An essential prerequisite to justify upscaling from sites to the globe requires homogeneous site-level characteristics (i.e. soil and vegetation properties). However, site-level characteristics are in practice rarely completely homogeneous, which might introduce variations in the magnitude of observed fluxes depending on wind wind direction (Griebel et al., 2016). Thus, some portion of the variability between interannual carbon budgets may purely be attributable to inconsistencies in the prevailing wind directions between years, which cause systematic differences in the area that is being sampled.
As no standard method exists to quantify site homogeneity, we developed a Heterogeneity Index (HI) based on the range of EVI values (Enhanced Vegetation Index) across individual 30 x 30 m pixels to assess the spatio-temporal variation of vegetation characteristics around each flux tower. Applying this to the FLUXNET2015 dataset identified that spatial homogeneity of site characteristics does not hold true for many flux tower locations.
Removing this unaccounted variability by attributing the contributions of surface heterogeneities to flux tower estimates would allow us to establish more reliable impacts of a changing climate on ecosystem properties, which will ultimately result in more accurate predictions of global productivity in future climates.