Use of Remote Sensing Data to Estimate Sugar Beet Crop Yield in the Doukkala Irrigated Perimeter

Abstract

Sugar beet is grown-up mainly for its root as a source of sucrose for human food. It contributes to job creation and participates in the growth of the economy. At the Doukkala level, it is the main crop in the agricultural rotation system. The main objective of this study is the estimation by remote sensing of root weight yields and sugar richness of this crop, based on empirical statistical relationships between yield and vegetation indices, and this for three spatial resolutions: 10m (Sentinel-2), 15m (pansharpned Landsat8) and 30m (Landsat 8). The vegetation indices that have a positive relationship with sugar beet crop yields are NDVI, NDWI, and NVI. Concerning crop yield root weight, the prediction results were still moderate; NDVI of Sentinel-2 (R2 = 0.496), NDVI of Landsat-8 (15m) (R2 = 0.340) NDWI of Landsat 8 (30m) (R2 = 0.400). For sugar richness, the annual sum of NVI (sentinel-2) and NDVI (Landsat-8, 30m) the prediction results values were R2 = 0.350 and R2 = 0.227, respectively, while NVI (Landsat-8, 15m) has a prediction value of R2 = 0.543. In conclusion, NDVI from Sentinel-2 allowed a better prediction of root weight yield with an RMSE prediction error of 9.93 t.ha-1; whereas for sugar content prediction the best value with an RMSE of 0.75% was obtained using NVI from Landsat-8(15m).