Alternative methods to predict actual evapotranspiration illustrate the importance of accounting for phenology – Part 2: The event driven phenology model

[摘要] Evapotranspiration (ET) flux constitutes a major component of both the waterand energy balances at the land surface. Among the many factors that controlevapotranspiration, phenology poses a major source of uncertainty inattempts to predict ET. Contemporary approaches to ET modeling and monitoringfrequently summarize the complexity of the seasonal development ofvegetation cover into static phenological trajectories (or climatologies)that lack sensitivity to changing environmental conditions. The Event DrivenPhenology Model (EDPM) offers an alternative, interactive approach torepresenting phenology. This study presents the results of an experimentdesigned to illustrate the differences in ET arising from various techniquesused to mimic phenology in models of land surface processes. The experimentcompares and contrasts two realizations of static phenologies derived fromlong-term satellite observations of the Normalized Difference VegetationIndex (NDVI) against canopy trajectories produced by the interactive EDPMtrained on flux tower observations. The assessment was carried out throughvalidation of predicted ET against records collected by flux towerinstruments. The VegET model (Senay, 2008) was used as a framework toestimate daily actual evapotranspiration and supplied with seasonal canopytrajectories produced by the EDPM and traditional techniques. Theinteractive approach presented the following advantages over phenologymodeled with static climatologies: (a) lower prediction bias in crops; (b) smaller root mean square error in daily ET – 0.5 mm per day on average;(c) stable level of errors throughout the season similar among different landcover types and locations; and (d) better estimation of season duration andtotal seasonal ET.

[学科分类] 地球化学与岩石