Subsidence hazard prediction for limestone terrains, as applied to the English Cretaceous Chalk

Abstract Soluble carbonate rocks often pose a subsidence hazard to engineering and building works, due to the presence of either metastable natural solution features or artificial cavities. There is also an inherent danger to the public and lives have been lost because of unexpected ground collapses. Although site investigation techniques are becoming increasingly elaborate, the detection of hazardous ground conditions associated with limestones is frequently difficult and unreliable. Remedial measures to solve subsidence problems following foundation failure are expensive. It would be advantageous if areas liable to subsidence could be identified in a cost-effective manner in advance of planning and ground investigation. Hazard mapping could then be used by planners when checking the geotechnical suitability of a proposed development or by engineering geologists/geotechnical engineers to design the type of ground investigation best suited to the nature and scale of the potential hazard. Recent research focussed on the English Chalk outcrop has led to the development of two new models to predict the subsidence hazard for both natural solution features and artificial cavities. The predictive models can be used to map the hazard at any given chalkland locality, as a cost-effective precursor to ground investigation. The models, although created for the Chalk outcrop, have important implications for all types of limestone terrain. The basis of the predictive modelling procedure is an analysis of the spatial distribution of nearly 1600 natural solution features, and more than 850 artificial cavity locations, identified from a wide varietyy of sources, including a special appeal organized by CIRIA. A range of geological, hydrogeological and geomorphological factors are evaluated to identify significant relationships with subsidence. These factors are ranked, numerically weighted and incorporated into two quantitative subsidence hazard model formulae. The models can be applied to perform hazard mapping.