Subsidence Detection Using Integrated Multi Temporal Airborne Imagery

Multi temporal aerial photography and airborne hyper spectral imagery have been integrated for the detection and monitoring of coal mining subsidence hazards. Digital elevation models derived from successive epochs of aerial photography provide estimates of topographic change which may be indicative of the collapse of abandoned underground mine workings in the study area. Ground disturbed by subsidence can also be identified in hyper spectral imagery from soil moisture anomalies or vegetation stress. Archive photography originally acquired for topographic mapping over the last forty years was scanned and processed in a digital photogrammetric workstation. Since uncertainties in surface stability preclude the use of conventional ground control points for controlling historic photogrammetric models, each model was processed only to relative orientation stage. A control surface, created from a photogrammetric model comprising present-day imagery with contemporary ground control, was used in conjunction with a surface matching algorithm to provide the absolute orientation for each archive model. Subsidence features were then identified by subtracting the control DEM from each of the archive DEMs. Three epochs of airborne hyper spectral CASI and ATM imagery were acquired for the study area during a twelve month period. In the vegetated areas the Red Edge Position (REP) and parameters of the chlorophyll absorption feature were mapped. In the areas identified as exposed soil the thermal band of the ATM imagery is enhanced to show soil moisture variations. The results of the photogrammetric and hyper spectral processing were integrated to produce a subsidence hazard map of the study area. * Corresponding author.