Search

Context and background Wheat is one of the oldest cultivated plants in the world and has always been one of the most important staples for millions of people around the world and especially in North Africa, where wheat is the most used crop for typical food industry. Thus, an operational crop production system is needed to help decision makers make early estimates of potential food availability Yield estimation using remote sensing data has been widely studied, but such information is generally scarce in arid and semi-arid regions such as North Africa, where interannual variations in climatic factors, and spatial variability in particular, are major risks to food security. Goal and Objectives: The aim of this study is to develop a model to estimate wheat yield based on phenological metrics derived from SENTINEL-2 NDVI images in order to generalize a spatial model to estimate wheat yields in Morocco's semi-arid conditions Methodology: The 10 m NDVI time series was integrated into TIMESAT software to extract wheat phenology-related metrics during the 2018-2019 agricultural season, the period in which ground truth data was collected. Through the multiple stepwise regression method, all phenological metrics were used to predict wheat yield. Moreover, the accuracy and stability of produced models were evaluated using a K-fold cross-validation (K-fold CV) method. Results: The results of the obtained models indicated a good linear correlation between predicted yield and field observations (R2 = 0.75 and RMSE of 7.08q/ha). The obtained method could be a good tool for decision makers to orient their actions under different climatic conditions

Context and background Wheat is one of the oldest cultivated plants in the world and has always been one of the most important staples for millions of people around the world and, especially in North Africa, and especially in Morocco where wheat occupies an important place in the agricultural sector. Thus, an operational crop production system is needed to help decision makers make early estimates of potential food. availability Yield estimation using remote sensing data has been widely studied, but such information is generally scarce in arid and semi-arid regions such as North Africa, where interannual variations in climatic factors, and spatial variability in particular, are major risks to food security. Goal and Objectives: The aim of this study is to develop a model to estimate wheat yield based on phenological metrics derived from SENTINEL-2 NDVI images in order to generalize a spatial model to estimate wheat yields in Morocco's semi-arid conditions. Methodology: The 10 m NDVI time series was integrated into TIMESAT software to extract wheat phenology-related metrics during the 2018-2019 agricultural season, the period in which ground truth data was collected. Through the multiple stepwise regression method, all phenological metrics were used to predict wheat yield. Moreover, the accuracy and stability of produced models were evaluated using a Kfold cross-validation (K-fold CV) method. Results: The results of the obtained models indicated a good linear correlation between predicted yield and field observations (R2 = 0.75 and RMSE of 7.08q/ha). The obtained method could be a good tool for decision makers to orient their actions under different climatic conditions. [ABSTRACT FROM AUTHOR]