Comparison of energy balance modeling schemes using microwave‐derived soil moisture and radiometric surface temperature

Abstract

A Two‐Source (soil + vegetation) Energy Balance (TSEB) modeling scheme has been developed to use either microwave‐derived near‐surface soil moisture (TSEBSM) or radiometric surface temperature (TSEBTR) as the key remotely sensed surface boundary condition for computing spatially distributed heat fluxes. Output of the surface heat fluxes from both two‐source schemes have been validated using tower‐ and aircraft‐based flux observations. However, these observations rarely provide the necessary spatial information for evaluating heat flux patterns produced by spatially based models. By collecting microwave and radiometric surface temperature observations concurrently during the Southern Great Plains 1997 (SGP97) experiment conducted in Oklahoma, USA, heat flux estimates by the two modeling schemes were compared on a pixel‐by‐pixel basis. This provided a unique opportunity for evaluating the consistency in spatial patterns of the heat fluxes. Comparisons with radiometric surface temperature observations helped to elucidate factors contributing to discrepancies between TSEBSM and TSEBTR output, because the TSEBSM modeling scheme computes an effective surface temperature. Results from the heat flux comparisons and simulated versus observed surface temperatures suggested revisions to TSEBSM parameterizations are needed to better constrain flux predictions from the soil and vegetation. When the revisions are made, TSEBSM accommodates a wider range of environmental conditions. The revisions involve an adjustment to the soil evaporation algorithm for differential drying of the near‐surface soil layer and adopting the Priestley–Taylor coefficient estimated from the TSEBTR model. It was also found that areas with high fractional vegetative cover conditions, TSEBTR estimates of energy partitioning between sensible and latent heat flux at the soil surface (expressed in terms of the soil Bowen ratio, BOS), were uncorrelated to the remotely sensed near‐surface soil moisture. This contributed to inconsistencies in BOS patterns estimated by TSEBTR during a dry down period. A ∼20% change in the maximum fractional vegetation cover estimated using the remote‐sensing‐based algorithm is shown to dramatically impact BOS values estimated by TSEBTR for the densely vegetated areas while having little effect on TSEBSM‐derived values. This result suggests that under certain environmental conditions, energy balance partitioning at the soil surface over densely vegetated areas may be tenuous using the TSEBTR scheme.