The Caatinga biome, located in the northeastern region of Brazil, is the most populated dryland region on the planet and extremely vulnerable to land degradation due to climatological and anthropogenic factors. Energy partitioning substantially influences the local climate and affects the water cycle, which is of utmost importance for the economy and livelihood of the region. Recently, eddy covariance (EC) towers were installed in the area; thus, the scientific community can thoroughly assess the water and energy fluxes over this unique biome. While EC towers have a high degree of accuracy, they only measure energy fluxes over a small land footprint. Given the biome spatial heterogeneity, the use of EC-based techniques has the limitation of not comprehensively representing water and energy fluxes profiles over the entire region. Incorporating remote sensing (RS) data into the landscape analysis is a feasible solution to overcome this issue, given that satellite data can capture the phenomena represented by the EC measurements across large spatial scales. Our research studied the capability of the Surface Energy Balance Algorithm for Land (SEBAL) and MOD16-ET products to represent the EC measurements regarding energy and mass exchange, with an ultimate objective of applying the best approach to assess these fluxes regionally. We applied the SEBAL model using only remote sensing data from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor. The MOD16-ET model uses a different approach but is also based on MODIS data. Our analysis was based on three years (2014–2016) of data, which was limited to the availability of the EC tower data. We found that for the EC-based measurements, energy balance closure (EBC) achieved an average of 0.84, which is considerably high for the region. This is possibly due to the EC tower being installed on a preserved Caatinga plot, with reduced heterogeneity and higher plant density. When analyzing RS-based products to represent ET profiles in the region, we found that the SEBAL model accurately represented water fluxes during the wet season but not the dry season, whereas the MOD16-ET showed a better agreement with EC-based water fluxes throughout all the seasons. SEBAL inaccuracy in drylands is partially due to the narrow range between the cold and hot pixels in an image, as the algorithm relies on this range for input parameters, especially in the dry season. Therefore, we concluded that MOD16-ET is capable of better-representing water fluxes in the Caatinga region. We analyzed the fluxes regionally and quantified annual ET for the three years. These results are especially relevant for local policymakers on dealing with water and landscape issues in a region where the livelihood and well-being of the population is inextricably bound to water availability.