Forecasting the wheat powdery mildew (Blumeria graminis f. Sp. tritici) using a remote sensing-based decision-tree classification at a provincial scale

Powdery mildew (Blumeria graminis) on wheat (Triticum aestivum) is one of the most common and devastating foliar diseases, which has resulted in significant reductions in wheat production. The study discusses an assessment of Moderate Resolution Imaging Spectroradiometer (MODIS) time-series data products for forecasting the incidence of wheat powdery mildew at a provincial scale. Firstly, the wheat areas were identified using 8-day interval Normalized Difference Vegetation Index (NDVI) dataset at 250 m resolution. A decision tree was then constructed to identify four infection severities (healthy, mild, moderate and severe) using three kinds of forecasting factors including wheat growth situation (NDVI), habitat factors (land surface temperature, LST) and meteorological conditions (rainfall and air temperature). The results show that the coefficient of determination (R2) is 0.999 between the remote sensing based and the statistical data. Wheat-growing areas were primarily distributed in Fuyang, Bozhou, Suzhou and Huaibei of Wanbei (54.38%) and the northern part of Wanzhong. The overall forecasting accuracy was 83.33% and the infected wheat areas showed a spatial spread from the capital city to surrounding regions. The overall infection rate of Anhui Province was 15.64% and the mildly affected wheat areas accounted for 65.07%.