Impact of land-surface variability on turbulent surface fluxes in multi-scale atmospheric simulations

Using multiscale atmospheric simulations with the WRF model, coupled to the unified NOAH LSM - UCM, we investigate land-atmosphere exchanges over heterogeneous and complex terrain and their sensitivity to surface variability, grid resolution, and turbulence intensity. These numerical simulations have horizontal grid resolutions ranging from 12.15 km for the coarsest mesoscale model down to 50 m for finest LES model and from 10km to 80 m. We simulate different meteorological scenarios and compare them to local measurements of an eddy covariance station and to measurements from New Jersey mesonet stations network. The study area is a typical suburban zone composed of a variety of urban surfaces (roads, houses, commercial development, etc) and a patchwork of natural areas (lake, forests, crops, grassland, etc.) in a 25 km² domain. The results indicate that local turbulent fluxes over grid cells with identical land-cover types vary depending on the immediate surrounding of the area, and hence on upstream flow and turbulence conditions. Using estimates of the variability length scale of surface parameters and fluxes based on the autocorrelation function, we investigate how local and domain-averaged turbulent fluxes depend on the composition and spatial distribution of the land cover.