Two energy balance closure approaches: applications and comparisons over an oasis-desert ecotone

Abstract

Studies of energy balance that rely on eddy covariance (EC) are always challenged by energy balance closure, which is mainly caused by the underestimations of latent heat flux (LE) and sensible heat flux (Hs). The Bowen ratio (BR) and energy balance residual (ER) approaches are two widely-used methods to correct the LE. A comprehensive comparison of those two approaches in different land-use types is essential to accurately correcting the LE and thus improving the EC experiments. In this study, two energy balance approaches (i.e., BR and ER) were compared to correct the LE measured at six EC sites (i.e., three vegetated, one mixed and two non-vegetated sites) in an oasis-desert ecotone of the Heihe River Basin, China. The influences of meteorological factors on those two approaches were also quantitatively assessed. Our results demonstrated that the average energy closure ratio ((LE+Hs)/(Rn–Gs); where Rn is the surface net radiation and Gs is the surface soil heat flux) was approximately close to 1.0 at wetland, maize and village sites, but far from 1.0 at orchard, Gobi and desert sites, indicating a significant energy imbalance at those three latter sites. After the corrections of BR and ER approaches that took into account of soil heat storage, the corrected LE was considerably larger than the EC-measured LE at five of six EC sites with an exception at Gobi site. The BR and ER approaches yielded approximately similar corrected LE at vegetated and mixed sites, but they generated dissimilar results at non-vegetated sites, especially at non-vegetated sites with low relative humidity, strong wind, and large surface-air temperature difference. Our findings provide insight into the applicability of BR and ER approaches to correcting EC-based LE measurements in different land-use types. We recommend that the BR-corrected and ER-corrected LE could be seriously reconsidered as validation references in dry and windy areas.