A Landscape-scale Approach for Evaluating Cumulative Impacts

Preliminary Study Results

Peter R. Claggett, John R. Pomponio, and Emily C. Clifton

Canaan Valley Institute, Valley Forge, Pennsylvania

Abstract

This paper demonstrates how coarse scale landscape indicators and Geographic Information Systems (GIS) can be used for assessing the cumulative environmental impacts from spatially extensive projects such as mountaintop mining. Twelve landscape indicators are analyzed for three different land cover scenarios based on varying degrees of mining. These indicators are then combined to form composite indicators relating landscape changes to habitat for generalist species, habitat for interior-forest species, and water quality. Additional data is in the process of being gathered and a complete report will be finalized in the Fall 2000.

Introduction

Our landscape is continuously changing due to both natural and human disturbances. Natural disturbances such as tree falls, fires, floods, windstorms, and climatic fluctuations may alter stream morphology and the distribution of plant, animal, and human communities. Human disturbances such as the clearing of forests, extraction of mineral resources, and the construction of roads, buildings, and dams may lead to further changes in plant and animal communities, topography, hydrology, and socio-economic conditions. Whether natural or human processes cause disturbances, such changes directly impact people living in the Appalachian Highlands. Landscape disturbances affect the abundance and diversity of fish and game resources, drinking water quality and quantity, and the character of human communities. Landscape changes often occur gradually over time as a series of small, localized events. For these reasons, landscape changes are frequently unmanaged until significant changes occur to fish and game populations, air and water, and other valued resources.

In order to pro-actively manage landscape change, the Canaan Valley Institute (CVI) has adopted a landscape assessment approach based on ecological principles to compliment CVI's ongoing efforts to address watershed-related issues. CVI’s landscape assessment approach incorporates ecological principles and techniques derived from the science of “landscape ecology”. Landscape ecology is the study of the ecological and societal causes and consequences of variation in the landscape (Klopatek and Gardner, 1999). Using satellite imagery, aerial photography, and geographic information systems computer technology, landscape ecologists are able to examine how the landscape has changed over time and how it is likely to change in the future. Once landscape changes are identified or predicted, the causes and the ecological and societal consequences of such changes can be examined.

CVI is currently experimenting with these techniques as tools to evaluate broad-scale environmental disturbances in the Mid-Atlantic highlands. CVI selected the Central Appalachian ecoregion as the initial area to prototype the use of the landscape assessment process and tools. The Central Appalachian ecoregion[i][4] extends through the center of CVI’s service area and through the heart of Appalachia, overlapping the boundaries between West Virginia, Virginia, Kentucky, and Tennessee (Figure 1). The Central Appalachians is largely forested and is dominated by two types of disturbances: timber harvesting and surface mining. Currently, the most extensive landscape perturbations in the region are those associated with the practice of “mountaintop” mining. This paper presents the preliminary results of an analysis of the cumulative impacts of mountaintop mining activities in the Central Appalachians and adjacent ecoregions. The landscape assessment process will be outlined followed

Background

Mountaintop mining involves the removal of large quantities of soil and parent material from the tops of mountains to expose underlying coal seams. While some of the overburden material is retained for reclamation purposes, the majority is deposited in adjacent hollows and coves creating “valley fills”. These activities directly impact terrestrial ecosystems through converting significant areas of forest to grassland and through changes in the topography. Valley fills also eliminate miles of ephemeral, intermittent, and perennial streams that may in turn result in decreased organic matter inputs and increased stream sediment loads. Local residents have complained of problems with dust, noise, and vibrations associated with blasting activities and restricted access to traditional hunting and gathering sites. Mountaintop mining and valley fill activities may have equally significant but less obvious cumulative impacts affecting entire watersheds and ecoregions.

Typically, multiple mines and numerous valley fills may be proposed in an individual watershed. The elimination of miles of headwater streams may result in cumulative downstream impacts caused by increased suspended sediments, decreased organic matter inputs, and hydrologic changes. Cumulative impacts from changes in topography and land cover may result in the elimination of large tracts of habitat for native forest-interior species, the invasion of exotic plant, animal, and insect species, and micro-climatic changes.

Mountain top mining and valley fill activities might also produce positive economic and environmental benefits. Proponents of mining are sure to point out the creation of jobs for local residents; the creation of hundreds of miles of access roads and gently sloping land for future development; the conversion of intermittent to perennial flow in headwater streams; and the creation of wildlife habitat for deer and other generalist game species.

The project study area (Figure 2) comprises about 675 watersheds and expands through southern West Virginia, southwest Virginia, and eastern Kentucky and Tennessee. The majority of all mining activity is restricted to the Central Appalachian ecoregion near the border of West Virginia and Kentucky. In total, mountaintop mining and valley fill activities have been proposed or are ongoing in 245 watersheds[ii][3] in Appalachia. The 1993 National Land Cover Data Set shows over 300,000 registered as either surface mines, quarries, or gravel pits.

In West Virginia alone, approximately 200,000 acres of land was permitted for mountaintop mining operations between 1980 and 1999. As of 1999, there were approximately 11,500 permitted valley fills in the region encompassing about 100,000 acres and potentially impacting 1,400 miles of headwater streams, though not all of these permitted valley fills have been constructed[iii][1]. Due to the extent of current and proposed future mountaintop mining activities in the Central Appalachians, federal agencies have been charged to prepare a Programmatic Environmental Impact Statement (PEIS) on all past, present and future mountaintop mining activities. In preparing the PEIS, state and federal agencies have compiled a wealth of publicly available data on the study area. CVI relied on this data in addition to other publicly available datasets in order to analyze the cumulative impacts from mountaintop mining.

Methods

To assess the cumulative impacts from mountaintop mining and valley fill activities, CVI followed a landscape assessment process consisting of ten major steps (Table 1). The first two steps were accomplished mostly through public meetings held as part of the scoping phase of the PEIS. CVI participated in some of those meetings and

Table 1. Landscape Assessment Process

1. Meet with watershed stakeholders to determine issues and goals
2. Define the study area
3. Identify environmental and socio-economic endpoints of concern
4. Identify relationships between endpoints, stressors, and landscape condition
5. Obtain and refine data
6. Characterize the study area
7. Model landscape changes
8. Evaluate impacts to endpoints of concern
9. Develop and model alternatives to achieve goals
10. Provide results to stakeholders

met additionally with staff from the West Virginia Department of Natural Resources, and with researchers from the Penn State Cooperative Wetlands Center and the University of Maryland’s Appalachian Laboratory. Based on feedback received during these meetings, CVI was able to identify several ecological endpoints of concern in the study area. Ecological endpoints of concern are those species, groups of species, and/or ecosystem processes and functions that are of concern to stakeholders and at risk of being degraded by landscape disturbances. The following ecological endpoints were identified: habitat generalist species (e.g., bobcat, black bear, white-tailed deer, wild turkey, grouse); interior-forest species (e.g., pileated woodpecker); water quality; fish and bait species (e.g., smallmouth bass, minnows); salamanders; and non-timber forest products (e.g., ginseng, wild leeks, morels) were identified. For this preliminary assessment, habitat generalists, interior-forest species, and water quality were selected as the initial ecological endpoints used to focus the study. The other endpoints identified above will be used in future analyses of mountaintop mining.

After identifying endpoints of concern, the next step in the process is to determine the relationship between the ecological endpoints and landscape conditions. For this purpose, CVI reviewed the scientific literature to identify landscape indicators relating forest loss with water quality and terrestrial and aquatic species. Landscape indicators may be direct measures of economic, social, or environmental parameters (e.g., household income, ethnicity, temperature); aggregations or weighted combinations of parameters (e.g., Gross National Product, Index of Biological Integrity); or indirect measures based on established relationships with individual parameters (e.g., the amount of agricultural land in a watershed as an indirect estimate of non-point source nutrient loadings to streams). Landscape indicators can be used to evaluate and monitor the overall condition of ecosystems.

To identify landscape indicators relevant to the endpoints of concern, CVI also drew from the array of landscape indicators developed by the U.S. Environmental Protection Agency (Jones, et al., 1997). The U.S. Environmental Protection Agency used thirty-three landscape indicators to characterize the environmental condition of the mid-Atlantic region (Ibid). Almost half of these measures (15) are indicators of terrestrial ecosystem integrity and nine are measures of water quality and aquatic ecosystem integrity. The remaining measures serve as indicators of demographic pressure, air quality, erosion potential, and soil loss.

In order to analyze the cumulative impacts from mountaintop mining, CVI selected ten of the indicators used by the USEPA (Table 2). The ten indicators were selected based on their relevancy to the ecological endpoints of concern and based on their sensitivity to land use changes resulting from mountaintop mining activities. The Institute used two additional landscape indicators (i.e., potential ecological condition, streams through mines) that are particularly sensitive to the types of impacts caused by mountaintop mining. In addition to the landscape indicators, CVI relied on five descriptive measures to characterize the study area under current and future land use conditions (Table 3).

Table 2. Landscape indicators used in mountaintop mining case study and related ecological endpoints (G=game species, F=fish and bait species, S=salamanders, B=interior-forest birds, P=non-timber forest products) .

Forest-related indicators	Ecol. EndPts.	Water quality- related indicators	Ecol. EndPts.	Other indicators	Ecol. EndPts.
Percent forest cover*	All	Roads along streams*	F, S	Population density/ wtrshd (1990)*	All
Forest edge*	G, B	Streams through forests*	F, S	Population change/ wtrshd 1970 -1990*	All
Potential ecological condition	All	Streams through mines	F, S	Road density*	All
Interior forest (16 acre min. patch)*	G, B			Human use index*	All
Interior forest (145 acre min. patch)*	G, B

* Landscape indicators used by the USEPA (Jones, et al., 1997).

Table 3. Descriptive measures of the study area

Stream length

Stream length directly impacted by mining

Land use/ land cover (histograms, pie charts)

Slope

Mined area per watershed

The above listed indicators were used to define the data needs for the study. The ten indicators depend on seven critical datasets: land cover, elevation, streams, roads, watersheds, ecoregions, and mine permit and valley fill polygons. All of these data were provided in Esri ArcView shape file or ArcInfo Interchange formats. Esri’s ArcView 3.2 with the Spatial Analyst 2.0 extension were used to calculate the majority of the indicators. Esri’s ArcInfo 8.0 was used to calculate focal statistics and other processor intensive calculations.

Land cover data were used from The National Land Cover Dataset (NLCD). The NLCD was developed from 30-meter Landsat Thematic Mapper data by the Multi-resolution Land Characterization (MRLC) Consortium. The Consortium includes the U.S. Geological Survey, U.S. Environmental Protection Agency, U.S. Forest Service, and the National Oceanic and Atmospheric Administration. The Landsat TM scenes used in the NLCD classification for the study area include leaf-off imagery from 1993, 1994, and 1997 and leaf-on scenes from 1992 and 1994. The NLCD differentiates sixteen cover types in the study area (Table 4).

Table 4. National Land Cover Data Classes

11. Open Water

21. Low Intensity Residential

22. High Intensity Residential

23. Commercial

31. Bare Rock/ Sand/ Clay

32. Quarries/ Strip mines/ Gravel Pits

33. Transitional

41. Deciduous Forest

42. Evergreen Forest

43. Mixed Forest

81. Pasture/ Hay

82. Row Crops

83. Small Grain

85. Urban/ Recreational Grasses

91. Woody Wetlands

92. Emergent Herbaceous Wetlands

Digital elevation data were derived from the USGS National Elevation Dataset with a 30m resolution. Stream and road data were derived from 1:100,000 scale Wessex Tiger files. Watershed delineations for West Virginia, Kentucky, Virginia, and Tennessee were based on the Natural Resource Conservation Service’s 11-digit code hydrologic units. Discrepancies between watershed boundaries along state lines were resolved through a visual inspection and on-screen rectification of boundaries using the stream network and digital elevation model to guide revisions. Ecoregion maps published by Omernik (1995) were used to group watersheds in the study according to ecological region. Omernik’s ecoregions were used over those published by Bailey and others because Omernik incorporated current land use characteristics as part of his ecoregion identification criteria (Omernik, pers. com., 12/1999).

Valley fill and mine permit polygons for the study area were provided by individual state agencies and distributed by the US Office of Surface Mining (OSM). The valley fill data represent all fills recorded in the study area from April 1981 to December 1999. A small percentage of these fills are associated with refuse piles and other non-mountaintop mining activities. Mountaintop mining permit boundaries were provided by OSM for all mines in West Virginia that were permitted from December 1980 to December 1999. There are an estimated 57 to 100 additional permits scheduled to be issued in West Virginia which were not included in the permit data analyzed in this study. OSM is in the process of compiling permit data for Virginia, Tennessee, and Kentucky.

To calculate the ten indicators, CVI generally followed the methods outlined in Jones et al. (1997). These methods and a justification for selecting each indicator are provided in Appendix A. After calculating each indicator, the results were summarized by watershed and all watersheds were grouped by ecoregion[iv][2]. Since six of the ten indicators are forest based, grouping watersheds by ecoregion enabled the comparison of watersheds composed of similar forest types. For each watershed, the percent rank of that watershed relative to other watersheds in the same ecoregion was calculated for each indicator. For example, a watershed with a 75th percentile rank for the road density indicator means that the watershed has a higher road density than 75% of other watersheds in the same ecoregion.

The percent ranks of selected indicators were weighted and combined to form composite indicators for habitat generalists, interior-forest species, and water quality (Table 5). Due to the preliminary nature of the base data, the selection of indicators and

Table 5. Composite Indicator Formulas

Water Quality = ((3 * SF) + (1 * F) – (3 * SM) - (2 * RS) – (2 * HU)) + 3)/ 7

Habitat Generalists = ((1 * RD) + (2 * SF) + (3 * E))/ 6

Interior Forest Habitat = (3 * EC) + (2 * SF) + (1 * IF18) + (1 * IF200) + (1 * F) – (1 * E) – (2 * RD)

– (2 * SM) – (3 * HU) + 1

(E = % Edge, F = % Forest, EC = Potential Ecological Condition, HU = Human Use Index, IF18 = Interior Forest 18, IF200 = Interior Forest 200, RD = Road Density, RS = Roads along Streams, SF = Streams through Forests, SM = Streams through Mines)

weights pertaining to each composite indicator was completed by CVI using an iterative consensus process. For the final study, CVI will survey experts in the fields of wildlife management and water quality using the same iterative consensus process to select and weight indicators for each of three composite indicators. Additional composite indicators relative to fish, salamanders, and non-timber forest products will also be developed.

Eight of the ten landscape indicators and the three composite indicators were calculated under three different land cover scenarios. The eight indicators were selected based on their likelihood of changing as a result of additional mining activity and include: Percent forest cover, Percent edge, Interior forest 18, Interior forest 200, Potential ecological condition, Streams through forest, Streams through mines, and the Human use index (Appendix A). Scenario one represents baseline conditions depicted by the 1993 National Land Cover Dataset. Mountaintop mines and surface mines are most likely represented in class #32 which includes quarries, gravel pits and other surface mines. No attempt was made to differentiate mountaintop mines from other types of mines, pits, and quarries. Valley fill polygons were added to the NLCD to form the second land cover scenario. Permit area polygons were then combined with the valley fill polygons and added to the NLCD to form the third land cover scenario. Valley fills and permit areas were burned into the NLCD and classed as grassland. The second and third land cover scenarios are used to model the potential cumulative impacts to generalist species, interior-forest species, and water quality that might result if all permitted valley fills are constructed and all permitted mining areas are deforested.

Results

Mountaintop mining and valley fill activities are either on going or proposed in approximately 245 watersheds (11-digit Hydrologic Units) located in southern West Virginia, southwest Virginia, and eastern Kentucky and Tennessee. The majority of mountaintop mining activity is occurring in southern West Virginia and eastern Kentucky in the Central Appalachian ecoregion. Also impacted by mountaintop mining are the Western Allegheny Plateau, Central Appalachians, and Southwestern Appalachians as defined by Omernik (1995). The Central Appalachian ecoregion is a rugged plateau that is extensively forested with mixed mesophytic, Appalachian oak, and northern hardwood trees. Bituminous coal seams are embedded in the underlying rock composed of sandstone, shale, and conglomerate. Agricultural activities are very limited in this ecoregion due to the rugged terrain, cool climate, and infertile soils. The Western Allegheny Plateau ecoregion consists of hilly and wooded terrain composed of mixed mesophytic forests and mixed oak forests. Agricultural activities in the region are concentrated in the valleys and the horizontally bedded, sedimentary bedrock has been mined for bituminous coal. The Southwestern Appalachians is composed of low mountains. Mixed mesophytic forests are located in the deeper ravines and escarpment slopes while mixed oaks and short leaf pine dominate the upland forests. Other land uses in the region include cropland and pasture. Mountaintop mining is restricted to the northern portion of this ecoregion, where it abuts the Western Allegheny Plateau and Central Appalachians (Omernik 1999).

The dominant land cover in all three ecoregions is forest, particularly in the Central and Southwestern Appalachians. Agricultural lands and urban areas constitute significant land uses in the Western Allegheny Plateau, which includes southeastern Ohio and the city of Pittsburgh, Pennsylvania. Surface mining, quarries, and barren lands are most prevalent in the Central Appalachians, encompassing about 200,000 acres and occurring within 60% (162/272) of the watersheds in that ecoregion. Approximately 100,000 acres of mined, quarried, and barren land is located in 17% (60/351) of the watersheds in the Western Allegheny Plateau and just over 2200 acres of such lands are located in 45% (23/51) of the watersheds in the northern Southwestern Appalachians.

There are approximately 120,000 miles of streams[v][5] in the study area. About 1,400 miles of headwater streams are potentially impacted by existing or proposed valley fills. Eighty-nine percent (1,254/1,404) of the affected stream reaches fall within the Central Appalachian ecoregion.

The results of the baseline analysis for the three composite indicators are presented in Figures 3 - 5. In Figure 3, green areas represent watersheds with relatively high proportion of habitat for generalist species. In Figure 4, green areas represent watersheds with relatively high proportion of habitat for interior-forest species and in Figure 5, green areas represent watersheds with potentially better water quality. Conversely, red areas in Figures 3 – 5 represent watersheds with the opposite characteristics. These maps only depict the baseline conditions. Additional maps and tables representing land scenarios 2 and 3 will be presented at the Esri 2000 conference along with a discussion of the preliminary conclusions.

Appendix A Landscape Indicator Descriptions

Population density and change

Population density and change (1970 - 1990) serve as indicators of the degree of current human interaction with and modification of the landscape. Landscapes with dense human populations are likely to have been highly modified by humans and are typically classified as urban, residential, and agricultural land uses. Landscapes experiencing high population growth rates are likely to be undergoing land use change, e.g., from forest to agricultural/residential and from agricultural to urban/commercial.

Road density

Roads serve as animal movement corridors and both connect and fragment the landscape. Roads may disrupt, restrict, or enhance the dispersal of plants and animals. They are a source or conduit of noise, dust, and pollutant runoff. Roads improve human access to the land and facilitate development. The impact of roads on the landscape is a function of road size, traffic volume, and type of use.

Roads along streams

Streams and roads have been shown to have an important relationship at the landscape level. Roads constructed near watercourses pose potential risks to water quality and may alter stream hydrology. Run-off from roads contains vehicle-related chemicals and pollutants such as sediments and nutrients, oils and greases, salts etc. that may enter adjacent watercourses. Run-off from roads also may increase peak stream flows.

Streams through forest

Forested riparian areas stabilize stream banks, slow flood waters, and filter out some nutrients and pollutants from overland flow before they enter a stream. Streams through forested land uses have been shown to receive few pollutants from the landscape. They also receive the benefits of natural energy, pollution buffering, shading and food sources from the forests. Notwithstanding point sources of pollution, streams through forests are typically high quality streams.

Streams through mines

Streams that flow through valley fills will be virtually eliminated. Streams that flow through mine permit areas will likely be re-routed and/or degraded through increased sedimentation.

Percent forest cover

Forest cover is an indicator of stream condition and of human development. Forests protect the land from erosion and filter out some chemical contaminants. Local economies in heavily forested watersheds are typically based on forest use, extraction of timber and non-timber forest products, and tourism.

Forest edge

Fragmented forests have more edge habitat (areas along the boundaries between different types of land cover) than non-fragmented forests. Irregularly shaped forest patches have more edge habitat than regularly shaped forest patches. Small amounts of forest edge positioned naturally within the landscape can be beneficial to both the forest itself and wildlife. The edges provide ecotones where food sources, habitat, and energy sources are enhanced. The creation of more forest edge habitat often corresponds to an increase in local species diversity as “edge” species are attracted to the region. However, the creation of edge habitat can also lead to the elimination of forest interior species and the encroachment of disease and invasive exotic species. More often then not in systems disturbed by man, increased edge means decreased forest and wildlife values.

Too much edge in the wrong places provides access for forest pests, disease, unnatural predators, invasive plants and animals, and pollution.

Interior forest

Interior forest species require large tracts of land with continuous forest cover. Different species perceive interior forest differently- therefore interior forest habitat is assessed at three scales: 7 has., 65 has., and 600 has.

Potential Ecological condition

Indices of biotic integrity measure the capability of an area to support and maintain “a balanced, integrated, adaptive community of organisms comparable to that of relatively undisturbed habitats of the region. O’Connell et al. (1998) developed a Bird Community Index (BCI) as an indicator of habitat disturbance applicable to the Mid-Atlantic Highlands region which encompasses the entire state of West Virginia and the western portions of Virginia. The BCI is based on response guilds of songbirds, e.g., groups of birds requiring similar habitat, prey, and other conditions for survival. The relative proportions of specialists and generalists represented in each response guild serve as indicators of ecosystem condition. O’Connell et al., found a high correlation between the types of response guilds found in an area and the level of human impact.

Human Use Index

The human use index represents the proportion of watershed area with urban, agriculture, mining, transitional, and grassland land uses. High values of the human use index are indicative of environmental problems due to contaminated runoff, habitat loss, and other human-caused disturbances.

References

Bailey, R.G. 1995. Descriptions of the Ecoregions of the United States, 2^nd edition. Miscellaneous Publication No. 1391, U.S. Department of Agriculture, Forest Service, Washington, D.C., 108 pp.

Bryce, S.A., J.M. Omernik, and D.P. Larsen. 1999. Ecoregions: A geographic framework to guide risk characterization and ecosystem management. Environmental Practice 1(3): 1-15.

Cleland, D.T., P.E. Avers, W.H. McNab, M.E. Jensen, R.G. Bailey, T.King, and W.E. Russell. 1997. National hierarchical framework of ecological units. In: Boyce, M.S. and A. Haney (eds.), Ecosystem Management Applications for Sustainable Forest and Wildlife Resources. Yale University Press: New Haven, CT, 181-200.

Jones, K.B., K.H. Riitters, J.D. Wickham, R.D. Tankersley, Jr., R.V. O’Neill, D.J. Chaloud, E.R. Smith, and A.C. Neale. 1997. An ecological assessment of the United States Mid-Atlantic region: A landscape atlas. EPA/600/R-97/130. U.S. Environmental Protection Agency: Washington, D.C.

Klopatek, J.M., and R.H. Gardner, eds. 1999. Landscape Ecological Analysis: Issues and applications. Springer-Verlag: New York.

O’Connel, T.J., L.E. Jackson, and R.P. Brooks. 1998. The bird community index: a tool for assessing biotic integrity in the Mid-Atlantic Highlands. State College, PA: Penn State Cooperative Wetlands Center. No. 98-4.

Omernik, J.M. 1995. Ecoregions: A framework for environmental management. In: W. Davis and T. Simon (eds.), Biological Assessment and Criteria: Tools for Water Resource Planning and Decision Making. Lewis Publishers: Boca Raton, FL.

Omernik, J.M. [1999]. Primary distinguishing characteristics of level III ecoregions of the Continental United States, forthcoming.

Omernik, J.M. 1999. Personal conversation with J.R. Pomponio and P.R. Claggett, 16 December.