Quantifying net ecosystem exchange by multilevel ecophysiological and turbulent transport models

To quantify the interplay between scalar sources and sinks (S_c) and net ecosystem exchange (NEE), “forward” and “inverse” approaches have been proposed. The canonical form of forward approaches is a one-dimensional ecophysiological-radiative transfer scheme coupled to turbulent transport theory. In contrast, inverse approaches strictly rely on turbulent transport theory and mean scalar concentration as their primary input to infer S_c and NEE. While the formulation of both approaches have evolved over the past decade, no systematic comparison between them was undertaken for the same data set, and over a wide range of atmospheric conditions. Our objective is to compare the predicted S_c and NEE from these two approaches with eddy-covariance measurements. The results show that the forward method outperformed all three inverse methods for unstable and neutral conditions on short time scales (30 min) but yielded comparable results at longer time scales. Poor agreement was obtained under stable conditions for all models. Hence, for modeling event-based flux variations, forward models are preferred. Since the forward method requires detailed knowledge of ecophysiological, drag, radiative transfer and other canopy attributes, all of which are difficult to obtain on a routine basis, a symbiotic use of forward and inverse approaches is most advantageous.