Sensitivity of roughness length for heat transport (zoh) on evapotranspiration derived from SEBAL

Thermal infrared remote sensing has greatly contributed to the development and improvement of remote sensing based evapotranspiration (RS-ET) mapping algorithms. The radiometric temperature derived from the thermal sensors were inherently different than the aerodynamic temperature required for solving the bulk formulation of sensible heat (H) based on the Monin-Obukhov similarity (MOS); this posed a critical problem. The TSM (Two Source Model), SEBS (Surface Energy Balance System) and SEBAL (Surface Energy Balance Algorithm) forms the three most widely applied RS-ET algorithm's differing in their conceptualization and parameterization of the soil-canopy-air heat exchange mechanism addressing the issue arising from aerodynamic-radiometric temperature differences. The scalar roughness length zoh, representing heat transport and described by the dimensionless parameter kB-1, was used as a correction factor to accommodate the discrepancy between radiometric and aerodynamic temperatures. In this study we looked into the sensitivity of zoh on the ET estimates using the SEBAL approach. ET estimates from four approaches namely, (i) zoh derived from constant kB-1 of 2.3, (ii) zoh=0.1, (iii) zoh=0.01, and (iv) zoh from kB-1 parameterization, were compared. SEBAL was executed for 10 high resolution airborne images acquired during BEAREX07-08 (Bushland Evapotranspiration and Agricultural Remote Sensing Experiment) field campaign and validated against large precision weighing lysimeters installed on two irrigated and two dryland fields. Statistical tests revealed no significant differences between the first three approaches, however, the fourth approach of kB-1 parameterization produced significantly different results. Model performance evaluation for all the components of the energy balance was conducted. Percent root mean square error (%RMSE) for instantaneous ET estimates from the four approaches were 33.7, 26.9, 27.7 and 23.2 respectively. Evaluation of the SEBAL performances on irrigated and dryland fields separately revealed that all four approaches showed good agreement with the irrigated field observations, however, large deviations were observed for dryland fields. Model performance for the dryland fields from first three approach produced NSE of -0.91, -0.31, and -0.39 respectively, indicating unacceptable model performance, however the fourth approach yielded positive NSE of 0.11 indicating acceptable level of performance. Overall, results indicated that a physically based excess resistance to heat transfer parameter (kB-1) incorporated into SEBAL significantly improves the ET estimates.