Evaluation of soil and vegetation heat flux predictions using a simple two-source model with radiometric temperatures for partial canopy cover

A two-source model developed to use radiometric temperature observations for predicting component surface energy fluxes from soil and vegetation was evaluated with data from a row crop (cotton). The total or combined heat fluxes from the soil and vegetation agreed to within 20% of the observed values, on average. Component heat flux predictions from the soil and vegetation indicated that soil evaporation was generally higher than canopy transpiration. This result contradicts an earlier study which showed that soil evaporation was ∼1/3 of canopy transpiration rates with a significant source of sensible heat from the soil being advected to the canopy ( Kustas, 1990). Moreover, the modeled derived canopy temperatures were ∼6 K higher and soil temperatures were ∼4 K lower than the radiometric temperature observations. In order to obtain more physically realistic soil and vegetation component heat fluxes and better agreement between the predicted and observed soil and canopy temperatures, two model parameterizations required modification. One adjustment was to the magnitude of the Priestley–Taylor coefficient α PT used in estimating canopy transpiration. The magnitude of α PT was increased by ≈50% from its `universal constant' α PT ∼ 1.3 to α PT ∼ 2. The other modification was to the free convective velocity, U CV, defined as constant in the original formulation for estimating soil resistance to sensible heat flux transfer, R S. The new formulation is based on the recent experimental results from Kondo and Ishida (1997)who found that U CV ∝ Δ T 1/3 where Δ T is the surface–air temperature difference. Both of these modifications are shown to be supported by observations from the literature and therefore are not considered merely model `tuning'. Furthermore, component heat fluxes predicted by the model using canopy and soil radiometric temperature observations support the higher α PT value and new free convective formulation for estimating R S. Two other changes to model algorithms are described which are relevant to all dual-source modeling schemes. One is replacing the commonly used Beer's law type expression for estimating the divergence of net radiation in partial canopy covered surfaces with a more physically-based algorithm. The other is a simple method to address the effects of clumped vegetation (common in row crops and sparse canopies) on radiation divergence and wind speed inside the canopy layer.