Assessment of land cover degradation due to mining activities using remote sensing and digital photogrammetry

Abstract

Appropriate environment management requires an understanding of how mining activity alters environmental characteristics and how those changes affect an area. Therefore, to reduce the adverse effects of mining activity on the land, it becomes crucial to have relevant information about responses to environmental degradation. This study aims to assess the impact of semi-mechanised and artisanal mining activities on the land cover using remote sensing data and photogrammetric analysis, in the Mbale locality, Northern Cameroon. For this purpose, the maximum likelihood classification algorithm of the supervised classification method combined with field surveys was used to map environmental changes, based on Sentinel-2 images of 2019, 2021, and 2023. Normalized Difference Vegetation Index (NDVI), Normalised Difference Water Index (NDWI), Brithness index (BI), and Soil crust index (SCI), were calculated to assess changes in vegetation, bare soil, water body, and exploited area. The orthophoto obtained from photogrammetric processing was performed to outline river network change through visual interpretation techniques and to calculate the volume of pits created by mining. The result of classified images indicated that vegetation cover decreased by 11.74% over the studied years. However, bare soil and exploited areas increased by 9.2% and 5.4% respectively. The calculated spectral indices show that between 2019 and 2023 the locality of Mbale considerably lost its vegetation cover, in favor of bare soil. The color of the soil and the granulometric size of the topsoil have also changed. The photogrammetry analysis highlighted the deviation of the main river and estimated the volume of pits created by mining activity to 22188.7 m³. The mining activities caused a loss of the vegetation cover, generated big pits, and multiple deviations of the Lom River from its natural course, which have a substantial negative influence on the ecosystem. Such data can be used for long-term environmental management, reclamation and rehabilitation monitoring, and mining area restoration.

Introduction

Mining is an activity that is practiced in almost every country in the world. The considerable global expansion in mineral resources since the end of the 20th century has led to the proliferation of large mining projects pushing the limits of what is acceptable further because mining activities constitute sources of socio-economic benefits (Ettler 2016; Maconachie 2012). In the strategy document for growth and employment, the Cameroonian government makes exploiting mineral resources one of the pillars of the country’s economic take-off and emergence in 2035. The discovery of iron deposits at Mbalam, aluminum at Ngaoundal and Minim-martap, gold at Colomine and Betare-oya, diamonds at Mobilong, cobalt, and nickel at Lomie, and gold along the Lom River (Djouzami, Wakasso, Mbale, etc.), has attracted not only foreign investors but also local communities to mining.

However, the expansion of mining activities raises numerous challenges, one of which is environmental protection. Indeed, mining activity modifies the functioning of the ecosystem more or less profoundly. It frequently results in significant changes to the ecology and landscape, including deforestation, farmland degradation, pollution of the soil and water, biodiversity loss, and landscape shaping (Taylor et al. 2004; Greffre et al., 2006; Joshi 2006; Musa & Jiya 2011; Andriamasinoro and et Angel 2012; Panagos et al. 2013; Petropoulos et al., 2013; Khalil et al. 2014; Merrem et al. 2017; Onyemachi et al. 2022).These consequences have an impact on the health and life of the population (Ouedraogo 2006; Issah 2022).

The environmental deterioration caused by mining occurs mainly as a result of inappropriate and wasteful working practices and rehabilitation measures (Kamga et al.2018), as well as the lack of national legislation and policies to guarantee sustainable land management. Apart from poor environmental governance as highlighted in the Africa Mining Vision, many African countries lack a precise assessment of the impacts of abandoned and active mining activities in their territories (Kamga et al. 2018).

Thus, understanding the spatiotemporal dynamics of the land use/ land cover around mining sites is critical for developing mitigation strategies for environmental impact. Environmental degradation due to mining has been addressed by many researchers around the world. Hilson (2002); Betancourt et al. (2005); Zwane et al. (2006); Gomes (2010); Guédron et al. (2017); Strady et al. (2017) reported the degradation of the physicochemical quality of water resources due to mining activity. Li and Wu (2008); Malaviya et al. (2010); Zhai et al. (2012) and Anderson et al. (2020) have used remote sensing to quantify and map land use/land cover degradation around mining sites. This method has been proven to be effective at the regional level. It is faster, cost-effective, and can easily cover large surface areas, and the uncertainty level of the maps generated can be determined.

In the East and Adamawa regions of Cameroon, most of the studies on the assessment of environmental degradation consisted of the physicochemical characterization of water and sediments from mining areas. (Rakotondrabe et al. 2018; Ayiwouo et al. 2020; Ngounouno et al. 2021; Bella Atangana et al. 2023). Very few studies have been conducted using remote sensing. One of the few in the Adamawa region has been carried out by (Ngounouno et al. 2023). The authors have used the Leopold matrix, the Fecteau grid, and the remote sensing approach to assess the environmental impact of the Ngankombol gold mine. By calculating the normalized vegetation index, they noted that mining activity in this area has mainly resulted in the loss of vegetation cover and an increase in bare soil. Remote sensing is therefore an appropriate tool for such a study. However, no study integrating photogrammetry with drone images has been conducted in the Adamawa region.

The present study assesses the environmental impact of mining activities in Mbale (Adamawa Region) using remote sensing integrating photogrammetry. Gold mining has become widespread in recent years in the locality of Mbale. This locality lacks information on the real extent of land use/land cover degradation due to mining operations. The exploitation leads to the digging of open pits which are generally left open after the mining process. Photogrammetry is a method used to estimate the volume of these pits, to guarantee good restoration. Scientific studies in this area are scarce and deserve to be addressed to provide appropriate information to the government, to ensure sustainable land management.

Method

Study area

As shown in Fig. 1, the study area is a polygon of approximately 108.47 km² located between Longitude 14°20’00’’ and 14°27’35’’ E and Latitude 6°5’05’’ and 6°12’35’’ N. In administrative terms, it is located in the Adamawa region of Cameroon, Mbere division, Meiganga District. Mbale locality is characterized by diverse topographic conditions like plateaus dissected by U-shaped valleys and V-shaped valleys, terrains with steep slopes, and gentle slopes. The elevation ranges from 900 m to 1500 m. The climate is humid tropical, with two major seasons, the rainy season (7 months) and the dry season (5 months). The annual average temperature of the area is 23. 3 °C; the annual average rainfall is 1530 mm. The study area is mainly drained by the Lom River and we note the presence of a few small rivers distributed on both sides of the main river. The geology is made up of anatexia granite; ancient syntectonic granite and granodiorites; gneiss and concordant granulites and granites (Issohou 2015). The most common types of soil are ferralitic, hydromorphic, humiferous, suitable for agriculture and grazing in the lowlands. Sandy-clayey soils are also observed, mainly in the valleys. The vegetation is not dense because it is made up of a shrubby and grassy savannah.

Fig. 1

Location map: (a) Mbere Division and Adamawa Region (b Meiganga Subdivision in Mbere (c) study area

Remote sensing data processing and analysis

The study uses Sentinel-2 images from January to March for 2019, 2021, and 2023 respectively, with T33NVG as the scene, selected and downloaded from the Copernicus website. To reduce seasonal variations and cloud influence, the chosen images were taken throughout the same season (dry season). For the case of Mbale, this period (January to March) displayed the best quality and precision in sentinel images. Pre-processing of the images was carried out before performing the classification of the remote sensing data (Demissie et al. 2017; Meer and Mishra 2020; Ewunetu et al. 2021). Processing data entails:

• Atmospheric and geometric corrections: to correct geometric distortions caused by sensor-Earth geometry variations and conversion of the data to real geographic coordinates.

• Layer stacking and suitable band selection combines multiple image layers into a single image, to avoid information redundancy. Bands with different spatial resolution were resampled to 20 m resolution. The colour composite 4-3-2 bands, for RGB channels were chosen and enabled to enhance the identification of features to select training set or classification signatures to be used.

• Image enhancements improve the appearance of the image to assist in visual interpretation and analysis.

To guarantee consistency among datasets, all data were projected to the World Geodetic System 84 datum and the Universal Transverse Mercator projection system (zone 32 N). Then, these data were employed to perform supervised classification using the ENVI 5.3 software and to analyze the variation between spectral indices.The Maximum Likelihood Classification was chosen because it is the most widely used per-pixel method with remote sensing image data (Liu et al. 2011). In addition to the reflectance values, it also takes into account the covariance of the information contained in the sensors’ spectral bands of land use and land cover classes (Gupta 2013). In total, more than 28,392 pixels as reference points were collected in 2019, 2021, and 2023, covering four land use and land cover classes that were defined for each image: vegetation, bare soil area, water body, and exploited area.

In remote sensing, assessing the quality of a classification result is crucial because it shows how well the classifier can extract the necessary items from the image. The confusion matrix is a general method for evaluating the accuracy of remote sensing image classification, which provides the correspondence between the land use land cover classification results and verification data (Liu et al. 2020). In this study, the verification of classification accuracy was made by overall accuracy, kappa coefficient, producer’s accuracy, and user’s accuracy (Jensen 1996; Mathur and Foody 2008; Belay and Mengistu 2019). Overall accuracy represents the total classification accuracy. User’s accuracy is the probability that a value predicted to be in a certain class is that class. Producer’s accuracy is the probability that a value in a given class was classified correctly.

The Kappa coefficient expresses the proportionate reduction in error generated by the classification in comparison with a completely random process. It measures the agreement between the model predictions and reality (Tilahun and Islam 2015).

The interpretation of the Kappa coefficient follows the guidelines provided by Bogoliubova and Tymków (2014) as presented in Table 1.

Table 1 Interpretation of agreement for kappa coefficient (Bogoliubova and Tymków 2014)

Spectral indices are widely used in various applications of remote sensing due to providing relevant class information with less computational overload software (Bramhe et al. 2018). All indices, whether vegetation indices, soil indices, water column indices, etc., are based on an empirical approach based on experimental data (Koffi et al. 2017). There are a multitude of spectral indices that have been developed over time by scientists and users of optical satellite images, but the present work will be limited to a few. Table 2 summarises the indices selected.

Table 2 Spectral indices used in this work

Drone image processing (photogrammetry)

To provide high-resolution images quickly and accurately of the desired sites even inaccessible, drone flights were used. The optical photogrammetry is used with the drone data set. The drone was operated automatically using a flight plan. The flight was carried out on February 2023, at a height of 100 m above ground, with constant speed. It moved in horizontal lines and captured images at continual intervals. The data obtained by the drone were processed using Pix4D software. The image processing consists of initial point cloud generation, 3D mosaic of the image production and ortho-image generation. This ortho-image was used to carry out the Digital Elevation Model (DEM) of the current state of the slopes, from which the volume of the pit generated by mining activities, was calculated. Also, the ortho-image allowed us to visualise river channel deviations.

Results

Land cover classification

The spatial patterns of land cover changes for the years 2019, 2021, and 2023 in the Mbale area are displayed in Fig. 2. A progressive decrease in the area of the vegetation cover was noticed: 6.98% between 2019 and 2021, and 4.76% between 2021 and 2023 (Table 3). However, the change rate of bare soil and the exploited area is observed to grow continuously throughout the study period, with a net rate change of 9.2% for bare soil and 5.4% for the exploited area. Between 2019 and 2021, the rate change of the water body was − 0.38%, and between 2021 and 2023, it was + 1.61. According to the analysis, the bare soil shows the most positive changes between 2019 and 2023, while the vegetation records the negative changes (Fig. 3). So, the land cover changes were performed gradually over the study period. Thus, mining activities are the main actors of environmental degradation.

Table 3 Land used/ land cover classes and their changes in 2019, 2021 and 2023

Fig. 2

Land cover maps a) land cover of 2019, b) land cover of 2021, c) land cover of 2023

Fig. 3

Rate of land cover change

Accuracy evaluation

An accuracy assessment was done to confirm classification using the confusion matrix. In 2019, 2021, and 2023, the overall image classification accuracies are 96.30%, 97.40%, and 98.64% respectively. The corresponding kappa coefficients for 2019, 2021, and 2023 are 0.94, 0.95 and 0.97. Almost perfect agreements are indicated by kappa coefficients for 2019, 2021, and 2023. Table 4 displays the kappa coefficients, overall classification user’s accuracy, and producer’s accuracy.

Table 4 Accuracies of classified images

Analysis of the degradation of environmental components by spectral indices

The study used NDVI, NDWI, TGSI, BI, and SCI indices to measure the degree of environmental degradation in the study area, from 2019 to 2023 using space-based observation.

Normalized difference vegetation index (NDVI)

The Normalised Difference Vegetation Index shows the vegetation for 2019, 2021, and 2023 in the study area (Fig. 4). On the NDVI scale which runs from − 1 to + 1, the dense vegetation is represented by a value of + 1. The NDVI value of water is typically less than zero. In this study, the highest value which is highlighted in green color represents the vegetation status, whereas the negative values indicate the water body. The analysis of the NDVI maps revealed that gold mining has a significant impact on the vegetation cover in Mbale. The NDVI shows considerable variations over the different years, ranging from − 0.35 to 0.82 in 2019, from − 0.31 to 0.87 in 2021, and from − 0.29 to 0.84 in 2023, reflecting a loss and decline in the overall health of the vegetation. In 2019 the vegetation was rich in shrubs. A slight decrease in shrubs is observed in 2021 and finally, almost no vegetation on the banks of the River Lom, along the N1 national road, and in the villages of Mbale and Bata in 2023, due to mining and human activity. To carry out mining-related activities (prospecting, creation of roads, construction of the settlement, and gold mining), the shrubs on the site must first be felled.

Fig. 4

NDVI maps: a) NDVI 2019; b) NDVI 2021; c) NDVI 2023

Normalized difference water index (NDWI)

The NDWI values also range from − 1 to + 1. The NDWI value is greater than 0 if the cover type is water and less than 0 for others (Mc Feeters 1996). Figure 5 shows the NDWI for the study area. The presence of a water body is being identified through this index for the years 2019–2023 which is represented in blue color showing higher values. The map of the 2019 NDWI (Fig. 5a) shows that only the River Lom in its original bed and certain places on the banks of the Lom are shown in blue. Figure 5b and c, on the other hand, show an increase in the number of pits containing stagnant water, mainly due to the multiple deviations of the riverbed and the filling of the pits with rainwater to exploit the alluvial resource, leading to an increase in the water regime, disruption of the aquatic ecosystem and the creation of traps for livestock.

Fig. 5

NDWI maps. a) NDWI 2019; b) NDWI 2021; c) NDWI 2023

Topsoil grain size index (TGSI)

The TGSI was used to assess the grain size and texture of the soils in the mining areas and also to highlight the disorganization of the granular classes. Negative or near-zero TGSI values represent vegetation cover and water, while values close to 0.20 indicate large quantities of sand. (Xiao et al. 2006). Figure 6 shows the various TGSIs in the study area for 2019, 2021, and 2023 respectively. In 2019, the TGSI varies from 0.014 to 0.37, in 2021 it varies from − 0.02 to 0.35 and finally in 2023 the TGSI values are between − 0.05 and 0.33. The decrease in TGSI values to negative values close to zero highlights a relative increase in different particle sizes in the mined areas. This is because the quartz pebbles accompanying the gold mineralization are extracted at depth and then brought to the surface above the ground in mounds, thus creating a disorganisation of the particle size distribution.

Fig. 6

TGSI maps. a) TGSI 2019; b) TGSI 2021; c) TGSI 2023

Brightness index (BI)

The thinning that results from the increase in BI highlights phenomena such as the loss of vegetation cover, the loss of soil roughness, and desertification.

In the years 2019, 2021, and 2023, the BI varies respectively from 732 to 4578, from 652 to 5122, and from 1985 to 9084. Before the mining boom, the surface area occupied by the road and the village were the areas with the highest BI values. Similarly, the banks of the Lom River, which are the future mining areas, have relatively low BI values due to the vegetation and relatively dark lateritic soils. During the peak mining years (2021–2023), the mined areas have the highest BI. This is because the pebbles extracted from the mineralized zone are relatively light in color, tending towards white, which increases the reflectance of the surfaces.

The maps in Fig. 7 show the BI values for Mbale in 2019, 2021, and 2023 respectively.

Fig. 7

Maps of BI. a) BI 2019; b) BI 2021; c) BI 2023

Soil crust index (SCI)

One of the most important criteria when characterizing soils is their colour, as this is the result of their mineralogical composition and organic matter content (Irons et al. 1989) and can be used to assess the state of degradation of the soil. The SCI has been calculated to evaluate soil colour and its temporal variation, and the maps in Fig. 8 show the results obtained.

The SCI varies from − 0.23 to 0.32 in 2019, from − 0.30 to 0.26 in 2021, and finally from − 0.29 to 0.27 in 2023. An overall decline in soil color is therefore evident, with amplification in the mined areas. The recovery of ore involves the preliminary stripping of vegetation and topsoil, which is rich in organic matter and much more colourful, thus predisposing the soil to erosion, which can either damage the surface horizon rich in organic matter, making the soil lighter and brighter or destroy the soil and expose the parent rock at the surface, whose colour can be different from that of the intact soil (Haboudane 1999). In 2023, the colour of the soil around the Lom River is different from that of 2019 and 2021, which means that the soil in this area has been destroyed and the bedrock has risen to the surface.

Fig. 8

SCI maps. a) SCI 2019; b) SCI 2021; c) SCI 2023

Assessment of environmental degradation using photogrammetry

The photogrammetry results provided a panoramic view of the state of the environment at the abandoned mining sites. Two sites were therefore selected to show the deviation of the river and the volume of pits created by mining activity in the study area. Figure 9 shows the physical state of an abandoned site with the deviation of the watercourse bed and on this same site with an overlay of the old bed of this watercourse marked in blue and obtained by digitising a Google Earth image of the area in 2015.

Fig. 9

Deviation of the Lom River bed. a) Image of an abandoned site taken by a drone. b) Google Earth image of the old river bed superimposed

Mining activity in the Mbale area takes place along the river and this generally leads to a change in the river channel. Figure 9b shows that the river has widened and deviated from its usual course, as indicated by the arrow. As a result of this change, aquatic fauna is disturbed and displaced, and a new floor is created, giving way to totally white, polluted lakes.

Figure 10 shows the pit created by mining and the digital elevation model of a site in full operation.

Fig. 10

Pit created by mining a) pit created by mining b) digital elevation model

Mining activity in the study area generally leads to the creation of gigantic pits; the digital elevation model enables us to estimate the volume of this pit at 22188.7 m³. The creation of pits from gold mining represents a real danger to the safety of the local inhabitants, their livestock, fauna, and even the acoustic environment.

Discussion

The scientific community widely accepted the use of data from remote sensing to assess ecosystem degradation. In this study, the adopted methodology was used to capture and analyse environmental changes due to gold mining activities. By analysing classification images from 2019 to 2023, changes in the spatial distribution of land cover classes were detected. Land cover dynamics is categorised into four classes: vegetation, bare soil, water body, and exploited area. About 11.74% of the vegetation cover of the study area was changed to bare soil and exploited area from 2019 to 2023. The increase in bare soil is related to more abandoned sites, where vegetation has been destroyed. To have enough space for their activities, artisanal gold miners cut down trees, which results in deforestation and exposes more land (Azinwi Tamfuh et al. 2024). The increase in water bodies between 2021 and 2023, is associated with the creation of ponds during panning-related artisanal gold mining due to the expansion of mining activity. It is also related to pits filled with rain or river water. This trend is nearly identical to those found in the literature of (Nodem et al. 2018; Kamga et al. 2020; Ngounouno et al. 2023; Azinwi Tamfuh et al., 2024).

The overall classification accuracy is almost in perfect agreement according to Bogoliubova and Tymków (2014), with 96.30% (2019), 97.40% (2021), and 98.64% (2023). The corresponding Kappa values are stated to be 0.94, 0.95, and 0.97. According to Anderson et al. (1983), effective land use land cover change detection recommends the minimum threshold value for overall accuracy needs to be a Kappa coefficient of 85%. Therefore, this study is acceptable in accuracy.

The spatial changes in the land cover are also detected in terms of spectral indices. The analysis of soil indices shows that mining has significantly changed the soil characteristics in Mbale. The soil is brighter in 2023 compared to 2019, with accentuated grain disorganisation in 2023.

Vegetative areas show a decreasing trend in the NDVI from 2019 to 2023, while the water body shows an increasing trend in the NDWI. This result corroborates the findings of Ngounouno et al. (2023) which was conducted in Gankombol, Adamawa Cameroon.

The photogrammetry analysis estimated the volume of pits created by mining activity to 22188.7 m³. The existence of pits in the areas where mining activities are concentrated, which are frequently filled with water, could harbor pathogens, and also present risks to the lives of humans and animals. Because of this, it is necessary to properly restore the minefields to lessen the risks that already exist and stop environmental deterioration.

According to the study’s findings, mining activities led to changes in land cover. They have a negative influence on the ecosystem, including deforestation, digging of pits, deviating water courses and obstructing another river network, having a variety of effects on water bodies, and alteration of soil profiles. While just taking up a small portion of the land’s surface, mining does have major and frequently irreversible repercussions (Snapir et al. 2017). There is obvious environmental deterioration in the Mbale locality. This situation is encouraged by the fact that there is a lack of national laws and policies to ensure sustainable land management, as well as the ignorance of the indigenous communities in the area regarding the negative impacts of environmental degradation caused by mining operations. If appropriate, serious action is not taken, and by which it was monitored to be done, the environment will become much more damaged. So, this work does not only presents the seriousness of the current situation of the environment of Mbale in the face of artisanal gold mining but also provides appropriate information to the local population for awareness and to the government to take adequate measures to restore this site.

However, the limitation of this work lies in the fact that the evaluation of land cover degradation was made over a short time difference of two years. A long period is necessary to perceive long-term fluctuations of environmental degradation because climatic conditions could cause the same degradation to a certain extent. Drone images must be taken continuously over several years to quantify and model the deviations of the Lom River. This is to propose a better restitution measure of the natural environment of aquatic fauna and flora.

It is recommended that similar research be investigated in the future with a long time difference this should depict the long-term effects of degradation. Strong environmental regulations on mining operations are advised, and initiatives to inform the public about the different facets of mining should be supported by the government.

Conclusion

The study aimed to assess the environmental degradation caused by gold mining in the locality of Mbale. The Maximum Likelihood Classification, spectral indices, and photogrammetry have been used to analyze land cover changes in the Mbale gold mine area. Four land cover classes have been identified from the mining area of Mbale. Out of the 108.47 km² total size of the examined area, the result obtained reveals that the area covered by water bodies, bare soil, and the exploited area increased by 1.23%, 9.17%, and 5.4% respectively. The area covered by vegetation decreased by 11.74%. Mining activity has created pits with a volume estimated to be 22188.7 m³, and deviation of watercourse in many sections of the Lom River. The result of NDVI analysis shows that vegetative areas decreased from 2019 to 2023 and NDWI analysis shows an increasing trend of water bodies. The Topsoil Grain Size Index indicates a disorganization of the granular classes in the soil. The Soil Crust Index shows a decline in soil colour in the studied period and the Brightness index indicates that soils become lighter as mining progresses. It is then clear that the environment of Mbale has undergone fairly significant disturbance since the activity took off between 2019 and 2023.

The application of this method in the study area has seriously illustrated the capabilities of remote sensing imagery and its methods for determining the patterns of change in land cover in the areas affected by the mining influx. The study produced a basic document that would help the government and environmental policy for potential environmental changes in the area. It will provide the foundational data for further environmental issues.