Topographic Controls on the Duration of Sinkhole Flooding in Central Tennessee, USA

Abstract

Abstract Daily 3-m cell size Dove multispectral satellite images were used to estimate the duration of flooding in topographic depressions in the approximately 12-km2 Shores Road focus area, central Tennessee. Flooding happened after rainfall events February 4–6 and 10–12, 2020. The filling and draining of 60 topographic depressions (mostly sinkholes) were observed by visually inspecting images for eight dates. The average duration of inundation was approximately 15 days, and all depressions were dry 30 days after the end of the first event. Satellite observations were consistent with ground-based water elevation measurements in three sinkholes and eight wells. Depression volume, elevation, and depth were estimated with a 0.76-m cell size LiDAR 3D Elevation Program (3DEP) digital terrain model (DTM). Long-duration (21 or more days) inundation was statistically more likely in larger-volume, lower-elevation, and deeper depressions at the p < 0.01 level. When combined in a probit model, these three factors accurately sorted 80 percent of depressions into long-duration and short-duration categories, and the accurately classified depressions accounted for approximately 95 percent of depression volume. Duration of inundation was not related (p > 0.05) to percentage of depression covered by slow-permeability soils, as determined from the Soil Survey Geographic Database. This study shows how to use repeat Dove satellite imagery and a LiDAR 3DEP DTM to assess how multiple topographic factors contribute to the duration of inundation in sinkholes.