How can we use remotely sensed imagery to explore land cover
dynamics in a heavily disturbed area?
A case study at the Hobet Mine in Boone County, West Virginia
Mountaintop removal is a destructive form of surface mining that has scarred large swaths of the Central Appalachians. In this landscape-altering process, explosives are used to blast off the tops of mountains and reveal the coal seams below. The rubble is deposited into adjacent valleys, essentially leveling the topography. Mining companies are required by federal law to restore the site to its “approximate original contour” and replant the land to reduce soil erosion and mitigate water pollution, but in reality these efforts have largely failed to restore the healthy forests that once blanketed the Appalachian landscape. Companies tend to plant aggressive invasive species because they spread quickly and hold down the soil, and many of the region’s reclaimed mine sites have become weedy expanses of brush, shrubs, and grasses growing on heavily compacted soil.
Operating from 1974 to 2015 in southern West Virginia’s Boone County, the Hobet Mine was one of the state’s largest mountaintop removal coal mines. Satellite imagery from Landsat 5, 7, and 8 span a wide enough range of time to track the progression of the mine across the landscape for the entire duration of its operation, and the more than fifteen square miles of impacted land is so expansive that fine spatial resolution is unnecessary.
The three maps in this app represent imagery from the summer months of three different years in the history of the Hobet Mine. In 1984, mining was well underway in the initial stages of operation in a small area east of the Coal River. 2002 was the mine’s peak of output, producing a record 5 million tons of coal in a single year. At this point, the mine had advanced westward and laid waste to a large swath of land in the Connelly Branch watershed. The final map layer shows the most recent Landsat imagery from the summer of 2021, allowing us to examine the current state of deforestation as well as revegetation in the mine’s wake.
While the contrast between the barren mine site and the green surrounding forest is undeniably clear in natural color imagery, these maps display an additional spectral band to further illuminate land cover changes. The study site is presented in a standard false color composite, meaning that near infrared surface reflectance is displayed through the red color channel, while visible red and green reflectances are displayed through the green and blue channels, respectively. This type of composite is useful for visualizing changes in forest cover and land use types; vegetation appears in shades of red, while areas of human development are cyan blue. Examining the maps, we can see that untouched areas that remained in their original forested state are deep red, indicating healthy, broad-leaf vegetation. In contrast, the lighter red color of the reclaimed mine sites suggests that these areas consist mainly of grassland or other sparse vegetation.
The spectral signatures of the sample points shown on the map provide further insight into the reforestation status of the former Hobet Mine. The green points serve as a control, representing areas that remained forested throughout the entire time period. The blue points, on the other hand, lie within areas that were forested in 1984, mined in 2002, and revegetated in 2021. The spectral signatures on the left-hand side of the app represent the surface reflectances of different wavelengths in the electromagnetic spectrum for each of the six points. Examining the charts in chronological sequence tells the story of land cover change at this mine site.
In the 1984 chart, all six spectral signatures are nearly identical, each one representing a patch of healthy forest. In 2002, however, a dramatic change occurs. Areas that were once forested are now active mine sites that display drastically different spectral signatures from the untouched land throughout the entire measured portion of the electromagnetic spectrum. The 2021 chart shows a promising trend towards revegetation, as the former mine sites now begin to resemble the healthy forest one more. Upon closer inspection, however, it is evident that the reclaimed sites have not fully recovered the characteristics of the untouched forest. The most conspicuous difference between the spectral signatures in this chart occurs in the near-infrared range (around 750-1200 nm), which indicates differences in cell structure. Healthy plants are known to reflect more near-infrared radiation than stressed plants, allowing us to infer a deficiency of healthy vegetation in the reclaimed sites. This lower near-infrared reflectance among the reclaimed points also explains the lighter red hue in the false color composite map.
The results of this comparison of false color composite imagery and spectral signatures match the findings of the Central Appalachian Mine Reforestation Assessment. This report assessed reforestation progress on formerly mined lands by using spectral indices from Landsat imagery to measure forest health based on vegetation greenness and moisture. The report’s data from the Hobet Mine region indicate that much of the reclaimed site consists of grassland or poor forest, and that the land has yet to recover its previous forest characteristics.
Remotely sensed imagery and spectral signatures clearly illustrate the destructive advancement of the Hobet Mine across the Central Appalachian landscape and the unsuccessful ecological restoration in its wake. However, there is only so much information we can glean from the near-infrared spectral band of the Landsat data, and we are left with only a vague concept of vegetation health. An examination of various other bands and indices from those bands could provide a more comprehensive picture of ecosystem structure in reclaimed mine sites. Ultimately, remote sensing will need to be followed up with an on-the-ground ecosystem assessment and reclamation plan if these landscapes are ever to be restored to their former glory.