1. Monitoring Landcover changes of small-scale alluvial mining in Columbia Based on Multi-source Remote Sensing Data
- Author
-
Lifan Xiong and Jingyi Jiang
- Abstract
Small-scale and artisanal mining is widespread in many countries and often causes severe eco-environment problems. The dynamic nature of this extractive activity and its spread in rural and remote areas make its mapping and monitoring very challenging. Therefore, the use of multi-source remote sensing data provides possibilities to monitor landcover changes as well as the destruction and restoration of vegetation in these regions. The study area of this paper is located at the border of El Bagre and Zaragoza, in Bajo Cauca, Colombia, which is greatly affected by small-scale alluvial gold mining. This place has a typical tropical climate with frequent high cloud coverage. Therefore, the multi-source remote sensing images (Sentinel-2, Landsat-8 and Sentinel-1) are selected for this study to detect and map small-scale mining of gold placers. For both Sentinel-2 and Landsat-8 data, clouds are removed according to the information of pixel QA bands through the google earth engine (GEE). Then multiple images within each two months are aggregated after cloud removal. For Sentinle-1 data, post-processing including noise removal, calibration and orthorectification using DEM data is applied with GEE. When cloud-free data is not available within two months, Sentinle-1 data is added to improve the accuracy of the classification. The support vector machine classifier (SVM) is selected as the supervised classification approach. And to avoid possible classification errors, the post-classification based on local geographic information is also applied. Results show the landcover changes of mining areas in recent years and detect the location of mining areas, including mined bare soil and mining ponds and other important land cover. Furthermore, the change of vegetation cover in this area is monitored.
- Published
- 2023