil and Water Analysis Tool (SWAT). A sensitivity analysis was performed first to identify the parameters that both stream flow and ET are sensitivity to. Then the model was calibrated, using an automated global search algorithm, under different scenarios designed to reveal the effects of integrating remote sensing data on improving the hydrologic simulations: calibration against stream flow alone, ET alone and both variables at the same time with equal weights. Preliminary results suggested that (1) multi-objective calibration converges faster than single-objective calibration and result in comparable improvements of stream flow prediction as to the calibration of stream flow alone, i.e. integrating remote sensing ET data led to higher optimization efficiency; (2) calibration of evapotranspiration alone also resulted in improved runoff estimation, especially in the monthly time step. Although it did not produce as good hydrograph as direct calibration of stream flow, the annual water balance was more realistic. The results of this work can help enhance watershed or regional scale conservation practice assessment through improved simulation of hydrologic processes as well as nutrient and sediment loads.