|Kustas, William - Bill|
Submitted to: Agricultural and Forest Meteorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/15/2009
Publication Date: 11/3/2009
Citation: Gonzalez-Dugo, M.P., Neale, C., Mateos, L., Kustas, W.P., Prueger, J.H., Anderson, M.C., Li, F. 2009. A comparison of operational remote sensing-based models for estimating crop evapotranspiration. Agricultural and Forest Meteorology. 149:1843-1853.
Interpretive Summary: The integration of remotely sensed data into models of actual evapotranspiration (ET) has broadened the field of application of these models from point to basin and regional scales. Operational applications of remote sensing-based ET models to hydrology and agriculture is considered essential for large scale water resource monitoring. Two types of remote-sensing based modeling approaches for estimating crop ET have been successfully applied in agricultural water use studies. The first approach computes a surface energy balance using the radiometric surface temperature as a key variable with several levels of complexity for modeling land-atmosphere energy exchange and input requirements. The second approach relies on the ability of vegetation indices to trace the crop growth and estimate a crop coefficient used to adjust a reference ET estimated from local weather station data. These models were evaluated and compared using remote sensing data and ground truth ET observations collected during the Soil Moisture-Atmosphere Coupling Experiment [SMACEX] conducted over rain fed corn and soybean crops in central Iowa. The surface temperature-based energy balance models provided reasonably accurate ET estimates with the more complex modeling scheme giving slightly better performance. The performance of the crop coefficient-based model was comparable to that of the surface temperature-based energy balance models, although the correlation coefficient between predicted and observed ET was lower. In addition the crop coefficient-based model is unable to consider reduction in ET due to plant water stress. Hence the surface temperature-based models are more suitable for estimating crop ET under water limited conditions.
Technical Abstract: The integration of remotely sensed data into models of actual evapotranspiration has allowed for the estimation of water consumption across agricultural regions. Two modeling approaches have been successfully applied. The first approach computes a surface energy balance using the radiometric surface temperature for estimating the sensible heat flux, obtaining the evapotranspiration as a residual of the energy balance. The main difference between existing energy balance models proposed in the literature is in the method of estimating sensible heat flux (H). This paper compares the performance of three different approaches for estimating H: an empirical one-source energy balance model, a second one-source model calibrated using inverse modeling of extreme fluxes in the satellite scene and a two-source energy balance model. The second approach is based on the ability of vegetation indices derived from canopy reflectance to trace the basal crop coefficients (reflectance-based crop coefficients) that are used in conjunction with the reference evapotranspiration to obtain crop evapotranspiration. Local meteorological and soil data are required as a water balance in the root zone of the crop must be maintained. Model output was compared to sensible and latent heat fluxes measured during the Soil Moisture-Atmosphere Coupling Experiment [SMACEX] conducted over rain fed corn and soybean crops in central Iowa. The root mean square differences (RMSD) of the estimation of instantaneous latent and heat fluxes were less than 50 Wm-2 for the three energy balance models. The two-source energy balance model generally produced better agreement with measured fluxes than did the one-source approaches. For these models, daily evapotranspiration was obtained using an evaporative fraction that scales instantaneous to daily values, yielding RMSDs in daily evapotranspiration that were less than 0.60 mm d-1. The RMSD for the reflectance-based crop coefficient predictions was comparable to those of the energy balance models. However, crops were generally not under water stress conditions during the modeling period. Crop stress can be directly detected using radiometric surface temperature, but ET approaches based vegetation indices will not be sensitive to stress until there is actual reduction in biomass or changes in canopy geometry.