Landslides triggered by the storm of November 3-5, 1985, Wills Mountain Anticline, West Virginia and Virginia

Abstract

More than 3,000 landslides were triggered by heavy rainfall in the central Appalachian Mountains of West Virginia and Virginia, November 3-5, 1985. These landslides provided the opportunity to study spatial controls on landslides, magnitude and frequency of triggering events, and the effects of landslides on flood-induced geomorphic change. The study area consists of parts of the Wills Mountain anticline, a major NE-trending structure in the central Appalachians, and a portion of the adjacent Appalachian Plateau. Across the anticline and adjacent plateau, bedrock lithologies vary markedly and include pure marine limestone, marine shale, deltaic mudstone/sandstone sequences, and orthoquartzites. Because of the geologic structure, bedrock lithology varies little along strike. The spatial distribution of landslides triggered by the storm was controlled primarily by rainfall, bedrock lithology, surficial lithology, land cover, and slope morphology. The triggering rainfall was of moderate intensity and long duration. Two-day storm totals varied from 170 mm to more than 240 mm in the study area. Most landslides occurred at the northeast end of the study area, where 48-h rainfall totals were in excess of 200 mm. Different rainfall thresholds are apparent for triggering landslides on different bedrock lithologies. The highest density of landslides occurred in shallow colluvium and residuum of the Reedsville Shale (Ordovician), followed by regolith of the Greenbriar and Mauch Chunk Groups (Mississippian). Most of the landslides in these fine-grained regoliths were shallow slides and slumps, many of which transformed to mudflows and delivered sediment directly to streams; a smaller number of debris avalanches were triggered high on quartzite ridges. Instability of colluvium and residuum derived from the Reedsville Shale, compared with regolith from four other fine-grained bedrock lithologies, is attributable to its low strength combined with moderate infiltration rates that allowed soil moisture to accumulate under the moderate intensities of the rainfall. Slopes covered by coarse, cobbly debris flow and alluvial deposits, mostly of Pleistocene age, were very stable due to their low slope angles and high frictional strength. For a particular bedrock lithology, the spatial distribution of landslides appears controlled by interdependent influences of slope morphology and land cover. On the Reedsville Shale, most landslides occurred on north- to northeast-facing slopes, which might have had higher antecedent levels of soil moisture; these slopes have also been preferentially cleared because they produce better pasture forage for livestock. A secondary concentration of landslides on south- to southwest-facing slopes cannot be explained by conventional soil-moisture models. Landslide density was 100--200 percent higher on cleared land than on forested land. On pastured land, most landslides occurred on laterally planar slopes, but on forested land, most landslides occurred in slope positions that were laterally concave (hillslope hollows). Compared with other documented Appalachian storms that have triggered landslides, the November 1985 storm had lower rainfall intensities over longer durations. Comparison with these other storms suggests that the anomalously high degree of slope instability in 1985 is due to the long duration of low-intensity rainfall on fine-grained regolith derived from shale; the triggering rainfall can be approximated by the 48-h storm total. Landslide density in Reedsville Shale regolith is linearly related to the varying 48-h rainfall along the anticline. These data define a probabilistic model that estimates return intervals of 43 to 300 yr for landslide densities ranging from 1 to 70 landslides/km 2 . Analysis of flood-induced geomorphic changes in 79 small drainage basins that received 210-240 mm of rainfall showed a clear local association between landslides and channel erosion or deposition adjacent to where the landslides delivered sediment to the stream. When channel change was quantified using an index evaluated at each basin mouth, most of the channel change was attributable to the influence of basin morphology on flood discharge. Landslide density in the basins was of secondary, although measurable, importance in explaining flood-induced channel changes at the basin scale.