EVAPOTRANSPIRATION ESTIMATION FROM MODIS IMAGERY FOR THE CENTRAL RIO GRANDE

In semi-arid environments, evapotranspiration (ET) from stream channels and riparian vegetation often represents one of the most significant losses of water in the entire hydrologic system. Understanding the magnitude of this loss is critical to managing scarce water supplies and in mitigating the damage to riparian zones resulting from deficits in water supply caused by population growth and climatic variability. In our modeling approach, we use MODIS imagery to estimate daily potential evapotranspiration (PET) for two reaches of the Rio Grande in central New Mexico. Our ET model is currently calculated in ARCGIS but will be migrated to a TerraScan processing and modeling system to become a daily product. The approach applies an energy balance model to a 30-meter Digital Elevation Model to yield estimates of shortwave and longwave radiation fluxes. The minimum scaled reflectance for channels 1, 3 and 4 of the MODIS imagery is used to supply instantaneous albedo estimates while surface temperatures are approximated using a scaling of MODIS bands 31 and 32. All MODIS estimates are scaled to the underlying 30-meter resolution of the energy balance model. Air temperature and humidity estimates are incorporated from local weather station hourly data. Model output is calculated every three hours from 6 AM to 6 PM. Intermediate values are then interpolated to produce a set of PET estimates for each daylight hour. For the May 31, 2002 test case, we calculated a daily mean estimate across the central Rio Grande bosque of ~6.5 mm/day with extremes ranging from 3.4 to 8.5 mm/day depending on vegetation type and local topographic variability. We currently validate our model by comparing our ET estimates to ET measurements from multiple flux towers. Tower data for this date and site is ~7.0 mm/day suggesting that our model slightly underestimates ET in comparison to the flux tower measurements.