OR/16/004 Model development log

From Earthwise
Jump to navigation Jump to search
Barron, A J M. 2016. GSI3D model metadata report for HS2 Area 3 (Newton Purcell to Thorpe Mandeville). British Geological Survey Internal Report, OR/16/004.

All modelling done by A J Mark Barron.

14/3/13: Drew proposed lines of Area 3 cross sections in ArcGIS through as many as possible of the bores preselected during the borehole coding process, whilst retaining reasonably straight lines and a rectilinear grid (Figure 4). This excluded only five of the preselected boreholes.

Figure 4 Original cross section lines and selected boreholes
Centreline (proposed HS2 route) in red, dotted lines are Area 2 and Buckingham 1:50 000 cross sections. For cross-section labels and helper sections added later to aid calculation see Figure 5 and Figure 6.
Figure 5 Cross sections and helpers — Area 3 north-west part. Proposed centreline in blue, model boundary in green. Contains Ordnance Survey data © Crown Copyright and database rights 2014.
Figure 6 Cross sections and helpers — Area 3 south-east part.
Contains Ordnance Survey data © Crown Copyright and database rights 2014.

Appendix 1 gives a dated section-by-section, version-by-version account of the model build. NB section comments (‘bubbles’) were used sparingly (e.g. NWSE_2; Centreline). Additional comments as follows.

The model was correlated throughout down to 30 m below OD as per project instructions. This level is very widely very close to the base of the Charmouth Mudstone Formation (CHAM), which is usually below, but in places just above (EWrung_14; Centreline; EWrung_4; EWrung_5). Here the underlying Penarth Group and Mercia Mudstone Group strata (WLI, PNG, MMG) are locally correlated but were not calculated.

14/5/13: Lenses of GFDMP and TILMP1 were created in several cross-sections in the south from v73 onwards to v152. 4/6/13 S Thorpe advised not to use lenses if unit daylights at all. Need a basal GFDMP unit.

15/5/13: Faults were drawn as NONE lines in cross sections from v77. See 14/6/13 below.

16/5/13-6/6/13: Polygons (polys) from the DiGMapGB-50 layer were added to their respective geological units (GU), apart from very minor outliers that are not intersected, and envelopes built and modified as indicated in Appendix 2. Exceptions/rules as see Model assumptions, modifications in order of creation below:

  • FMB-LSMD and BWC: DiGMapGB-50 polys do not fit across 202/219 boundary. Redrew as per MG Sumbler 1:10 000 map SP63NW (Figure 7).
  • HEAD: Some modifications for sub-ALV coverages v141 (Figure 8).
  • MGR: 31/5/13, 3&4/6/13 new polys drawn for railway embankments).
  • WMGR: 31/5/13, 3/6/13 new polys for backfilled railway cuttings.
  • ALV: 3/6/13 joined polys and extended under LSGR at EWrung 6 W2900 and EWrung 7 W2550*.
  • LSGR: 3/6/13 extended at EWrung 6 W2900 v149*
  • WGR: 4/6/13 made a poly for railway cutting.
Figure 7 Modifications to BWC geological unit against DiGMapGB-50 at the Towcester-Buckingham sheet boundary.
Merged FMB geological unit highlighted. Proposed centreline in blue, model boundary in green. NB BW (yellow) against White Limestone Formation. Contains Ordnance Survey data © Crown Copyright and database rights 2014.
Figure 8 Examples of modifications to HEAD GU under ALV GU.
  • TILMP1: till unit created that lies above the most extensive GFDMP but not separated/ distinguished on survey maps from subjacent TILMP. 5-6/6/13 — coverages built above GFDMP and surface closures constructed where GFDMP is missing (Figure 9).
Figure 9 TILMP1 GU in bright blue superimposed on DiGMapGB-50 superficial layer.
DiGMapGB TILMP polygons in pale blue, proposed centreline in blue, model boundary in green. Constructed base of TILMP1 indicated by red rectangles. Contains Ordnance Survey data © Crown Copyright and database rights 2014.
  • WHM: GU modified vs DiGMapGB-50 in EWrung 14 W1100*.
  • MRB: GU modified vs DiGMapGB-50 in EWrung 14 W1300*.
  • GFDMP1: generally overlies TILMP1, but locally coverage surface closure was constructed cross-cutting very complex glacigenic stratigraphy (Helper_4 W7000–8000) in order to close the GU (Figure 10).

* requires modification to DiGMapGB, check 50 and 10.

Figure 10 TILMP1 (bright blue) and GFDMP1 (bright pink) GUs superimposed on DiGMapGB-50 superficial layer in SP 66 33 showing cross-cutting lines. DiGMapGB TILMP polygons in pale blue, GFDMP in pale pink, model boundary in green. Contains Ordnance Survey data © Crown Copyright and database rights 2014.

30/5/13: GFDMP lenses fixed and colour up v138.

10/6/13: by now all lenses translated to non-lens correlated lines v194. 11/6/13: first attempt at calculation of all failed.

14/6/13: excluded all CAIP cross sections and two Buckingham cross sections from calculation. Calc all — stalled at WHM, then at GOG. S Thorpe suggested the join between GOG and its component formations could be the problem — made all these into near vertical ‘L’ forms with GOG v208 (e.g. Figure 2). Also need to make ‘L’ forms at faults where coverage ends — did up to EWrung 4 v214.

14/6/13: H Kessler suggested make proper modelled faults which can be used in GoCAD. Reopened v208, created faults HS2_Area3_F1_AJMB to HS2_Area3_F31_AJMB (Figure 11) and fixed faulted line v224. 20/6/13: Calculate all, stalled at NS. Requested help from S Thorpe.

Figure 11 Faults drawn and modelled.
Total of 32 faults drawn. These were later removed as they hindered model calculation.

28/6/13: S Thorpe reviewed v224:

“I can see that it halts when trying to calculate the Whitby Mudstone. This is usually a result of one, or a combination of, the following reasons:

  • The Whitby Mudstone isn’t consistently snapped/tied into control points
  • The unit(s) above which create the top of the WHM aren’t snapped/tied into control points
  • There are units which don’t make geological/topological sense (you pointed one of these out to me when we last spoke) there are a number of these in your model

From looking through your sections and envelopes I’d say that most of the above reasons are applicable. However, I don’t want you to think I’m criticising your model as it looks like a huge amount of effort has gone into it, and this is clearly reflected!!

One issue I do have is the splitting of units along the sheet boundary. The Forest Marble and Blisworth Clay boundary is completely arbitrary for example, and I don’t think this should be replicated in the modelling especially as you have clearly shown that they share a common depth and shape in section. The other that I found was the White Limestone and Blisworth Limestone, in fact these four units share this common boundary. Is there a way to remove this manmade boundary, perhaps by subsuming one unit into the other for each pair? Or maybe creating a new unit to hold both descriptions? (WHL_BWL for example). This would remove a lot of ambiguities from the calculation, in addition to removing a lot of hard meticulous work on your part to get the pairs of units to sit neatly underneath each other where sections don’t constrain them precisely.”

3/3/14: opened v224. Merged all BWC into FMB, fixed in all cross sections v225. ST recommends not to merge WHL and BWL but TY still won’t calc — problem is WHL/BWL.

5/3/14: Merged all BWL into WHL, fixed in all cross sections v228.

5/3/14-7/3/14: Repeated Calculate all commands stalled at various bedrock units — HYSA, NS, WHM.

10/3/14: ST also says that modelled faults will not assist calculation — need to join same lines across (Ctrl+click) as steps and ensure these do not cross ones above (Figure 12). May also need to densify nodes here and elsewhere to assist calculation.

Figure 12 Problems at fault planes of correlation line crossovers.

17/3/14: v246 calc all, stalled at STAM, sent to 3D. Layers needing help: STAM SHHB TY RLD1 WHL FMB CB KLB TILMP GFDMP TILMP1 GFDMP1 HEAD ALV.

18/3/14: ST advises “a few more helpers are needed. You also need to check your snapping and/or envelopes in the areas ... where some units are faulted and GSI3D thinks they should be interconnected somehow. The bottom right corner showing blue triangles could be the edge of the DTM, in which case you may need to just enlarge your envelopes outside the area to ensure a tighter calculation before the cookie-cutting takes place.”

20/3/14-21/3/14: Added Helper sections 1 to 4 (0; Figure 5; Figure 6). Calc all excl CAIP sections, stalled at FMB, with CAIP sections stalled at WHL.

24/3/14: del CAIP old sections, added in new from her v1_93. Edited various bedrock coverages: STAM NS RLD1 WHL FMB CB KLB v254. Added PET GU. Mod GFDMP in NE and NS below STAM v256.

25/3/14: extended NS envelope south into Area 2, ref 10ks SP52NE (2m thick), SP52SE (2.5 m) and SP62NW (0–1 m) and Sumbler (2001, p.8)[1], and GOG envelope SE across CAIP Rung 16 v259.

28/3/14: added nodes to fault steps and to the joins between GOG and component formations: EWrungs 1–14, NWSE1–4, Centreline v263.

28/3/14: added in CAIP rung 16 from her v112, calc all, stalled WHM. Extended HYSA and WHM beyond DTM v265.

31/3/14: extended ALV, HEAD Gus. Added Helper_SD_12 and 13 v266. Deleted old CAIP sections, added new from her v113.

31/3/14 calculate all (excl all CAIP sections) completed (17 mins), sent to 3D, various bleed problems, extended envelopes of all bedrock units from KLB to DYS outside DTM, calc all stalled at WHM.

1/4/14: Added Helpers ALV_14 to WHM_19 v275.Calc stalled at WHM. Densified various bedrock lines v277.

2/4/14: calc all stalled NS v279.

3/4/14: calc all completed v282

4/4/14: calc all completed v284. Snap to Helpers and synthetic render check of all main sections — all ok except EWrung 8, no colour v287.

7/4/14: calc all completed v288.

8/4/14: calc all completed v290, v294.

9/4/14: ST looked at EWrung 8’s rendering problem and redrew an MGR line v295.

12/5/14: calc all complete, sent to 3D render probs in NE, added nodes to Lias fms here, del phantom polys in GFDMP that were TILMP. Added Helpers ALV_30, ALV_31 in NW.

13/5/14; mod TILMP under ALV N of EWrung 9 v297.

14/5/14: deleted small TILMP1 poly v298.

15/5/14: calc all stalled at TY. Various snaps improved v299.

21/5/14: requested Ben Wood look at v302 model to determine reason for part-calculation and stall. Is it connected with ‘Points not used’ e.g. for TY? Ben examined and advised that lines from DiGMap were too node-dense, and to apply the ‘Simplify all map units command’ (‘Points not used is not a problem — refers to coverage outside). We tried with a tolerance of 10 m — makes lines rather polygonal. Calculation successful. Recommends do not re-snap correlated lines to outcrop flags, but save simplified calculated model as a distinct version and reopen previous version for further edits. Line simplification used is based on the Douglas-Peucker algorithm http://en.wikipedia.org/wiki/Ramer%E2%80%93Douglas%E2%80%93Peucker_algorithm. The implementation used is from a code library called the Java Topology Suite (JTS) http://www.vividsolutions.com/jts/JTSHome.htm.

22/5/14: opened v305, applied Simplify lines with 5 m tolerance — very quick, looks ok, calculation successful, saved as v305_calc_complted_tol_5. Extended all GU coverages out beyond ends of sections (as appropriate within outcrops) to preclude edge effects v311. Added Helpers GFDMP_39, _40, NS_41 checking GU coverages bottom up: CHAM, DYS v312.

23/5/14: opened v312, applied Simplify lines with 5 m tolerance, saved as v312_calc_complted_tol_5. Various edits calc and save v313_calc_complted_tol_5. Added MRB lines to Helpers 5, 6 and 20 v314. Added Helpers NS_42, checking GU coverages MRB, WHM, NS, GOG, added Helpers HYSA_43 correlated d/t NS, Helper_TY_44 correlated d/t WHM v315, Helper_TY_45 correlated d/t GOG, Helper_TY_46 correlated d/t NS v316.

27/5/14: opened v316, auto-densified MRB base in Helpers 2, 3 v317, NWSE_2 and 3, EWrungs 1, 2, 7, 6, 8, 14, Helpers 18 and 28 v318, in EWrung 12, auto-densified WHM in EWrung 12 and WHM and MRB in Helper 15, and NS in NWSE_1. Checked GUs: GOG, HYSA, added Helper_HYSA_47 drew d/t MRB, checked GUs STAM, RLD, SHHB (simplified linework to 5 m), WBRO and TY, densified TY lines in H_NS_41, NWSE_1. Added Helper_TY_48, drew d/t WHM v319. Checked GUs: WHL, BLAD, FMB, auto-densified FMB in EWrung 6 and Helper_4, added Helper_FMB_50, drew d/t FMB. Checked GUs: CB, KLB, GFDMP0, TILMP, GFDMP, TILMP1, GFDMP1, HEAD, ALV, extended H_HYSA_34 NW and drew d/t WHM, densified ALV in Helpers ALV 30 and 31 v320.

30/5/14: cross-snapped Helpers 47 to 50 v321.

31/5/14: Modified DYS and MRB coverages at north end v322. Fixed problems with TILMP and WHL coverages in NE, calculation successful v323.

16/6/14: calculated GU grids sent to Arc. Many discrepancies seen compared with thicknesses cited by Horton et al. (1987)[2] and Sumbler (2002)[3], and likely variations inferred from modelling. Require interventions.

18/6/14: all croplines simplified to 5 m tolerance, detaching many from correlated lines. These were autosnapped section-by-section with some manual intervention and saved to v324-v334.

19-23 23/6/14: thicknesses of various GUs assessed and action to fix taken as follows:

  • GFDMP where >15 m in extreme SE — Area 2 Rung 16 included in calculation fixes this: cross section truncated at SW end to within DTM and resaved as Area_2_Rung_16_CAIP_trunc v341.
  • CB where >4 to 10 m in south — mods to Helper_SD_13 fix.
  • FMB where >7 to 17 m in SE — added Helper_FMB_51 drawn d/t FMB thro SP63SE7 and sl mod to TILMP coverage to fix v336
  • WHL where 15 to >25 m in SE — extndd Helper_FMB_51 to Centreline, drew d/t WHL in and densified WHL here, in Rung 1 and Helper_TY_45 v336.
  • RLD1 where <17 m — densified in Rung 4, Centreline; Fault F32 added and drew displacement in Centreline (3 m) and Rung 6 (3 m); extndd Helper_4 north to Rung_9 and drew d/t WHM v337.
  • TY where 7 to >18 m — added Helper_TY_53 drew d/t TY v338 and x-snapped; densified TY in Rung_11, NWSE_3. Disregard at W end Rung_5 as edge effect.
  • SHHB where 4 to 11 m — amended TY coverage at SE end Helper_41 and in Rung_2, extndd Helper_49 to SE, added MRB, all densified v339.
  • HYSA where 7 to 15 m — W end Rung_8 OK, HYSA densified in Rung_5 and Helper_3 v339.
  • NS where 6 to 25 m — disregard at W end Rung_5 as edge effect; 20/6/14 densify in NWSE_2 and Rung_3 v340.
  • MRB — up to 6 m in NW ok. Holes in south: densified in Rung_7, Rung_6, added and densified in Helper_NS_22, Helper_GOG_33, Helper_TY_45 and 48 v340.

23/6/14: calculation completed v341.

28/7/14: raster backdrop deleted from EWRung_14,

NB the model version v342 submitted for QA on 28/7/14 includes the two Buckingham map cross- sections and eight cross-sections form Cath Cripps’ Area 2 model (+1_HS2, +2_HS2, -1_HS2, — 2_HS2, 0_HS2, TILL_helper_1, TILL_helper_2 and TILL_helper_3. All were used to inform correlation and the Area 2 sections are fitted to, but all are excluded from the calculated model. It may be appropriate to remove these from the delivered model.



5/6/13: v5 GFDMP0 added below TILMP GLEG

2/6/14: changed GOG to more orange and SHHB to more bluish to distinguish. GLEG saved as W:\Teams\GLE\GLETeamLeader\Data\HS2\HS2\GSI3D_Modelling_files\HS2_GLEG_v5.GLE G.

GSIPR (post QA)

7/1/15: Technical check by Ian Cooke v343_tol_5 m_ilc.gsipr.

21-27/5/15: QA undertaken and QA form compiled by Helen Burke (HBU) v344. 8-12/6/15: AJMB undertook post-QA amendments as Appendix 2 v360.

12/6/15: Steve Thorpe tested Area 3 for calculation v360_post_HBU_QA_AJMB_ST_fortesting_ 1_3.gsipr.

23/6/15: AJMB added Helper_ALV_54 v362 calc stalled TY.

24-30/6/15: S Thorpe has got a version to calculate: HS2_Area3_3D_Model_v364_post_HBU_ QA_ST_TEST_AJMB.gsipr. Use this for packaging for LithoFrame viewer.


  1. SUMBLER, M G. 2001. Geology of the Bicester area (SP 52 SE). British Geological Survey Technical Report, WA/01/10.
  2. HORTON, A, POOLE, E G, WILLIAMS, B J, ILLING, V C, and HOBSON, G D. 1987. Geology of the country around Chipping Norton.Memoir of the British Geological Survey, Sheet 218 (England and Wales).
  3. SUMBLER, M G. 2002. Geology of the Buckingham district — a brief explanation of the geological map. British Geological Survey Sheet Explanation, Sheet 219 (England and Wales).