Energy, water, and carbon fluxes in a loblolly pine stand: Results from uniform and gappy canopy models with comparisons to eddy flux data

Abstract

This study investigates the impacts of canopy structure specification on modeling net radiation (R-n), latent heat flux (LE) and net photosynthesis (A(n)) by coupling two contrasting radiation transfer models with a two-leaf photosynthesis model for a maturing loblolly pine stand near Durham, North Carolina, USA. The first radiation transfer model is based on a uniform canopy representation (UCR) that assumes leaves are randomly distributed within the canopy, and the second radiation transfer model is based on a gappy canopy representation (GCR) in which leaves are clumped into individual crowns, thereby forming gaps between the crowns. To isolate the effects of canopy structure on model results, we used identical model parameters taken from the literature for both models. Canopy structure has great impact on energy distribution between the canopy and the forest floor. Comparing the model results, UCR produced lower R-n, higher LE and higher A(n) than GCR. UCR intercepted more shortwave radiation inside the canopy, thus producing less radiation absorption on the forest floor and in turn lower R-n. There is a higher degree of nonlinearity between A(n) estimated by UCR and by GCR than for LE. Most of the difference for LE and A(n) between UCR and GCR occurred around noon, when gaps between crowns can be seen from the direction of the incident sunbeam. Comparing with eddy-covariance measurements in the same loblolly pine stand from May to September 2001, based on several measures GCR provided more accurate estimates for R-n, LE and A(n) than UCR. The improvements when using GCR were much clearer when comparing the daytime trend of LE and A(n) for the growing season. Sensitivity analysis showed that UCR produces higher LE and A(n) estimates than GCR for canopy cover ranging from 0.2 to 0.8. There is a high degree of nonlinearity in the relationship between UCR estimates for A(n) and those of GCR, particularly when canopy cover is low, and suggests that simple scaling of UCR parameters cannot compensate for differences between the two models. LE from UCR and GCR is also nonlinearly related when canopy cover is low, but the nonlinearity quickly disappears as canopy cover increases, such that LE from UCR and GCR are linearly related and the relationship becomes stronger as canopy cover increases. These results suggest the uniform canopy assumption can lead to underestimation of R-n, and overestimation of LE and A(n). Given the potential in mapping regional scale forest canopy structure with high spatial resolution optical and Lidar remote sensing plotforms, it is possible to use GCR for up-scaling ecosystem processes from flux tower measurements to heterogeneous landscapes, provided the heterogeneity is not too extreme to modify the flow dynamics.