Mapping changes in artisanal and small-scale mining (ASM) landscape using machine and deep learning algorithms. - a proxy evaluation of the 2017 ban on ASM in Ghana

Assessing the Applications of Earth Observation Data for Monitoring Artisanal and Small-Scale Gold Mining (ASGM) in Developing Countries

“…An effective band/feature selection process would result in enhanced performance of the model in terms of costs and accuracy of the results [45]. For instance, it has been shown in a case study in Ghana that the Sentinel-2 Band 5 (band center 705 nm) was the highest contributor to a land cover classification and, more importantly, it contributed most to delineating mining sites [19]. Classification models can also use multi-band indexes as input data such as the normalized difference vegetation index (NDVI) [13,19,31].…”

“…For instance, it has been shown in a case study in Ghana that the Sentinel-2 Band 5 (band center 705 nm) was the highest contributor to a land cover classification and, more importantly, it contributed most to delineating mining sites [19]. Classification models can also use multi-band indexes as input data such as the normalized difference vegetation index (NDVI) [13,19,31]. However, note that NDVI is influenced by many environmental factors such as topography, bare soil conditions, atmospheric conditions, vegetation association, rainfall, and non-photosynthetic materials [50].…”

“…Various bands and indexes have proven useful in detecting water turbidity and suspended sediment and can be used to identify the impacts of ASGM along rivers, especially, when considering long time series of satellite data [17,51]. Examples include the modified normalized difference water index (MNDWI), the Band 8A-VRE 4 and the Band 3 (Green) in Sentinel-2-A data [19]. Another example is the combination of Landsat 8 bands; (4,3,2) to distinguish deep water from shallow water, (5,6,4) to distinguish water from ground and, (6,5,2) to distinguish bare ground from ponds [43].…”

“…RS data can be integrated with environmental data to better analyze the influence of the mining process on the biota/mercury content in the environment. For instance, the location of mining sites from existing datasets or from field work can be utilized to create reference data that train supervised classification models or to refine model results with the assumption that a non-existing mine in recent high-resolution images implies that there was no mine present in the same location even in the past [19,44,[52][53][54]. Similarly, the turbidity of water may be used as a proxy for mercury content in water if the RS data are combined with in-situ water sample data that are taken on specific dates that correspond with the available RS data [17,20].…”

“…The World Database of Protected Areas also used RS tools to map about 700 ha of protected areas that have been disturbed by ASGM activities in Latin America and Africa [13]. Other examples include the works of Bruno, et al [14], Isidro, et al [15], Ibrahim, et al [16], Lobo, et al [17], de Lucia Lobo, et al [18], Nyamekye, et al [19], and Telmer and Stapper [20]. These studies have used both optical spaceborne RS data of high and medium resolution to identify and follow the evolution of artisanal and small-scale mines across space and time.…”