Determining adequate averaging periods and reference coordinates for eddy covariance measurements of surface heat and water vapor fluxes over mountainous terrain

Abstract

Two coordinate rotation approaches (double and planar-fit rotations) and no rotation, in association with averaging periods of 15 - 480 min, were applied to compute surface heat and water vapor fluxes using the eddy covariance approach. Measurements were conducted in an experimental watershed, the Lien-Hua-Chih (LHC) watershed, located in central Taiwan. For no rotation and double rotation approaches, an adequate averaging period of 15 or 30 min was suggested for better energy closure and small variations on energy closure fractions. For the planar-fit rotation approach, an adequate averaging period of 60 or 120 min was recommended, and a typical averaging period of 30 min is not superior to that of 60 or 120 min in terms of better energy closure and small variations on energy closure fractions. The Ogive function analysis revealed that the energy closure was improved with the increase of averaging time by capturing sensible heat fluxes at low-frequency ranges during certain midday hours at LHC site. Seasonal variations of daily energy closure fractions, high in dry season and low in wet season, were found to be associated with the surface dryness and strength of turbulent development. The mismatching of flux footprint areas among flux sensors was suggested as the cause of larger CF variations during the dry seasons as that indicated by the footprint analysis showing scattered source areas. During the wet season, the underestimation of turbulent fluxes by EC observations at the LHC site was attributed to weak turbulence developments as the source area identified by the footprint analysis was closer to the flux tower than those scattered in dry season.