OR/17/003 Discussion and recommendations

From MediaWiki
Revision as of 09:59, 4 May 2017 by Ajhil (talk | contribs) (→‎Uncertainty in geological models)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
Terrington, R, Thorpe, S and Jirner, E. 2017. Enköping Esker pilot study - workflow for data integration and publishing of 3D geological outputs. British Geological Survey Internal Report, OR/17/003.

Input data

DTM

The DTM that was provided was in Integer format, which meant that values were rounded up or down to the nearest metre. This in turn means that the ground level can have an accuracy discrepancy of up to 1 metre. This was rectified in the 2nd phase of the pilot study as a DTM with floating data values (decimal places) was made available to the BGS when checking the geological model.

Boreholes

The three borehole datasets that were provided by SGU had Easting and Northing information but lacked start height data (the Z elevation at which drilling commenced). This is a vital piece of data, as borehole start height elevations, if measured accurately, can be compared against modern day DTMs to ascertain whether the land level has changed in elevation from when the borehole was drilled. Boreholes may have been drilled prior to some kind of engineered construction such as a road or railway embankment, or for mineral assessment before extraction. Sometimes, there might be more than one type of Artificially Modified Ground (AMG) change. Therefore, start height data can be used map landscape evolution with regards to land-use and the type of potential AMG deposited or removed (Terrington et al, 2015[1]). For this study, all boreholes were given a start height elevation applied from the LiDAR DTM and then further changed once the floating point DTM was made available.

As stated in Boreholes datasets, the borehole datasets had the drillers log information written. For consistency it would be preferential to have a coding schema in place such as the unlithified coding scheme (Cooper et al, 2006[2]) used at the BGS. This is a computer-coding scheme for unlithified deposits, commonly also referred to as superficial deposits, unconsolidated deposits or engineering soils and is approved in the BS5930 documentation. These include clay, silt, sand, gravel, cobbles, boulders and peat plus all the combinations of these deposits (Figure 34). The report describes the former BGS system for coding such deposits and details a logical system for coding many hundreds of lithological mixtures by the simple use of up to seven letters in various combinations. This makes the process for assessing the dominant lithology easier and colouring up of the boreholes using established legends.

CLAY: C
SILT: Z
SAND: S
GRAVEL: V
COBBLES: L
BOULDERS: B
PEAT: P

Figure 34    Fishnet density plot so the number of boreholes per square kilometre that were used in the construction of the cross-sections.

The Association of Geotechnical and Geoenvironmental Specialists (https://ags.org.uk/) are in the process of adopting this coding schema in their latest guidance for coding and distributing boreholes data.

Bathymetry

No bathymetry data was provided, however a water body (vatten) was incorporated into the geological model using estimation. Bathymetry could be added to the terrain model to improve the geometry and thickness of the underlying superficial units in these areas.

Workflow recommemendations

In the future, it might be preferential to start all cross-section correlation in Groundhog Desktop to utilise the cross-section snapping capability and then go to the SubsurfaceViewer for calculation and subsequent checking of thicknesses in ArcGIS. This will speed the checking process up as mismatches between cross-sections will be reduced and the tolerance at which correlation lines are snapped to outcrop and subcrop will be increased.

Image sections are easier to locate and move in the backdrop of a cross-section in Groundhog Desktop compared to GSI3D and the SubsurfaceViewer, which will save time.

The base of the model needs to be at a consistent level for neatness of outputs. This can either be done at the correlation stage or by adjusting the values in the xml project file.

Section names can also be adjusted in this way, or renamed in Groundhog Desktop if necessary. The provided schema at BGS is to have the project name followed by section orientation and then number, and the ID of the geologist doing the correlation (e.g. Enköping_WE_1_RTE). These have been re-named by BGS in the Enköping GSIPR.

Stochastic modelling

In some areas where the data density and detail is sufficient, stochastic modelling might be preferable to the deterministic methodology described in this report. This methodology has been used at the BGS to produce lithofacies models using GOCAD-SKUA®. The reference below has further information and detail about this modelling methodology:

Kearsey, T, Williams, J, Finlayson, A, Williamson, P, Dobbs, M, Marchant, B, Kingdon, A, and Campbell, D. 2015. Testing the application and limitation of stochastic simulations to predict the lithology of glacial and fluvial deposits in Central Glasgow, UK. Engineering Geology, 187, 98–112.

https://nora.nerc.ac.uk/509487/1/Kearsey%20et%20al%202015.pdf.

DINOloket — Provides access to information and data from GeoTOP, a voxel model that (100 x 100 x 0.5 m) that goes to a depth of 50 m and has split the lithostratigraphy up in lithological classes based lithology and grain size. This methodology has been developed by TNO and has been widely documented (Stafleu et al 2012[3]): https://www.dinoloket.nl/en/want-know-more

Uncertainty in geological models

‘The standard uncertainty layer is the display of the location of all input data and section locations, such as in the form of borehole fishnet density plots. Other methods can be applied if agreed upon in discussion with the client and costed appropriately’ — this is what is set out in this report and is stated in the ‘Specification guidance for input and output data formats and deliverables for commissioned 3D geological models’ report (Kessler et al, 2016[4]).

A simple example of a fishnet density plot is in Figure 34, which shows the number of boreholes per square kilometre. This does not take in account the depth of the boreholes. Further fishnet density plots can be produced to see how many boreholes intersect major horizons on a square kilometre basis using the depth of the base horizons and the length of the borehole drilled. This will give an indication of how constrained each horizon by boreholes.

References

  1. Terrington, R L, Thorpe, S, Burke, H F, Smith, H, and Price, S J. 2015. Enhanced mapping of artificially modified ground in urban areas: using borehole, map and remotely sensed data. Nottingham, UK, British Geological Survey, 38pp. (OR/15/010).
  2. Cooper, A H, Kessler, H, and Ford, J. 2006. A revised scheme for coding unlithified deposits (also applicable to engineering soils). British Geological Survey, 45pp. (IR/05/123).
  3. Stafleu, J, Maljers, D, Busschers, F S, Gunnink, J L, Schokker, J, Dambrink, R M, Hummelman, H J, and Schijf, M L. 2012. GeoTOP modellering (in Dutch). TNO Report 2012 R10991, 216 p.
  4. Kessler, H, Terrington, R, Wood, B, Ford, J, Myers, T, Thorpe, S, and Gow, H. 2016. Specification guidance for input and output data formats and deliverables for commissioned 3D geological models. Nottingham, UK, British Geological Survey, 6pp. (OR/16/052).