On comparison of modeled surface flux variations to aircraft observations

Evaluation of models of air–surface exchange is facilitated by an accurate match of areas simulated with those seen by micrometeorological flux measurements. Here, spatial variations in fluxes estimated with the parameterized subgrid-scale surface (PASS) flux model were compared to flux variations seen aboard aircraft above the Walnut River Watershed (WRW) in Kansas. Despite interference by atmospheric eddies, the areas where the modeled sensible and latent heat fluxes were most highly correlated with the aircraft flux estimates were upwind of the flight segments. To assess whether applying a footprint function to the surface values would improve the model evaluation, a two-dimensional correlation distribution was used to identify the locations and relative importance of contributing modeled surface pixels upwind of each segment of the flight path. The agreement between modeled surface fluxes and aircraft measurements was improved when upwind fluxes were weighted with an optimized footprint parameter φ, which can be estimated from wind profiler data and surface eddy covariance. Variations of the flight-observed flux were consistently greater than those modeled at the surface, perhaps because of the smoothing effect of using 1 km pixels in the model. In addition, limited flight legs prevented sufficient filtering of the effects of atmospheric convection, possibly accounting for some of the more prominent changes in fluxes measured along the flight paths.