Modeling the impact of averaging on aggregation of surface fluxes over BOREAS

Abstract

Surface meteorological observations, soil and vegetation parameters, and satellite-retrieved surface radiation fluxes are used as input to a biosphere-atmosphere flux exchange model to estimate the fine-scale surface fluxes over a 75,000 km2 region within the large-scale BOREAS study area under cloud-free conditions. The size of the study area chosen for the analysis is synonymous with the characteristic grid size of a modern global climate model. The variables used to force the model are area-averaged and applied in such a fashion to assess the errors in area-wide fluxes that would arise in flux aggregation schemes based on the area-averaged inputs. In the absence of large spatial variance of downwelling radiative fluxes, characteristic of the clear sky case under study, progressive area averaging of the input parameters indicates that the most important quantities influencing the aggregated fluxes are root layer depth, soil moisture, and soil type, all of which are interrelated in the formulation of the transpiration capacity. Meteorological variance over the site and variability in a number of other parameters relating to vegetation characteristics and initial soil temperatures are not critical to the aggregation process. Notably, major errors in aggregated surface fluxes arise due to area-averaging soil moisture alone, even when all other parameters are not area-averaged and allowed to maintain their spatial variability. The errors materialize as significant phase shifts in the frequency distributions of the fine-scale surface fluxes calculated after input averaging. Application of a soil mosaic averaging technique, based on soil type and the soil moisture associated with each soil type, restores the characteristic mean values of the baseline area-averaged fluxes. This procedure also simplifies the surface flux distributions to dependence on the percentages of grid area covered by each soil type but without altering the basic multimodal character of the baseline distributions. It is demonstrated and explained why such a soil-type designation over the area forms the kernels for the surface flux distribution spectrums, around which the other input parameters act to disperse the frequency distribution properties of the aggregated surface fluxes.