Seismic investigation of natural coal fires: A pre-fieldwork synthetic feasibility study.

Discussion

Our modeling suggests that, with a suitably high-frequency source and optimal field conditions, useful information can be gained from seismic surveys at the Durango coal fire site. We could realistically hope to discern changes across the burn front such as diffractions; these would be difficult to interpret because fissures would produce other, possibly stronger, diffractions. Separately imaging the top and bottom of the coal would be much more useful and would require frequencies at least as high as those used in the model. A lower source bandwidth would be inadequate.

Attaining frequencies even as high as those in the model (125 Hz) is difficult in practice. With sandstone at the surface, coupling of sources and geophones should be good and attenuation reasonably low. But in areas where dry rocky soil overlies the sandstone, coupling issues and attenuation would almost certainly filter the higher frequencies in addition to significantly impacting the propagating wavefront. A successful survey would require working directly on bare rock (likely gluing geophones into drilled holes) and quite possibly a high frequency source such as a small vibrator. The application of either of those, would significantly increase the field-work effort. Attaining sufficiently high-frequency energy with a hammer is not impossible at this site, but certainly challenging.

The model omits very important known heterogeneity at the site, a key problem that would surely impact the quality of any acquired data. The fissures extending from the burned zones to the surface would undoubtedly have a major impact on propagating wave fields, quite possibly masking the reflections from subsurface features in the burned areas. These fissures are less common (absent?) above the unburned coal, so we are more confident in the validity of our results in that part of the model. But shale layers exist throughout the sandstone in the field area, and could also cause reverberations and spurious reflections in the data. Thus even under ideal circumstances, where source and receiver coupling are optimal and source frequencies exceed 100 or even 200 Hz, the recorded waveforms might be very difficult to interpret.

In addition to the missing fissures, other aspects of the model may be wrong. For instance, we do not really know what parameters are reasonable for the burned coal; we simply divided the coal properties by four. But the ash might be better modeled as a void, which would make a major difference. We could expect stronger diffractions from the edge of the burned zone. But because the reflection from the coal is already quite strong, not much difference is expected between the reflections of the burned and not unburned parts of the coal seam. In any case, the known subsidence and rock changes that exist above the burned zone at this site and observed at other such sites (e.g., Wolf, 2006) would surely impact both wave propagation above, and within, the burned zone. The absence of this heterogeneity in the model is a major simplification of the real case.

We have not simulated an SH-wave survey at the site, an option that should be considered in any near-surface application. Being strongly sensitive to voids, shear waves might prove useful in this case. Their lower propagation speed would lead to shorter wavelength and better resolution for a given frequency, but generation of high frequency sources is even harder for SH-wave sources than it is for P-wave sources. In addition, the jumbled nature of the fissured sandstone would likely lead to many spurious S-P conversions that would complicate interpretation of the observed wave field. Accurately simulating the complexities of an SH-wave survey at the site would be difficult.

Seismic investigation of natural coal fires: A pre-fieldwork synthetic feasibility study.