Time Dependent Mining- Induced Subsidence Measured by DInSAR Jessica M. Wempen 7/31/2014 Michael K. McCarter 1

Introduction – Mining Region  The Green River Basin of Wyoming has significant resources of trona, an evaporite mineral used for the production of soda ash.  Trona production in this region has historically been carried out using traditional underground coal mining techniques with mining depths ranging from 245 to 460 m.  Currently there are five operations: two longwall, two room and pillar, and one solution. 2

Introduction – Subsidence 3  As the trona mines utilize longwall mining more extensively and progress toward higher extraction ratios the potential for subsidence increases.  In this region, subsidence impacts a relatively large area, and has the potential to affect future land use and local hydrology, and to damage structures.  Traditional methods of measuring subsidence, including GPS and aerial surveys, can have millimeter level accuracy, but the spatial extent is limited.

Introduction – DInSAR  D ifferential In terferometric S ynthetic A perture R adar is a satellite based remote sensing process that can generate subsidence data on a regional scale with a high data density.  DInSAR has the potential to quantify the impact of subsidence over large regions and long time scales.  DInSAR could be a useful method for enhancing traditional subsidence surveys. 4

Data – ALOS  The Japanese Satellite ALOS (Jan 2006 – Apr 2011) collected SAR data that is suitable for subsidence monitoring.  ALOS had a Phased Array type L-band Synthetic Aperture Radar, with a wavelength of 24 cm.  L-band provides good penetration of atmosphere and ground cover, and produces data with high coherence.  In this study, nine scenes acquired by ALOS over the period from Dec 2007 to Mar 2011were processed. 5

DInSAR – Process  DInSAR works by measuring the phase shift of a radar signal between two scenes acquired over the same area at different times.  When two scenes are paired, an interferogram that contains the phase shift data is generated.  The interferogram includes contributions from the topography, changes in the satellite position, ground movement, atmospheric affects and noise.  6

DInSAR – Process  In DInSAR, the phase shift caused by ground movement is isolated by removing or reducing the phase shift caused by topography, changes in the satellite position, atmospheric affects and noise.  Accuracy of DInSAR is most negatively affected by significant topographic relief, long temporal baselines, and large spatial baselines. 7

Measuring Subsidence  Progressive DInSAR derived displacement maps for four of the trona mines are presented in the following slides.  Subsidence troughs generated by longwall mining are clearly identifiable.  Measurable subsidence was also produced by solution mining. 8

Dec 4, 2007 – Jul 21,

Dec 4, 2007 – Oct 24,

Dec 4, 2007 – Jun 11,

Dec 4, 2007 – Jul 27,

Dec 4, 2007 – Sep 11,

Dec 4, 2007 – Oct 27,

Dec 4, 2007 – Jan 1,

Dec 4, 2007 – Mar 14,

Measuring Subsidence  From Dec 2007 to Mar 2011, longwall mining generated a maximum subsidence of 1.3 m.  Over the same period, solution mining produced a maximum subsidence of 0.5 m.  Since publication, the location and magnitude of the maximum longwall subsidence have been validated. The DInSAR results are within centimeters of a GPS survey. 17

Image Quality – Decorrelation  Image quality is affected by the temporal and spatial decorrelation.  Temporal decorrelation is caused by variation in ground cover over time. It can be influenced by vegetation growth, snow, or other changes on the surface.  Baseline decorrelation is cause by significant changes in the satellite orbital positions. 18

Image Quality  Long elapsed times between pairs of data lead to large differences in the path of the satellite and high change gradients on the surface, which can lead to poor quality maps.  High quality, long-term displacement maps can be generated by pairing SAR data over short periods and summing the displacements. 19

Dec 4, 2007 – Mar 14, Paired Cumulative

Image Quality  There is a significant difference between the maximum subsidence in the paired map and the cumulative map: Maximum subsidence is 0.79 m for the paired data. Maximum subsidence is 1.30 m for the cumulative data.  In the paired data, processing errors generated by the large temporal and spatial baselines reduce the image quality.  Overall, the quality of the cumulative data is better and this is reflected by the data coherence. 21

Image Quality – Coherence 22

Conclusions  DInSAR can be useful for evaluating subsidence over large regions with relatively long time scales.  The elapsed time between SAR pairs can significantly affect the accuracy of the data but high quality displacement maps can be generated by pairing SAR data over short periods and summing the data.  Displacement maps derived from DInSAR can quantify the development of subsidence over time, the impact of subsidence in a mining region, and identify when subsidence is complete. 23