Introduction and Goal Sustainable water management in arid and semi-arid regions requires accurate estimation of water balance components, especially actual evapotranspiration (ETa). ETa is critical for understanding water consumption by vegetation, particularly in irrigated agriculture. Due to limitations of field-based ETa measurements, satellite remote sensing models such as METRIC, SEBAL, and SEBS have gained popularity in recent years. This study aims to evaluate and compare the performance of these three models in estimating ETa over the Garbayegan Plain, Fars Province, Iran.For all three models (METRIC, SEBS and SEBAL), from the data of Landsat 8 and 9 satellite images including 9 reflective bands and 2 thermal bands with spatial resolution of 30 meters for reflective bands and 100 meters for thermal, meteorological data, image processing and geometric, radiometric and atmospheric correction were performed. Reflectivity and surface temperature data were extracted. The METRIC model works based on the energy balance of the earth's surface and the internal calibration method to calculate evapotranspiration-transpiration, the main steps of which include calculating net radiation (Rn), estimating soil heat flux, SEBS model is a physical model that estimates evapotranspiration-transpiration using energy balance equations and surface parameters.
Materials and Methods The study was conducted in Garbayegan Plain, located in southeastern Fars Province with an arid climate and an average annual rainfall of about 230 mm. Landsat 8 and 9 satellite imagery from the 2022–2023 agricultural season was used. After performing geometric, radiometric, and atmospheric corrections, the SEBAL, METRIC, and SEBS models were applied based on the surface energy balance equation. The models estimated net radiation, soil heat flux, sensible heat flux, and latent heat flux to calculate instantaneous and daily ETa. Field data, soil water balance, and irrigation records were used for validation. NDVI indices were applied to extract agricultural pixels and assess crop water use.
Results and Discussion The METRIC model demonstrated the highest accuracy in ETa estimation (R² = 0.87, RMSE = 0.85 mm/day), primarily due to its use of internal calibration via anchor pixels. The SEBS model also yielded acceptable performance (R² = 0.79), while SEBAL, despite its structural similarity to METRIC, showed lower accuracy (R² = 0.74) due to the absence of local calibration. SEBAL, however, provided smoother spatial maps of ETa. Model selection for ETa estimation depends on the intended application, availability of field data, and desired numerical or spatial accuracy. METRIC, with its use of reference pixel calibration, is more suitable for farm-scale and regional assessments. In contrast, SEBAL and SEBS can offer preliminary spatial insights under limited field data conditions. The superior performance of METRIC highlights the importance of integrating ground observations with satellite imagery. Conclusion The findings confirm that the METRIC model is a reliable tool for accurately estimating ETa in agricultural lands of the Garbayegan Plain. It can support water use monitoring, irrigation planning, and productivity assessments. Furthermore, NDVI-based pixel classification significantly improved estimation of agricultural water consumption. Future studies are recommended to utilize high-temporal-resolution satellites (e.g., Sentinel-2) and incorporate more field observations for model calibration and validation.
Keywords: Meteorological data, remote sensing, field measurements, soil water balance, pixels
Article Type: Research Article Acknowledgement This is a sincere expression of gratitude to all colleagues and experts in the remote sensing and water sciences fields who have assisted in data collection, satellite imagery processing, and result analysis for this research. We also appreciate the financial support and facilities provided by the Soil and Watershed Conservation Research Institute and the Agricultural and Natural Resources Research and Training Center of Fars. Without their cooperation and support, this research would not have reached this level of accuracy and quality. Conflicts of interest The authors declare that there is no conflict of interest regarding the authorship or publication of this article. |