Four upscaling methods for estimating daytime actual evapotranspiration (ET)from single time-of-day snapshots, as commonly retrieved using remotesensing, were compared. These methods assume self-preservation of the ratiobetween ET and a given reference variable over the daytime hours. Theanalysis was performed using eddy covariance data collected at 12 AmeriFluxtowers, sampling a fairly wide range in climatic and land cover conditions.The choice of energy budget closure method significantly impactedperformance using different scaling methodologies. Therefore, a statisticalevaluation approach was adopted to better account for the inherentuncertainty in ET fluxes using eddy covariance technique. Overall, thisapproach suggested that at-surface solar radiation was the most robustreference variable amongst those tested, due to high accuracy of upscaledfluxes and absence of systematic biases. Top-of-atmosphere irradiance wasalso tested and proved to be reliable under near clear-sky conditions, buttended to overestimate the observed daytime ET during cloudy days. Use ofreference ET as a scaling flux yielded higher bias than the solar radiationmethod, although resulting errors showed similar lack of seasonaldependence. Finally, the commonly used evaporative fraction method yieldedsatisfactory results only in summer months, July and August, and tended tounderestimate the observations in the fall/winter seasons from November toJanuary at the flux sites studied. In general, the proposed methodologyclearly showed the added value of an intercomparison of different upscalingmethods under scenarios that account for the uncertainty in eddy covarianceflux measurements due to closure errors.
展开▼