Celtic Sea Floating Offshore Wind – Local Soil Reconstruction & Subsurface Intelligence

Executive Summary

This study establishes the soil reconstruction layer of Morie Analytics by transforming sparse geotechnical information into engineering-ready soil profiles at anchor locations.

Starting from a layout generated in morie_layout, the workflow constructs a localized soil domain, generates a synthetic ground truth model, emulates tomographic sampling, and reconstructs vertical soil profiles from sparse sectional data.

The process is performed in a controlled environment where the true subsurface is known, enabling quantitative validation of reconstruction accuracy before application to real offshore datasets.

Site intelligence → Layout generation → Soil reconstruction → Mooring physics → Anchor verification → Cable optimization

Project Scope

• Local domain extraction from selected floater cluster
• Cropping of bathymetry and soil datasets
• Synthetic layered soil model generation
• Tomographic sampling using sectional planes
• Soil profile reconstruction via interpolation
• Boundary detection and validation
• Multi-anchor soil characterization
• Quantitative validation of reconstructed layer boundaries
• Multi-anchor statistical evaluation of reconstruction performance

This study converts layout intelligence into anchor-ready subsurface inputs.

Engineering Context

Following layout definition, the next geotechnical question is:

What soil profile is acting at each anchor location?

Anchor design depends critically on:

• Soil stratigraphy
• Layer boundaries
• Spatial variability
• Localized geotechnical interpretation

However, early-stage offshore projects typically rely on:

• Sparse CPT data
• Limited geotechnical sections
• Incomplete subsurface coverage

This workflow introduces a localized reconstruction methodology, where sparse sectional information is converted into soil profiles suitable for downstream engineering analysis.

To address this, the workflow is first applied in a controlled synthetic setting, where the true subsurface is known. This allows direct validation of reconstruction accuracy and provides confidence in the methodology before application to real geotechnical datasets.

Inputs and Data Sources

This study builds directly on upstream Morie Analytics outputs:

From morie_site

• Bathymetry and soil classification grids
• Lease boundary definitions

From morie_layout

• Selected floater cluster
• FOWT coordinates
• Shared-anchor typologies and coordinates
• Local layout footprint

Additional Inputs

• Layered soil profile definitions
• Cropped local-domain datasets
• Synthetic soil-model parameters

All inputs are aligned in a common projected coordinate system.

This provides the subsurface data framework for anchor-level characterization.

System Flow

Layout Geometry → Domain Extraction → Tomographic Sampling → Soil Reconstruction → Anchor Profiles

The architecture ensures traceability from spatial inputs to geotechnical outputs.

Processing Workflow

1. Load selected layout and soil model
2. Extract FOWT and anchor coordinates
3. Define cropped local domain
4. Generate synthetic soil model
5. Build tomographic sampling grid
6. Sample soil properties along planes
7. Reconstruct profiles at anchor locations
8. Detect layer boundaries
9. Validate reconstruction against truth
10. Select representative anchor location

This converts layout geometry into anchor-level soil profiles.

Local Domain Definition

The workflow begins by defining a localized engineering domain around the selected floater cluster.

Figure 1 – Lease boundary and cropped local soil domain highlighting the selected floater cluster and associated anchor positions.

Engineering Significance

This figure establishes:

• The relationship between lease-scale site context and local soil study area
• The cropped analysis window used for detailed subsurface reconstruction
• Spatial continuity from morie_site and morie_layout into geotechnical analysis

Figure 2 – Cropped bathymetry defining the local engineering domain.

Figure 3 – Cropped soil classification aligned with the local layout domain.

Engineering Significance

These local-domain products define:

• A focused computational region around the layout
• Alignment between geometry and subsurface inputs
• The starting point for localized truth-model generation

Synthetic Soil Model (Ground Truth)

A synthetic truth model is generated over the cropped local domain to provide a controlled subsurface reference. The three layer model displays two boundary layers.

Figure 4 – Synthetic ground truth soil model with spatially varying layer interfaces.

Model Characteristics

• Three-layer sand system
• Interfaces: Z1 and Z2
• Spatially varying layer boundaries
• Linear variation of soil parameters within each layer:
  • Friction angle, φ
  • Relative density, Dr
  • Unit weight, γ

Engineering Significance

The truth model provides:

• A controlled reference model
• Known boundaries for validation
• A benchmark to quantify reconstruction performance

Tomographic Sampling Framework

To emulate realistic geotechnical data availability, the workflow defines a sparse tomographic sampling framework using 4 planes per main direction.

Figure 5 – Tomographic acquisition grid representing sparse geotechnical sampling across the local domain.

Configuration

• 4 XZ planes
• 4 YZ planes
• Single-point validation location inside the cluster
• Full anchor set preserved for later reconstruction

Engineering Significance

This step represents:

• CPT-style investigation lines
• Sectional surveys
• Limited geotechnical campaigns under realistic coverage constraints

Unlike isolated borehole or CPT-based approaches, this configuration captures spatial continuity through sectional sampling, enabling reconstruction between investigation lines.

Soil Section Sampling

The synthetic truth model is sampled along each tomographic plane to generate the available sectional dataset.

Figure 6 – Example XZ section showing vertical layering and lateral variability.

Figure 7 – Example YZ section showing vertical layering and lateral variability.

Engineering Significance

These sections form the available dataset, capturing:

• Vertical layer stacking
• Lateral variation across the domain
• Sparse but structured information for downstream interpolation

Profile Reconstruction

A single-point controlled validation case is first used to assess the reconstruction methodology within the selected layout.

Figure 8 – Reconstructed vs ground truth soil profile at the single-point validation location.

Method

• Inverse Distance Weighting (IDW)
• Multi-plane interpolation from XZ and YZ sections

Engineering Output

• Linear variability of soil properties with depth - φ(z), Dr(z), γ(z)
• Local profile reconstruction from sparse sections
• Validation of the proof-of-concept interpolation approach

Boundary Detection & Validation

The reconstructed profile is then converted into discrete engineering layers through boundary detection.

Figure 9 – Boundary detection accuracy at the single-point validation location.

Within the synthetic validation domain, boundary detection errors at the validation point are on the order of decimetres. Across all anchors, mean absolute errors are approximately 0.2 m for both Z1 and Z2, with maximum deviations remaining below ~1 m. These metrics quantify reconstruction performance against the controlled truth model, not uncertainty against real geological variability.

Engineering Significance

This step transforms reconstructed profiles into:

• Discrete engineering layers
• Inputs for capacity models
• Quantified reconstruction accuracy

Multi-Anchor Evaluation

Following proof-of-concept validation, the workflow is applied across all anchors in the selected layout.

• 24 anchors evaluated
• Independent reconstruction at each anchor location
• Boundary comparison metrics computed for all anchors
• Ranking performed based on total boundary error

Selected Anchor

• fowt1b → selected as downstream handoff anchor for morie_anchor, ensuring traceability from soil reconstruction to anchor design

This selection is now formalized in the workflow and exported for the downstream case study. The structure represents the final engineering output of the soil reconstruction workflow.

Engineering Significance

This step ensures that the soil-reconstruction module produces actionable anchor-specific outputs, not only validation figures.

Outputs Generated

The workflow produces:

• Lease-to-local crop context figure
• Cropped bathymetry and soil grids
• Synthetic truth soil model
• Tomographic section plots
• Single-point profile validation plots
• Boundary detection results
• Anchor-level soil datasets
• Selected downstream anchor definition
• Engineering-ready soil profiles at anchor locations

These outputs are directly usable in the next engineering modules.

Engineering Applications

The outputs support:

• Anchor design input generation
• Soil-structure interaction modeling
• Early-stage geotechnical interpretation
• Validation of sparse-data reconstruction
• Integration with anchor capacity workflows
• Validation of geotechnical assumptions under sparse data conditions

This transforms sparse subsurface information into structured engineering inputs for early-stage assessment.

Within the controlled synthetic domain used in this study, the results demonstrate that tomographic reconstruction can recover layer boundaries with sufficient accuracy to support proof-of-concept anchor-level screening. Application to real offshore datasets requires calibration against CPT, borehole or geophysical investigation data.

Relationship to Other Morie Study Cases

This study is the subsurface intelligence bridge of the Morie Analytics workflow.

Receives from:

• morie_site → lease boundary, bathymetry and soil grids
• morie_layout → floater cluster and anchor coordinates

Feeds into:

• morie_anchor → selected anchor profile and boundary definition
• morie_mooring → soil-informed assumptions for mooring line seabed contact

It provides the geotechnical transition from layout geometry to anchor design.

Why It Matters Commercially

• Reduces uncertainty in early-stage geotechnical interpretation
• Enables engineering decisions with limited site data
• Improves reliability of anchor sizing inputs
• Supports scalable offshore development workflows
• Bridges the gap between GIS-scale data and anchor-scale design
• Enables anchor design decisions before full site investigation campaigns are available

This is where subsurface uncertainty begins to be structured, tested and reduced.

Aspects to Improve

• Integration with real CPT datasets
• Multi-soil systems (clay, sand, rock)
• Uncertainty quantification
• Adaptive sampling strategies
• Machine learning-based reconstruction

These extensions would move the workflow closer to project-grade geotechnical intelligence.

Design Philosophy

This study reflects the Morie Analytics approach:

• Physics-informed
• Modular
• Traceable
• Engineering-focused
• Scalable