ABSTRACT
The sinkhole distribution in Winona County, Minnesota was first mapped by Dahlgleish and Alexander (1984). They used the distribution of 535 sinkholes to define sinkhole probability zones. In 1992, a new survey identified an additional 73 sinkholes — 39 older sinkholes missed in the first survey and 34 new sinkholes that have developed since 1983. All 73 sinkholes are located in the two highest sinkhole probability zones, supporting the criteria by which the original zones were delineated.
The mean density of the Winona County sinkholes is 0.52 sinkholes/km2 in the area underlain by the Prairie du Chien. Sinkholes only occur over the Prairie du Chien Group. A moving window analysis, using a 1.69 km2 window and a 0.1 km step, was applied to determine the spatial distribution of sinkhole densities. The resulting sinkhole densities were contoured and range from 0 to 14.8 sinkholes/km2.
A nearest-neighbor analysis was applied to the sinkhole data set. The observed mean distance between nearest-neighbors is 385 m. The range of nearest neighbor distances is 6 m to 3999 m. Test statistics by Skellam (1952) and Clark and Evans (1954) were used to test the nearest-neighbor distances against the values expected from a spatially random data set. Random data sets were created and analyzed for comparison. The observed sinkhole locations are significantly different than a random distribution in visual spatial distribution, in nearest-neighbor distance histograms, and in tests for randomness. With near certainty, the observed sinkholes are clustered. The observed mean nearest-neighbor distance is about half the distance expected for a random population with the same number of sinkholes.
The 34 new sinkholes often develop in clusters and their nearest neighbors are usually another new sinkhole. The observed directions to nearest neighbor do not have preferred orientations along regional joint sets and are similar to the direction distribution calculated from random data sets. Sinkhole genesis does not appear to be structurally controlled in Winona County. Bedrock lithology, surficial geology, and topography of the land surface appear to be significant controlling factors.
