University of Georgia Georgia Disasters & Water Resources When the Truth Sinks In: Assessing Sinkhole Development in Georgia [music] [Start Scene] >>Matthew: Sinkholes are a risk to humans, infrastructure and environmental health. Specifically, groundwater pollution in Dougherty County and other areas of coastal plains in the southeast of United States. This project examines the formation of two different types of sinkholes: cover-subsidence--larger scale and cover-collapse--smaller scale [music] >>Kimberly: We utilized a time-series of digital elevation models to complete the first objective by compiling sinkhole inventory maps from the study period of 1999 to 2011. >>Kimberly: DEM data from NASA’s Satellite Terra, and SRTM were used to detect cover-subsidence sinkhole development. >>Kimberly: We produced an algorithm using ArcGIS ModelBuilder and Python code to identify possible sinkholes. Our team then manually eliminated any falsely identified sinkholes in urban areas or certain bodies of water by using high resolution imagery as a reference. >>Mohamed: To further analysis of the project, we decided to perform an on-site investigation of sinkholes by visiting Albany Utilities. There, we met up with the water production manager Jim Stolze, who showed us various sinkholes within the vicinity of the property. >>Mohamed: We geared up and took a hike within the wooded areas nearby, viewing and measuring different sinkholes. We learned that the majority of these sinkholes formed after rainstorms or flooding events. After our hike, we were given a map of all other human validated sinkholes around the property. We used this map to add more sinkholes to our sinkhole inventory maps that we didn’t find from our original analysis and deleted any false positives that we didn’t see on site. >>Kimberly: Next, we generated temporal difference maps, which displayed new or expanded sinkholes in the study period. These maps were later used as inputs to create sinkhole density maps, which showed the frequency of sinkholes within the study area. >>Tunan: The project completed sinkhole inventory maps, identified the primary factors influencing sinkhole development, and determined areas vulnerable to future sinkhole formation. >>Mohamed: The first analysis we performed was a ordinary least squares, which showed which independent variables were more influential in the formation of sinkhole as you can see from the chart over here. The second analysis we performed was a geographically weighted regression, we showed how much independent variables as a whole describe all the sinkhole formations >>Matthew: synthesizing the results from our regression analysis the team produced sinkhole susceptibility map where should pick the probability of the future sinkhole occurrence. This map is showing the sinkhole density then we look at the measured sinkhole density maps, we see the sinkhole susceptibility and the sinkhole density measured maps compare quite well. >>Matthew: Therefore the sinkhole susceptibility map can be use to help build capacity for end users and decision making processes >>Wenjing: Overall our model explained 80 to 90 percent of sinkhole formation, the sinkhole prediction maps compared well to the measured sinkhole results, we have achieved our tasks as sinkhole chasers! >>Matthew: Georgia disasters and water resources team would like to thank our project partners: Paul Forgey >>Mohamed: Randy Weathersby >>Wenjing: Jim Stolze [credits] [end]