Evapotranspiration is fundamental in the hydrological cycle and about 70% of the water that precipitates onto the surface of the Earth returns to the atmosphere through the process. Moreover, measuring evapotranspiration allows the estimation of the capacity of reservoirs or aquifers, the determination of watershed yields, the elaboration of irrigation projects aimed at agriculture, and the control of water availability to supply urban populations, for instance. During the 1990s and 2000s, remote sensing-based evapotranspiration algorithms, using information from visible, near-infrared, and thermal infrared bands, were developed for subregional applications, with a focus on irrigation or water resources management, and compared with the flux tower data, especially in North America and Europe. The aim is to learn how much water is used by cultivated or native vegetation and estimate the capacity of reservoirs and aquifers. "This allows the elaboration of more precise irrigation projects aimed at agriculture", explains the Embrapa Environment researcher Osvaldo Cabral, one of the authors of the work. With the measurements from different types of vegetation, using models, satellite images, and information on rainfall and temperature, it is possible to not only obtain irrigation estimates, but also learn the watershed yield and even their contribution to the supply of urban areas. According to Cabral, "the idea to use accurate evapotranspiration information for each surface type also touches on the issue of climate change. If we have an evapotranspiration forecast for the next 50 years, with these models we can estimate such effects", Cabral explains. In South America, a team of scientists used meteorological data from 25 flux towers to compare with the results of four remote sensing models: Priestley-Taylor Jet Propulsion Laboratory (PT-JPL), Global Land Evaporation, Amsterdam Model (GLEAM), Penman-Monteith Mu model (PM-MOD), and Penman-Monteith Nagler model (PME-VI), and the results showed that, in general, evapotranspiration can be satisfactorily predicted in all of them. South America, with 12% of the planet's land area, spans two hemispheres, and four major climate zones, from near the equator to sub-Antarctic regions, which makes it a geographically unique continent. Biomes in this continent range from tropical to deciduous forests, and contain ecoregions with high sensitivity to variability in water (e.g., the Caatinga and Humid Pampas) and energy availability (e.g., the Amazon, Valdivian temperate, and Magellanic subpolar forests). Moreover, five out of six of the terrestrial biomes not included in satellite-based evapotranspiration algorithm evaluations at a global scale are found in South America "We tried to answer whether such models can be applied consistently to reliably capture evapotranspiration in the continent, and if they are affected by climate, land cover type, or biome", the researcher restates from the paper. Our analysis provides essential information to identify model strengths and limitations across South America, allowing users to identify which model is more suitable for them. Knowing under which circumstances (e.g. land use or climate) each model is more reliable is necessary to address remaining research and applied science gaps concerning evapotranspiration at a local, regional and global level", Cabral underscores from the paper. The study showed that the model skill seems to be unrelated to land cover. Considering the good results obtained for 50% of the towers and the fact that,c ompared to the other models, PM-VI has a much simpler implementation, this model does have potential as long as sufficient data are available for calibration or, at least, validation. However, the need for local calibration is a hurdle for its implementation for most regions that are unsampled; therefore future studies are necessary to investigate which factors are most relevant in the determination of the model fitting coefficients, and to provide distributed reference values for its coefficients (e.g. based on land use dynamics). The paper was published on Water Resources Research.