SURFACE FLUX MEASUREMENT AND MODELING AT A SEMIARID SONORAN DESERT SITE

Continuous measurements of standard meteorological variables using an automatic weather station and intermittent measurements of the surface energy balance, carbon dioxide flux, and momentum flux using Bowen ra tio, eddy covariance, and sigma-T instrumentation were made for 13 mon ths at a semi-arid Sonoran Desert site just west of Tucson, AZ. Weathe r observations demonstrate typical semi-arid Sonoran desert conditions , with frequent clear skies, high radiation, a large seasonal and diur nal temperature range, low relative humidity, and intermittent precipi tation mainly of convective origin during a summer monsoon season. The substantial observational problems associated with surface flux measu rements in this environment are reported. Comparisons between measured fluxes made simultaneously with different instrumental systems show a cceptable agreement. Most of the incoming radiant energy leaves as sen sible heat, and latent heat fluxes are always low, but transpiration i s enhanced for about 10 days after rain. To investigate the influence of Crassulacean Acid Metabolism plants on carbon dioxide flux, measure ments were sustained through the night. Carbon dioxide uptake is low, typically with peak daytime uptake in the order 0.25-1.0 mu mol m(-2) s(-1) for the period for which data are available, and some carbon upt ake persists even at night. The observations were used to validate and calibrate the surface energy balance simulated by the Biosphere-Atmos phere Transfer Scheme. Using the default 'semi-desert' soil and vegeta tion parameters specified in the National Center for Atmospheric Resea rch Community Climate Model Version 2 resulted in a poor simulation of observations. However, using a set of site-specific parameters, inclu ding on-site observations to specify more realistic soil and vegetatio n characteristics, and optimized minimum surface resistance and plant wilting parameters, resulted in a substantial improvement in model per formance. The site-specific parameters reflect the fact that the veget ation fraction is greater than assumed in the default parameter set, t hat leaf area index and minimum stomatal resistance are less, soils at the study site contain more clay, but that the plants' wilting point is lower than this clay fraction would imply. The modified, site-speci fic parameters mon accurately describe the conservative character of t he semi-desert vegetation and the moderate nature of its response to t he seasonal water cycle.