With the support of daily acquired satellite images from SPOT and Pléiades, plus UAS monitoring, the agricultural waste burning sites can be identified by contrasting the differences of spectral response and carbon emission between harvested and waste-burning farmlands.
Based on a machine-learning algorithm, this study employs the object-based image analysis (OBIA), which segments image pixels into several objects in order to estimate the waste burning area.
Time-series imagery was used to automatically detect and map the burn sites after 2017’s second rice season and 2018’s first rice season. The results show the spatial distribution of the sites, which is invaluable to government agencies investigating this illegal practice.