INTERPRETATION OF SURFACE FLUX MEASUREMENTS IN HETEROGENEOUS TERRAIN DURING THE MONSOON 90 EXPERIMENT

A network of 9-m-tall surface flux measurement stations were deployed at eight sparsely vegetated sites during the Monsoon '90 experiment to measure net radiation, Q, soil heat flux, G, sensible heat flux, H (u sing eddy correlation), and latent heat flux, lambdaE (using the energ y balance equation). At four of these sites, 2-m-tall eddy correlation systems were used to measure all four fluxes directly. Also a 2-m-tal l Bowen ratio system was deployed at one site. Magnitudes of the energ y balance closure (Q + G + H + lambdaE) increased as the complexity of terrain increased. The daytime Bowen ratio decreased from about 10 be fore the monsoon season to about 0.3 during the monsoons. Source areas of the measurements are developed and compared to scales of heterogen eity arising from the sparse vegetation and the topography. There was very good agreement among simultaneous measurements of Q with the same model sensor at different heights (representing different source area s), but poor agreement among different brands of sensors. Comparisons of simultaneous measurements of G suggest that because of the extremel y small source area, extreme care in sensor deployment is necessary fo r accurate measurement in sparse canopies. A recently published model to estimate fetch is used to interpret measurements of H at the 2 m an d 9 m heights. Three sites were characterized by undulating topography , with ridgetops separated by about 200-600 m. At these sites, sensors were located on ridgetops, and the 9-m fetch included the adjacent va lley, whereas the 2-m fetch was limited to the immediate ridgetop and hillside. Before the monsoons began, vegetation was mostly dormant, th e watershed was uniformly hot and dry, and the two measurements of H w ere in close agreement. After the monsoons began and vegetation fully matured, the 2-m measurements of H were significantly greater than the 9-m measurements, presumably because the vegetation in the valleys wa s denser and cooler than on the ridgetops and hillsides. At one lowlan d site with little topographic relief, the vegetation was more uniform , and the two measurements of H were in close agreement during peak ve getation. Values of lambdaE could only be compared at two sites, but t he 9-m values were greater than the 2-m values, suggesting lambdaE fro m the dense vegetation in the valleys was greater than elsewhere.