Understanding fall armyworm infestation in maize fields of Bangladesh using temporal Sentinel-2 data

Fall armyworm (FAW), J.E. Smith Spodoptera frugiperda, is one of the most harmful crop pests that has caused a significant threat to food security worldwide. In Bangladesh, FAW was first detected in November 2018 and since then has affected the production of maize in the country. The study aimed to map the intensity of FAW infestation in maize fields across Bangladesh using Sentinel-2 data and machine learning algorithms. Field data was collected in six divisions of Bangladesh by CIMMYT -Bangladesh during the 2019 (December)–2020 (March) maize growing season. In total, 579 maize fields were sampled, and 6998 maize field samples were taken by means of weekly scouting across the divisions. Sentinel-2 spectral indices and bands were investigated to identify spectral features altered by the infestations. The Partial least squares discriminant model was trained using field-collected samples, and its accuracy was assessed. Our preliminary results show that FAW infestation intensity can be mapped using temporal Sentinel-2 data. Identifying FAW infestation intensity and hot spots using remote sensing is an effective and valuable approach for early estimation of damaged maize and yield and to plan crop management mitigations.