Detection of Spatio-Temporal Hotspot using NBR and dNBR for Quality and Strength of Clusters Dita Baitu Rahmawati(a), Rita br Purba(a), Indra Ranggadara(a*), Nia Rahma Kurnianda(a)
a)Faculty of Computer Science, Mercu Buana University, Indonesia
*indra.ranggadara[at]mercubuana.ac.id
Abstract
Hotspots are used as an indicator of land and forest fires. This study applies the NBR and dNBR index features to identify the hotspots spatio-temporal density based on the burn severity index. Image data in this study were obtained from the Landsat 8 OLI satellite used to detect fire areas in Katingan, Central Kalimantan, an area of peatland and forest. The image data period used is from July 2019 to December 2019, with the clipping process as preprocessing and NBR and dNBR as feature extraction to detect hotspots. Furthermore, the results of feature extraction hotspots are clustered using the DBSCAN algorithm, then analyzed using the Silhouette Coefficient to define the quality and strength of the clusters. This research provides a value of the highest Silhouette Coefficient is 0.90771, which is found in the dNBR index feature from July 2019 to August 2019, which results in two clusters. This shows that the cluster formed has a strong structure, so it is concluded that the dNBR index feature has better quality than NBR in identifying hotspot areas.
