Energy balance closure in the Southwest Amazon forest site—a statistical approach

Abstract

Energy balance closure is the main indicator of the quality of energy flux measurements obtained by the eddy covariance technique. Many researchers use a simple linear regression model between energy balance components to evaluate closure. However, these studies typically fail to verify the appropriateness of the statistical assumptions of regression analysis, which can lead to erroneous conclusions if the model is not satisfactory. Thus, the aim of this study was to calibrate and validate simple and robust with bootstrap and cross-validation linear regression models to verify the efficiency of energy balance closure in the Amazon ecosystem. Measurements of net radiation and latent, sensible, and ground heat fluxes were made from January to December 2008 in a tropical rain forest area in South West Amazonia in an experimental site belonging to the network of flux towers of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA). The results demonstrated that simple linear regression models are not appropriate for analyzing energy balance closure. However, robust linear regression models with bootstrap and cross-validation improved the fit to the data. Despite the better fit, there was an increase in energy balance closure residuals suggesting that the eddy covariance technique is underestimating the values of energy fluxes in Amazon forest areas more than what were reported in previous researches.