Research Leading to Forecasting of Sinkholes using Satellite Data

Karst subsidence or localized cover collapse of subsurface cavities (sinkholes) is a severe hazard in the United States and can be found in all 50 states. Although occurrences of subsidence or sinkholes are challenging to predict, karst areas are more susceptible and conducive to the formation of cover collapse sinkholes. With the expectation of the construction industry to increase by 5% every year, the need to build structures and roadways on undesired soil will also increase. Geospatial techniques can be used to predict the risk of sinkhole occurrence by developing a Sinkhole Hazard Assessment for Situational Awareness (SHASA) Model (Gutierrez et al., 2008). Satellite-based precipitation estimates combined with a susceptibility map, developed with the proper instinctive properties, can provide critical understanding into the initiation of subsurface anomalies like sinkhole phenomenon. The SHASA will be developed to provide near real-time assessment of the probable potential for sinkhole triggering during times of extreme hydrological change (rainfall and soil moisture) and moderate to very high susceptibility variables (geological data, ground–based precipitation and soil moisture data, ambient and soil temperature, and hydrologic data). Combined with deep learning modeling, this research will develop long term capabilities to predict future distribution of sinkholes, which can be used by land-use managers for designing hazard mitigation plans.