Mine void identification using Object-based Image Analysis (OBIA) of satellite imagery Sentinel 2 data

Abstract

Open pit mining is an extensively-used method in Indonesian coal mining. This method is characterized by the formation of mine void at the end of life-of-mine (LOM) due to insufficient material to backfill the mine-out areas. Mine voids are legally accepted as one of mine closure options and categorized as “<em>Reklamasi Bentuk Lain</em>” - a miscellaneous reclamation option (Decree of Minister of Energy and Mineral Resources/KepMen ESDM No.1827, 2018), However, unmanageable voids will exert negative impacts. The identification and mapping of mine voids spatially are imperative to give stakeholders ample information to construct viable mine voids management and benefit all stakeholders. In this research, Sentinel 2 satellite image data is used for land monitoring so that void can be mapped based on land cover classification. The land cover classification was carried out based on the Object-based Image Analysis (OBIA) method. This method has a good level of accuracy, ranging from 86.1 to 96.4%. Based on the land cover classification, potential voids are analyzed based on their shape, where potential voids have elongation values of 0.2-1.0 and circularity of 0.1-0.8. In addition, potential voids are analyzed based on the location where they are found (referred to as the Mining License Area/WIUP data). In 2018 there were 40 potential voids inside WIUP and 5 potential voids outside WIUP, while in 2020, 62 potential voids inside WIUP and 8 potential voids outside WIUP were identified in the study area. The final result of potential mine void, i.e. mine void-1 could not further be distinguished between mine sumps, voids, or mine ponds without additional data and analysis. On the other hand, mine void-2 could not be further assigned as natural water bodies or mine void from illegal activities. Subsequent studies using more elaborated data, processes, and analysis are important, to enhance the accuracy of void mapping using satellite images.