Due to increasing global population and increasing food demand, crop yield needs to be improved to forestall potential food shortages. To achieve optimal crop yield, irrigation is applied to replace water losses due to evapotranspiration (ET). Information on ET should be applied to optimize water allocations and water use. In this study, the correlation between ET and yield will be investigated on field-scale level to assess the potential of high resolution ET products as a tool to detect yield variation. The variability of crop yield and transpiration are caused by the variability in the topography, groundwater and soil properties.
Agricultural practices and field-scale water management demand high resolution (in meters) and high temporal resolution (daily to sub-daily) remote sensing products. With the arrival of new satellite platforms, such as Sentinel-2, the aforementioned remote sensing data can be improved significantly in spatial and temporal resolution. In order to compare the functionality of different remote sensing products, an assessment is executed for two satellite derived vegetation indices: normalized difference vegetation index (NDVI) and normalized difference water index (NDWI), and two satellite derived evaporation products: WaPOR (Water Productivity through Open access of Remotely sensed derived data) and a newly developed evaporation algorithm from VanderSat. Within this research, the focus lies on assessing which dataset is able to observe the spatial difference and temporal patterns on field-scale level. Using a large sugarcane plantation in Xinavane, Mozambique, as a case study, we demonstrate how the spatial variability of the remote sensing results are correlated to the sugarcane yield. To assist irrigated agriculture we demonstrate that a high resolution evaporation product is needed to incorporate spatial variability in evaporation estimates. The analysis shows that the high resolution satellite derived vegetation indices are related to the spatial variability of yield. Our results indicate that NDWI has a strong positive correlation of 0.73 with yield, but NDVI has only 0.64. The actual evapotranspiration estimates have a moderately positive correlation with yield of 0.5 for WaPOR and 0.57 for VanderSat. Evaporation estimates should be related to yield to control irrigation properly. WaPOR and VanderSat use NDVI as a input for crop stress, these existing evaporation algorithms should incorporate high resolution spatial imagery as NDWI instead of NDVI to assist irrigation adequately. In order to use the satellite derived evaporation algorithms for agricultural practices and field-scale water management, future research should be focus on improving the relation between satellite derived evaporation algorithms and yield.