Cahyo, FA, Dwitya, R & Musa, RH 2020, 'New approach to detect imminent slope failure by utilising coherence 
attribute measurement on ground-based slope radar', in PM Dight (ed.), Slope Stability 2020: Proceedings of the 2020 International Symposium on Slope Stability in Open Pit Mining and Civil Engineering, Australian Centre for Geomechanics, Perth, pp. 1401-1412, https://doi.org/10.36487/ACG_repo/2025_96
(https://papers.acg.uwa.edu.au/p/2025_96_Cahyo/)
Abstract: Detecting imminent failure is immensely important in slope monitoring, especially in open pit mining where the hazard potential can severely affect safety and production. The current, well-accepted method to detect an imminent failure is to monitor the slope behaviour in the wall and a prediction can be made from the interpretation of progressive slope deformation behaviour prior to collapse. This paper will convey a new approach that can be developed to detect an imminent failure with precise results by utilising the coherence attribute produced by the ground-based slope radar. Coherence in ground-based slope radar implies the normalised complex cross-correlation function of amplitude and range measurements. The coherence measurement can be utilised to determine how much the wall surface on each pixel has changed between successive radar scans. Development of slope deformation from a progressive trend into failure event will yield increasing percentage of low coherence pixel over time.
This paper summarises the result of the back-analysis of three slope failures, where the coherence value of each pixel on the area that had undergone failure is retrieved and arranged into three coherence threshold values of 0.990, 0.995, and 0.999. Then each group is summed to know the number of pixels and subsequently, the percentage of total failure areas which have coherence values shifting below those three coherence thresholds over the time of radar scans. A thorough analysis toward the shifting of low coherence pixels percentage over time, which generally shows a progressive trend on the chart, could detect the imminent failure and provide the notification prior to the real failure with varied time-lapse from 0.8 hours to more than 12 hours based on the selected study cases. Certain interpretation towards this coherence method, by taking into account its limitations as well, helps to ensure detection of imminent failure as well as complement other common failure detection methods that would otherwise be more complex and uncertain without it.

Keywords: failure, slope stability radar, coherence, range, amplitude, low coherence pixels, alarm threshold