OR/21/006 Introduction: Difference between revisions

From MediaWiki
Jump to navigation Jump to search
Line 24: Line 24:
   
   


[[Image:OR21006_fig1.pdf|thumb|center|500px|  '''Figure 1'''    Some issues in the derivation of discrete lithofacies from geophysical log data.  ]]
[[Image:OR21006_fig1.jpg|thumb|center|500px|  '''Figure 1'''    Some issues in the derivation of discrete lithofacies from geophysical log data.  ]]
      
      



Revision as of 09:01, 18 May 2021

Newell, A J, Woods, M A, Graham, R L, and Christodoulou, V. 2021. Derivation of lithofacies from geophysical logs: a review of methods from manual picking to machine learning. British Geological Survey Open Report, OR/21/006.

Contributor/editor: Kingdon, A

The importance of geophysical logs

Geophysical logs are continuous recordings of a physical parameter along a borehole. Many different types of parameter can be recorded such as natural radioactivity (the gamma-ray log), acoustic slowness, electrical resistivity and bulk density (see Rider and Kennedy (2011) for a complete discussion). They also include imaging tools which record images the of internal circumferal surface of the boreholes measured using light or false colour images derived from resistivity, acoustic pulses or measurement of active and passive radioactivity.

Geophysical logs are routinely used for the characterisation of rock formations in the subsurface where they may be important as aquifers, hydrocarbon reservoirs or potential sites for the storage of hydrogen or carbon dioxide. Geophysical logs are necessary because only a small fraction of boreholes are sampled by mechanical rock coring. For reasons of time and cost, coring may be limited to short intervals of a formation of interest, discrete side-wall cores or often omitted entirely. In many boreholes the only physical record of the subsurface geology comes from drill-bit cuttings which are retrieved from the circulating drilling fluid. Cuttings samples, however, leave a very imprecise record of the formations. Geophysical logs bridge the gap between cuttings and core by bringing additional ‘remotely-sensed’ data which can be used to reconstruct rock type and other rock properties such as fracturing at a high resolution. The relatively low cost of geophysical logging relative to continuous rock coring means that these data are routinely collected. Geophysical logs are probably the single largest source of continuous, quantitative stratigraphical data in the British Geological Survey archives and their potential as stratigraphic tools has long been known (Whittaker et al. 1985). They are, however, still relatively underexploited for this purpose particularly given the huge advances in log handling and analysis that has been made since these pre-digital studies. Geophysical logs offer the potential to not only pick gross lithostratigraphic contacts but to generate refined information on the internal lithological (or lithofacies) composition of a formation. The latter is the primary focus of this report.

Aim of this report

The aims of this report are to document:

  1. A range of methods that are currently used by the BGS stratigraphers to extract lithological information from geophysical logs (includes manual classification, cut-off analysis, mineral composition by linear inversion).
  2. Alternative methods which, at present, are not routinely applied but are sufficiently practical and accessible that they could become important, including unsupervised (k- mean clustering) and supervised machine learning approaches.

The report does not aim or claim to be a complete inventory of all possible methods to derive lithological information from geophysical logs. The authors welcome correspondence and information on any additional methods that are available or emerging.

The problem of deriving lithofacies from geophysical log data

The derivation of rock lithology or lithofacies using geophysical log data is complicated by a number of factors including (Figure 1) –

  1. Lithological (or lithofacies) classes are assumed to be discrete categories but in reality often represent a continuum e.g. sandstone merging to a mudstone by a progressive decrease in sand content/increase in mud content.
  2. Downhole sampling of lithologies by geophysical logging tools is imperfect and signals are often mixed (convolved) from multiple rock types where boundaries are crossed or the bed thickness is less than the relatively coarse resolution of the logging tools.
  3. Logging tools operate in difficult environments where variations in borehole diameter, drilling muds, pore-filling fluids, logging speed, tool calibration, high temperatures and pressures, borehole wall caving which affect contact between the lithology and some logging tools and other factors introduce many uncertainties in the process of determining lithology form logs signals.
  4. Geophysical log archives assembled over many decades are extremely heterogeneous in the quantity and quality of information they contain.
  5. Like all analytical equipment geophysical logs have ranges of optimum measurement resolution. Lithologies that record outside that range (such as very low gamma–ray values in very pure limestones, like the White Chalk Subgroup or high gamma–ray values in highly radioactive mudstone like the Kimmeridge Clay Formation) become much less reliable and measurements suitable for only qualitative rather than quantitative techniques.


Figure 1    Some issues in the derivation of discrete lithofacies from geophysical log data.