SURFACE-ENERGY BALANCE ESTIMATES AT LOCAL AND REGIONAL SCALES USING OPTICAL REMOTE-SENSING FROM AN AIRCRAFT PLATFORM AND ATMOSPHERIC DATA COLLECTED OVER SEMIARID RANGELANDS

Remotely sensed data in the visible, near-infrared, and thermal-infrar ed wave bands were collected from a low-flying aircraft during the Mon soon '90 field experiment. Monsoon '90 was a multidisciplinary experim ent conducted in a semiarid watershed. It had as one of its objectives the quantification of hydrometeorological fluxes during the ''monsoon '' or wet season. The remote sensing observations along with micromete orological and atmospheric boundary layer (ABL) data were used to comp ute the surface energy balance over a range of spatial scales. The pro cedure involved averaging multiple pixels along transects flown over t he meteorological and flux (METFLUX) stations. Average values of the s pectral reflectance and thermal-infrared temperatures were computed fo r pixels of order 10(-1) to 10(1) km in length and were used with atmo spheric data for evaluating net radiation (R(n)), soil heat flux (G), and sensible (H) and latent (LE) heat fluxes at these same length scal es. The model employs a single-layer resistance approach for estimatin g H that requires wind speed and air temperature in the ABL and a remo tely sensed surface temperature. The values of R(n) and G are estimate d from remote sensing information together with near-surface observati ons of air temperature, relative humidity, and solar radiation. Finall y, LE is solved as the residual term in the surface energy balance equ ation. Model calculations were compared to measurements from the METFL UX network for three days having different environmental conditions. A verage percent differences for the three days between model and the ME TFLUX estimates of the local fluxes were about 5% for R(n), 20% for G and H, and 15% for LE. Larger differences occurred during partly cloud y conditions because of errors in interpreting the remote sensing data and the higher spatial and temporal variation in the energy fluxes. M inor variations in modeled energy fluxes were observed when the pixel size representing the remote sensing inputs changed from 0.2 to 2 km. Regional scale estimates of the surface energy balance using bulk ABL properties for the model parameters and input variables and the 10-km pixel data differed from the METFLUX network averages by about 4% for R(n), 10% for G and H, and 15% for LE. Model sensitivity in calculatin g the turbulent fluxes H and LE to possible variations in key model pa rameters (i.e., the roughness lengths for heat and momentum) was found to be fairly significant. Therefore the reliability of the methods fo r estimating key model parameters and potential errors needs further t esting over different ecosystems and environmental conditions.