These profiles were used as primary integrated observed knowledge about the farm to create the probabilistic hot-spot map (Figure 9). Each pixel on the Google earth represents 15m-resolution analytical outcome from the big data integration. Soil moisture, soil surface temperature, water consumption and NDVI values of each pixel were used to train a simple data driven model to train a supervised knowledge predictor. The refined threshold of these variables was modeled based on the ground truth collected and also by close consultation with farm manager. This predictive system was trained based on farms historical records and hyperspectral ground truth data as training targets.