Synergistic contextual information and individual sample as mean training approach: paddy stubble burning mapping

Paddy stubble burning is a prevalent agricultural practice in India, particularly after paddy cultivation, making the country the second-largest contributor to crop residue burning (CBR) globally, releasing approximately 84 Tg/year of aerosols and pollutants, significantly exacerbating air quality and public health crises. This study aimed to enhance the identification of paddy stubble-burning activity at the field level by integrating the contextual possibilistic c-means (PCM-S) model and individual sample as mean (ISM) training approach. By analysing spectral and temporal data from PlanetScope and Sentinel-2, the study optimized the classification of burnt paddy fields. The contextual PCM-S model, which incorporates neighbouring pixel effects, was combined with the ISM training approach, which preserves individual sample characteristics during the training process. This integration, along with pre-burnt and post-burnt temporal data, effectively addressed noisy pixels and field heterogeneity caused by varying harvesting techniques. Moreover, it helped prevent the recurrence of burnt fields in subsequent observations and facilitated the identification of fields that were burned and immediately ploughed. The key findings demonstrated that among 155.42 sq. km of paddy fields in the vicinity of Patiala, 27.07 sq. km were burnt across ten mapped dates, constituting 83.99% of the total burning events mapped across 13 dates of harvested paddy fields. The results showed good accuracies and validation, with minimal intra-class mean membership difference (MMD), indicating negligible variability within the same class (almost 0), higher inter-class MMD, representing a clear distinction between classes (nearly 1), negligible variance (approximately 0.0001), minimal entropy (about 0.05), and a statistical F-score exceeding 0.9. These findings underscore the significant occurrence of paddy stubble burning, despite efforts to manage paddy crop residue, underscoring the urgent need for immediate measures to mitigate future occurrences.