Th.e presence of mining-related cavities (workingS, shafts and tunnels) or karstic (solution cavities and sinkholes in limestone) within the top 100m in the rock mass restricts land utilisation, and their migration to the surface may damage property or services or cause loss of life. Confirmation of features marked on existing plans prior to design and construction may be sufficient but it is often necessary to determine the detailed sub-surface structure. The standard method of site investigation is to drill a pattern of boreholes to locate the spatial extent of any cavities. However, unless the spacing is less than the cavity dimensions it is possible to miss it completely. A cavity may be filled with air, water, or collapse material resulting in a contrast in physical properties which may be detected using appropriate geophysical methods. One powerful technique is microgravity which locates areas of contrasting sub-surface density from surface measurements of the earth's gravity. Although the method is fundamentally simple, measurement of the minute variations in gravity (1 in 108) requires sensitive instruments, careful data acquisition, and data reduction and digital data analysis. Final interpretation must be performed in conjunction with independent information about the site's history and geology. This paper presents three examples in both mining and karstic environments demonstrating that microgravity is a very effective technique for detecting and delineating cavities in the sub-surface.

During the past three years a network of six permanent surface seismometer stations has been established over the Cynheidre workings. In parallel with this an 8-channel on-line microprocessor-based event detector has been developed in order to provide warnings of high rates of activity.The main results of this twofold approach are as follows:Considerable normal microseismic activity is generated from within Cynheidre with an excellent correlation between the patterns of daily extraction and the levels of activity which can exceed 600 events per shift. It is thought these events principally originate from failure of the roof cantilever and the gob. They have impulsive onsets and durations of 0.5-1.0 s.Observations during outbursts suggest a completely different type of activity. The envelope of the events is very emergent and almost monochromatic with a frequency of about 30 Hz. These events appear to originate from within the coal, to be related to the emission of methane and show clear peaks in activity immediately prior to an outburst. In fact this type of activity is extremely rare except before periods of gas emission or an outburst.The microprocessor has clearly shown increases in the automatically picked events prior to the outburst of 24 June 1985 particularly on the station adjacent to the event. The microprocessor also gave an alarm prior to the recent heavy gas emission (8,500 m3) associated with ‘pouncing’ (20 February 1986) which was controlled successfully without proceeding to a full outburst.

This case study was undertaken for a low-porosity fractured carbonate reservoir with a complex fracture network resulting from several phases of tectonic activity. The integration of the image log and seismic-derived interpretations was problematic due to the complexity of the image log signature and the variable quality of the surface seismic data. Earlier experience indicated that VSPs may provide information on faulting and/or fracturing that may otherwise be difficult to determine with confidence from other data sources. Consequently, specialist VSP processing techniques were used to identify and map reflectors in three-dimensional space. Data acquired in two wells were reprocessed to interpret structural features and determine their geometries. The interpreted VSP reflectors were validated and integrated with the analyses of image logs and the interpretation of surface seismic data providing a constrained structural model that allowed the interpretation of seismic data away from well control and provided a starting point for seismic interpretation in areas where structural geometries were poorly imaged on surface seismic. It is shown that VSP, including vertical incidence, data can contribute to the understanding of reservoirs and enables well-derived information to be extrapolated away from the wells.

The detection of subsurface cavities, including mine workings, mine shafts and solution features, is an essential component of any site investigation for major civil engineering works and often relies on drilling investigations to identify the presence of any cavities. However, there is no standard, cost-effective site investigation technique which can be readily used for the physical investigation of such features.Whilst a desk study may yield documentary information on the presence of recorded mine workings and shafts, the location of solution features is generally even more problematical. Two complementary approaches have been developed for the location of subsurface cavities. Firstly, closely spaced boreholes are drilled in a specific pattern to locate cavities. This method can prove prohibitively expensive, with no guarantee of intersecting all voids or cavities. Secondly, remote sensing geophysical techniques have been used. Such techniques rely on the existence of contrasts in physical properties between the rock mass and the cavities, which can be detected using suitable geophysical methods.This paper describes the application of the microgravity technique to the detection of solution cavities and mine workings with reference to three case histories. In the first and second examples the microgravity technique was used as a reconnaissance method for defining targets for subsequent physical investigation; in the third, the technique was used to define the extent of solution features, having been initially and unexpectedly encountered by a drilling programme. These examples demonstrate the applicability of the microgravity method in detecting and delineating both solution cavities and mine workings within differing geological settings.

No abstract