Space-Based Detection of Sinkhole Activities in Central Florida
Summary
Sinkhole activities, including ground subsidence or collapse, are major geologic hazards. In Florida alone, hundreds to thousands of subsidence cases are reported each year. Most occur in the covered karst of central Florida (Figure 1), where large anthropogenic changes in groundwater levels have increased the frequency of sinkhole activities (Tihansky, 1999). The main hazard induced by sinkhole activity is property damage. In August 2013 a sinkhole collapse destroyed a resort complex near Disney World. More minor events inflict a significant cumulative toll: the Florida insurance record for 2006-2010 indicates sinkhole related claims with a total indemnity of about $200 million per year (Florida Senate report, 2010). Sinkhole collapse can also cause a loss of life, as occurred on March 1st, 2013, when a resident of Seffner was "swallowed" by a sinkhole that opened beneath his bedroom.Monitoring sinkhole activities is a challenging task, because most of the activity occurs in the subsurface. Geophysical imaging techniques, as Ground Penetrating Radar (GPR), provide very useful images of the subsurface, but are practical only over small areas (hundreds of square meters) and cannot image the entire sinkhole active zone (thousands of square kilometers).
Sinkholes develop above dissolution cavities in Florida's thick carbonate deposits. The cavities have developed over geological time scales by dissolution processes, mostly during low sea level condition, which exposed the Florida peninsula to karst processes. The dissolution cavities serve as important conduits for groundwater movement. Newly developed sinkholes occur mainly in western-central Florida (Figure 2a), because the geological and hydrological conditions in the area enable rapid sinkhole growth. In this area a 10-60 m thick layer of clay and/or sand covers the carbonate deposits (Tihansky, 1999). The surface response to the limestone cavity depends on the properties of the cover sediments. When the cover consists mostly of sandy deposits, its cohesion is low, resulting in downward granular flow, or piping, into fractures and voids. Slow subsidence results in gentle depressions (Figure 2).
However, when the overlying layer consists mainly of cohesive clay deposits, the top of the roof erodes upward as sediments spall into the cavity (Figure 3) (Tihansky, 1999). The thinning coherent roof could theoretically also subside. Two mechanisms for advancing subsidence include decreasing flexural strength, as the roof thins, and poroelastic response to groundwater draining through the void. Eventually the void expands to the surface, resulting in abrupt surface collapse.
Both types of sinkholes can become major geological hazards, as they can cause severe damage to property and latter even loss of life, as mentioned above.
Methods
InSAR detection of sinkhole activities Space-borne InSAR is a very powerful technique that is capable of detecting small surface displacements (millimeter level) over wide areas. The method compares pixel-by-pixel SAR phase observations of the same area acquired at different times from roughly the same location in space to produce high spatial-resolution surface change maps. InSAR has been successfully used to monitor land subsidence both in urban (e.g., Amelung et al., 1999; Bawden et al., 2001; and rural areas (e.g., Baer et al., 2002; Galloway et al., 1998; Hoffmann et al., 2003) due to natural or anthropogenic causes, as groundwater withdrawal. A significant progress in InSAR technology was the development of permanent scatterer Interferometry (PSI) (Ferretti et al., 2000; 2001) and small baseline subset (SBAS) techniques (Berardino et al., 2002; Lanari et al., 2004), which use a large number of SAR observations to monitor displacement time-series using the successive InSAR observations (Figure 1). These techniques are very useful for monitoring slow and continuous deformation of the Earth's solid surface, in particular land subsidence. PSI works well in urban areas and was applied to monitor subsidence in many cities, as New Orleans (Dixon et al., 2006), Mexico City (Osmanoglu et al., 2012), Venice (Bock et al., 2012), and more. The technique is capable of detecting slow movements of individual structures, or even differential section of structure, as we observed in the subsidence of the Mexico Cathedral (Figure 1b - Osmanoglu et al., 2012). The SBAS technique is slightly less precise, but works well in both urban and open areas. It was applied to monitor land subsidence in many locations, as in Italy (Stramondo et al., 2007), Indonesia (Chaussard et al., 2013a), and Mexico (Chaussard et al., 2013b).
Although the InSAR technique has been available for two decades, it had limited success in detecting sinkhole activity. Several studies detected subsidence in sinkhole-prone areas, such as along the Dead Sea shores (Baer et al., 2002; Closson et al., 2005), Nevada (Al-Fares, 2005), Texas (Paine et al., 2009), Arizona (Conway and Cook, 2013), and Spain (Castaneda et al., 2009). However, successful detection of individual sinkhole subsidence occurred only over large sinkholes (area > 100x100 m2). This limited success reflects most likely the detection limitation of the first two generations of SAR satellites (ERS-1/2, JERS-1, Envisat, and ALOS), which acquire data with 10-20 m pixel resolution. A significant improvement in SAR satellite detection capability occurred when the third generation of SAR satellites after TerraSAR-X and Cosmo-SkyMed were launched. These current operating satellites have an amazing sub-meter level pixel resolution that is sufficient for detecting surface deformation within small and medium size sinkholes. Recently, Nof et al. (2013) detected precursory subsidence around individual sinkholes along the Dead Sea shores using high-resolution Cosmo-SkyMed observations. Their study shows a gradual surface subsidence in and around sinkholes several months prior to the sinkhole's collapse. The study of Nof et al. (2013) demonstrates that high spatial resolution SAR observations are crucial for detecting small, localized subsidence signal induced by sinkhole activity.