Estimating evapotranspiration of riparian vegetation using high resolution multispectral, thermal infrared and lidar data

Abstract

High resolution airborne multispectral and thermal infrared imagery was acquired over the Mojave River, California with the Utah State University airborne remote sensing system integrated with the LASSI imaging Lidar also built and operated at USU. The data were acquired in pre-established mapping blocks over a 2 day period covering approximately 144 Km of the Mojave River floodplain and riparian zone, approximately 1500 meters in width. The multispectral imagery (green, red and near-infrared bands) was ortho-rectified using the Lidar point cloud data through a direct geo-referencing technique. Thermal Infrared imagery was rectified to the multispectral ortho-mosaics. The lidar point cloud data was classified to separate ground surface returns from vegetation returns as well as structures such as buildings, bridges etc. One-meter DEM's were produced from the surface returns along with vegetation canopy height also at 1-meter grids. Two surface energy balance models that use remote sensing inputs were applied to the high resolution imagery, namely the SEBAL and the Two Source Model. The model parameterizations were slightly modified to accept high resolution imagery (1-meter) as well as the lidar-based vegetation height product, which was used to estimate the aerodynamic roughness length. Both models produced very similar results in terms of latent heat fluxes (LE). Instantaneous LE values were extrapolated to daily evapotranspiration rates (ET) using the reference ET fraction, with data obtained from a local weather station. Seasonal rates were obtained by extrapolating the reference ET fraction according to the seasonal growth habits of the different species. Vegetation species distribution and area were obtained from classification of the multispectral imagery. Results indicate that cottonwood and salt cedar (tamarisk) had the highest evapotranspiration rates followed by mesophytes, arundo, mesquite and desert shrubs. This research showed that high-resolution airborne multispectral and thermal infrared imagery integrated with precise full-waveform lidar data can be used to estimate evapotranspiration and water use by riparian vegetation.