Hotspot Clustering Using ST-DBSCAN Algorithm to Detect Forest Fire Prone Areas in Riau Province Based on Spatial and Temporal Aspects

Forest fires in Indonesia occur almost every year, the impact of which can be felt by all Indonesian people and neighboring countries. Therefore, it is necessary to have preventive methods, one of which is by means of prevention which can be done by identifying the distribution of hotspots, we can find areas that have a high risk of fires so that early preventive measures can be taken in these areas. The problem to be discussed in this study is how to classify areas prone to forest fires, especially in Riau Province, using the ST-DBSCAN algorithm. This study aims to determine the pattern of distribution of hotspots in forests in Riau Province from January 2016 – December 2022 using the ST-DBSCAN algorithm and Silhouette Score for method validation. The results obtained from this study are using a year trial that has a dense cluster pattern, the year is in 2019 with 16 clusters. As well as the results of the silhouette method testing with the calculation of the entire data to get a result of 0.741 where this value is included in a strong condition because it has an interpretation result close to the maximum silhouette result, namely 1.