Freshwater availability is a critical planetary boundary currently under threat from intensified climate change and agricultural over-exploitation. Agriculture accounts for over 70% of global water use, yet the ability to monitor crop water consumption accurately remains a significant challenge. Satellite remote sensing has become an essential tool for monitoring agricultural water use at regional to global scales, primarily through the estimation of evapotranspiration (ET). However, most operational ET studies rely on moderate-resolution satellite missions such as the Landsat program or Sentinel-2, which may limit the detection of short-term dynamics and field-scale variability in agricultural systems.

This thesis, conducted at KTH Royal Institute of Technology in collaboration with Planet Labs, evaluates the operational potential of three distinct evapotranspiration (ET) modeling frameworks: STIC (Surface Temperature Initiated Closure), PT-JPL (Priestley-Taylor, Jet Propulsion Laboratory), and GLEAM (Global Land Evaporation Amsterdam Model). The objective of this research is to assess how these widely used ET models perform when driven by high-frequency commercial satellite observations and to investigate whether combining their outputs can improve the reliability of agricultural water use estimation across diverse climatic conditions. By leveraging Planet's unique satellite constellation, this study explores the potential of high-temporal-resolution Earth observation data to enhance evapotranspiration monitoring at the field scale. This research benchmarks these models against in-situ AmeriFlux data across diverse climate zones in North America between 2020 and 2024.

Results indicate that the PT-JPL model demonstrated the highest overall consistency (R²: 0.64 and RMSE: 1.04 mm/day), driven by its robust eco-physiological scaling. While STIC proved highly accurate during peak biomass, but vulnerable to thermal inputs from bare soil during early growth. Conversely, GLEAM provided the most robust performances over the different climates, though sometimes underestimating the transpiration rates of mature crops.

A central contribution of this research is the development of a multi-model ensemble, which significantly improved validation metrics (R²: 0.68, RMSE: 0.96 mm/day) by mitigating individual model biases. The study further identifies climatic dependencies, showing higher accuracy in subtropical and continental climates compared to arid regions, where GLEAM emerged as the most resilient framework. By translating these biophysical fluxes into actionable Water Stress and Deficit Indices, this thesis demonstrates a clear bridge between satellite-based modeling and practical irrigation scheduling. Beyond model benchmarking, the results highlight the potential of high-frequency commercial satellite constellations to support operational agricultural water monitoring within the fields of geoinformatics and remote sensing. This research provides a foundation for scalable to similar climates, model-driven tools that can optimize agricultural water use, ensuring both economic viability for farmers and ecological sustainability in a water-constrained future.