|M.Sc Student||Michal Tune|
|Subject||BOTDR Based System for Automated Detection of Sinkhole|
|Department||Department of Civil and Environmental Engineering||Supervisors||Professor Linker Raphael|
|Professor Klar Assaf|
|Full Thesis text|
Thousands of sinkholes have appeared along the Dead Sea coast since the 1980’s. The most recent hypotheses for explaining their formation correlate their development to the drop in the Dead Sea water level and associate them to subsurface cavities formed by salt layer dissolution .
The salt dissolution process is gradual and does not occur instantaneously. During this period, which may last months or even years, mechanical deformations are expected to develop in the soil. As a result of the salt dissolution, stresses are added to the non-dissolving layer. When these stresses reach a level that causes the soil to become unstable, collapse occurs. The sudden collapse can lead to property damage and loss of life. Consequently, it is of interest to develop tools to detect sinkhole development sufficiently early before collapse occurs .
The objective of the present study was to determine the feasibility of a Brillouin optical time domain reflectometer (BOTDR) based system for automated detection of sinkhole formation. In order to develop and evaluate the potential performance of such a system the study included two parts: a numerical simulation based on sinkhole models and experimental field measurements for disturbance evaluation .
The numerical simulations used elastic soil models which were shown to be sufficient to estimate the strain signal that a developing sinkhole is expected to produce in an optic fiber. Simulations of more realistic signals were obtained by superimposing disturbance signals recorded during field experiments over ideal sinkhole signals. Wavelet decomposition was applied in order to filter out the disturbances and extract the sinkhole contribution. The analysis was based on signals over a range of 400m (with increment of 0.1m) and a subset of wavelet coefficients most affected by the sinkhole formation was identified. A detection criterion based on these coefficients and their integral over a 40m moving window was developed. This approach led to detection of spherical cavities with radii ranging from 2 to 6 m for cover depth ranging from 10 to 50 m, without any false alarm. Detection was achieved almost regardless of the subsurface cavity height (for cavity heights larger than 2m .(
In order to determine whether detection occurs prior to collapse, the stability of the subsurface voids was evaluated via the so-called "factor of safety". This analysis showed that detection occurred while the subsurface cavities were still stable (factor of safety 1.7-2.5) and early enough to enable the implementation of countermeasures.