On the representation of small-scale soil biophysical features for large-scale applications

In light of the complex interactions of multiple biotic and abiotic processes acting at different spatio-temporal scales and the high spatial variability in their biophysical features, our ability to describe catchment dynamics is still limited. Such a limitation stems, on one side, from the mismatch of scales between key small-scale processes and their current parameterization and the large scales of many modeling applications, and, on the other, from the challenge of considering the spatial configurations in an explicit manner. As a motivating example, we will discuss here the issue of representing the effect of biologically-induced soil structure in the parameterization of soil hydraulic properties (SHPs) for large scale applications. Currently available parameterizations based on soil pedotransfer functions (PTFs) do not account for the effects of soil structure, thus limiting their applicability in vegetated areas in which macropores are expected to significantly increase soil saturated hydraulic conductivity. Considering the strong links between vegetation and soil structure, we propose a systematic approach for incorporating structural effects on PTF-derived SHPs. We will show that, under certain soil and climatic conditions, small scale soil structure features prominently alter the hydrologic response emerging at larger scales and that upscaled parameterizations must explicitly consider the spatial variability of soil and vegetation attributes. Lastly, opportunities in the representation of multiple small‐scale ecohydrological processes for regional and global applications will be discussed. Progress on this front is key for establishing more complete causal links between landscape attributes and heterogeneities in physical properties, thus providing a mechanistic strategy for model parameterization and process description across scales and a path forward for more reliable large-scale modeling under future scenarios.