An independent framework-based evapotranspiration model (IFEM) for dual-source: From field to regional scale

The triangle/trapezoid framework model (TFM) is a widely used approach for mapping regional evapotranspiration (ET) based on thermal infrared remote sensing imagery. However, the homogeneity assumption for meteorological forcing and surface properties involved in TFM contradicts the characteristics of heterogeneous regions, leading to high uncertainty at regional scale, especially under energy-limited or strong advection conditions. Therefore, this study examined the linear relationship between soil moisture availability (Lambda SM), evapotranspiration efficiency (Lambda ET) and radiation surface temperature (Trad), and developed an Independent Framework-based dual-source ET Model (IFEM) for heterogeneous regions. By fixing two key meteorological forcing variables, air temperature (Ta) and vapor pressure deficit (VPD), and three surface characteristic parameters, aerodynamic resistance (ra), albedo (alpha) and the ratio of soil heat flux to net radiation (Gamma), the IFEM constructs a uniform fractional vegetation cover (fc)- Trad space. In this uniform fc-Trad space, isolines of Lambda SM and Lambda ET are aligned with those of Trad, and an independent trapezoid framework is constructed for each pixel of RS images. It avoids the systematic bias caused by the distorted trapezoid space involved in traditional TFMs. The proposed model was applied in a hydrologically semi-enclosed irrigation area, the Hetao Irrigation District, with Landsat and MODIS datasets from 2016 to 2020 to evaluate its performance at the field and region scales. Firstly, the IFEM was capable of retrieving latent heat with a mean absolute error (MAE) of 34.43 W m- 2 and a mean absolute percentage error (MAPE) of 8.68%, compared with the measurements from flux towers. Secondly, the IFEM reasonably separated the soil (MAPE = 21.33%) and canopy (MAPE = 14.33%) components from ET, compared with the soil evaporation observed by micro-lysimeters and the plant transpiration calculated by sap velocity upscaling technology. Thirdly, the IFEM showed high accuracy in estimating annual total ET with a percentage error range within +/- 3.0%, compared against the regional water balance calculations. Overall, the IFEM demonstrated high robustness and accuracy for field-scale and regional-scale applications, validating its potential as a reliable method for mapping regional ET in heterogeneous regions.