European Journal of Sustainable Agroecosystems

Background: Accurate crop evapotranspiration (ETc) measurement is crucial for sustainable water irrigation management, food safety, and hydrological planning due to the impact of rising climatic variability and water scarcity. Traditional Penman-Monteith method and lysimeter methods results in only point-source results and do not assess the crop's water consumption type across the region. The combination of remote sensing (RS) technology and Geographic Information Systems (GIS) stands out and has become the leading technique in spatial evapotranspiration measurement during recent decades.
Aim: The aim of this review is to gather issues researched by people in the field of RS combined with GIS methods and compare the surface energy balance, vegetation index, and machine learning methods, as well as highlight what questions are still not answered and what the latest achievements are.
Sources: 1, 842 records were found through a thorough search in Scopus, Web of Science, PubMed, ScienceDirect, and Google Scholar, of which 108 studies fulfilled the eligibility criteria and 34 studies were selected for corresponding thematic and comparative synthesis based on PRISMA selection.
Results: Energy balance modelling methods (SEBAL, METRIC, SSEBop, SEBS, TSEB) are amongst the most widely tested methods with deviations ranging from 10% to 20% compared to measured data from eddy covariance and Bowen ratio methods. However, machine learning and hybrid physics-informed modelling methods have started to show similar or even better results compared to classical energy balance methods, as well as a reduced need for meteorological data.
GIS provides the necessary spatial framework for interpolation of values, zoning, and delivery of decision support, but there is still a lack of compatibility between the raster-based ET data and vector-based irrigation data.
Gap in research: Existing gaps are still there; including repeated thermal sensor visitation and cloudy conditions, subjective hot/cold pixel identification, lack of validation in smallholder and diverse cropping system in the southern hemisphere, and limited use of WebGIS-based decision support tools.
Conclusion: Although the integration of RS and GIS has developed into a workable framework for crop water accounting, success in the future relies on high-resolution thermal satellite networks, physics-based machine learning, cloud computing for geoprocessing and improved validation systems in areas with limited data.