Satellite monitoring of post-mining rehabilitation and the quantification of backfill dynamics using differential SAR interferometry

Date
2018-03
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Mining has severe environmental and social consequences and post-mining rehabilitation needs to take place before land re-zonation can occur. Post-mining rehabilitation is the process in which previously mined areas are returned to some degree of their natural state or to a condition that can sustain an intended post-mining land-use. Common rehabilitation practices, specifically in open-pit mines include backfilling and subsequent re-vegetation. Due to the risks involved with open-pit post-mining rehabilitation and to comply with the legislation governing mine closure, long-term monitoring of backfill settlement dynamics and the success of revegetation strategies are required. To overcome the limitations of field-based approaches, differential Synthetic Aperture Radar interferometry (dInSAR) has been used to detect measure and monitor surface deformation with millimetre accuracy. In the first objective of this study conventional dInSAR techniques were used to detect surface deformation in a rehabilitated backfilled open-pit mine, using multi-sensor Synthetic Aperture Radar (SAR) data. The results indicated that although deformation measurements could be extracted, some anomalous patterns were observed across X-, C-, and L- wavelengths. In particular, heave features were observed as opposed to subsidence. Potential reasons for anomalous features were investigated and could be attributed to uncompensated topography and signal noise due to the presence of vegetation. Based on the results, recommendations included an updated Digital Elevation Model (DEM) for the successful removal of topographic phase contributions and the use of more advanced dInSAR techniques. The second aspect of mine rehabilitation involves the revegetation of backfilled areas. Vegetation indices derived from multispectral data, such as the Enhanced Vegetation Index (EVI), have been widely used to monitor the productivity of vegetation. However, their use is limited in areas with persistent cloud cover and inclement weather. Due to the all-weather observations provided by SAR sensors, revegetation practices can be monitored using SAR data. This second objective evaluated the use of C-band SAR parameters to model EVI (simEVI) derived from Landsat 8 data using Random Forest (RF) algorithms. The results indicated that using a combination of all the SAR parameters yielded the highest adjusted R2 (=0.715). Overall,the results indicated that there was an overestimation of simEVI in the dry season and underestimation in the peak of the growing season. Limitations of this investigation included the known saturation effects of C-band as well as the statistical phenomenon known as regression dilution bias. In general, modelled SAR EVI results mimicked the EVI trends and can therefore be considered as an alternative in areas where persistent cloud cover prevents the use of Stellenbosch conventional multispectral EVI. However, with no or limited optical data, the modelled SAR EVI may not be suitable to monitor vegetation productivity in this investigation. Future research into the use of remote sensing to monitor post-mining rehabilitation is recommended and could provide valuable information to assist various stakeholders in improving post-mining rehabilitation strategies. The findings could contribute toward the operational monitoring of post-mining rehabilitation.
AFRIKAANSE OPSOMMING: Die resultate het aangedui dat die gebruik van 'n kombinasie van al die SAR parameters die hoogste aangepaste R2 opgelewer het (= 0.715). Oor die algeheel het die resultate aangedui dat daar 'n oorskatting van simEVI in die droë seisoen en ‘n onderskatting gedurende die piek van die groeiseisoen is. Beperkings van hierdie ondersoek sluit die bekende versadigings-effekte van C-band asook die statistiese verskynsel wat bekend staan as regressie verdunnings vooroordeel in. Oor die algemeen het gesimuleerde SAR EVI-resultate egter die EVI-tendense gevolg en kan daarom as 'n alternatief beskou word. Veral in gebiede waar aanhoudende wolkbedekking die gebruik van konvensionele multispektrale EVI verhoed. In die algemeen het SAR EVI resultate die EVI tendense gevolg en kan dus beskou word as ‘n alternatief in areas waar aanhoudende wolkbedekking die gebruik van konvensionele multispektrale EVI verhoed. Hierdie ondersoek toon egter dat gemodelleerde SAR EVI nie geskik is om plantegroei produktiwiteit te monitor in die afwesigheid van genoeg optiese data nie. Toekomstige navorsing oor die gebruik van afstandswaarneming om mynbou rehabilitasie te moniteer, word aanbeveel. Dit kan waardevolle inligting verskaf om verskeie belanghebbendes in die verbetering van die mynbou-rehabilitasie strategieë by te staan. Verder kan die bevindings tot die operasionele monitering van mynbou rehabilitasie bydra.
Description
Thesis (MA)--Stellenbosch University, 2018.
Keywords
Restoration ecology, Remote sensing images, Image processing -- Digital techniques, Signal processing, Differential SAR interferometry, Backfill dynamics, UCTD
Citation