E5 · Publication Volume 27

Scientific Visualisation and Geoscience Software Interaction Design

Treats scientific visualisation and interaction as analytical instruments for geological reasoning.

Purpose and institutional neutrality

A general, institution-neutral tutorial that treats scientific visualisation and interaction as analytical instruments for geological reasoning, comparison, evidence inspection and uncertainty communication.

It has no relationship to any company or individual. All worked examples use synthetic evidence, generic roles and fictitious identifiers. Names of external people or organisations appear only in source lists so readers can trace the public evidence used to prepare the teaching material.

The website that hosts this tutorial is only a delivery host. It is not presented as the publisher, scientific authority, owner, sponsor or curriculum subject. No interface pattern, data structure or assessment depends on a particular commercial product, employer, jurisdiction, asset or named place.

Visualisation is part of the evidence chain

A scientific interface does more than display completed analysis. It selects variables, aggregates records, projects geometry, assigns visual channels, orders attention and changes the inspected population through interaction. Those operations affect what can be inferred and must therefore be declared, tested and reviewed like any other analytical transformation.

The governing question is not whether a view looks convincing. It is whether a reader can identify the claim, inspect eligible evidence, compare alternatives, recognise uncertainty, reproduce the state and reach a conclusion that the sources support. A blocked or unresolved conclusion is valid when required evidence is missing.

Learning outcomes

  • Translate geological decisions into observable visual tasks and testable success conditions.
  • Design a coherent geological visual grammar that separates observation, interpretation, alternatives and unknowns.
  • Choose colour, classification, legends and visual channels according to quantity, support and reader task.
  • Construct plan, section, three-dimensional, drillhole, core and tabular views with explicit transforms.
  • Link views through stable identity, deterministic state, visible filters and reversible interaction.
  • Represent uncertainty sources and alternative models without inventing confidence.
  • Trace every visible claim to source evidence, quality, derivation and limitations.
  • Audit precision, scale, smoothing, colour and perspective for visual misrepresentation.
  • Deliver keyboard, touch, screen-reader, responsive and static paths to the same scientific reasoning.

Prerequisites and conventions

The tutorial assumes the map, section and three-dimensional reasoning of A4 plus representative data workflows from Series B to E. Coordinate, semantic and spatial-computing principles from E1 to E4 are useful. It requires no particular operating system, programming language, graphics engine, database or software package.

Quantities retain units and support. Interval notation states closure. Selection, filtering, focus, visibility, camera, scenario and edit state are different. “Unknown” means evidence is insufficient or unavailable; it never means zero, absent or low confidence. Display precision never upgrades measurement precision.

Synthetic learning package

All exercises use SYN-VIS, a fictitious and location-free evidence package containing observations, interpreted boundaries, three alternative models, a small raster, drill traces, intervals, core strips, source records and quality findings. It intentionally includes occlusion, overlapping intervals, missing values, projection offsets, recovery gaps, ambiguous classes and contradictory evidence.

The package describes no real person, organisation, project, property or place. Received evidence remains immutable. Learners create versioned derivatives and can therefore compare a misleading portrayal with its corrected successor without rewriting history.

Design and evaluation method

  1. State the role, decision, task, eligible evidence, alternatives and consequence of error.
  2. Define entities, quantities, support, status and relationships before choosing marks.
  3. Specify visual encoding, interaction state, transformations and static fallback.
  4. Create ordinary, boundary, invalid and unresolved fixtures for the same task.
  5. Measure correctness, evidence citation, completion, reading error and accessibility by mode.
  6. Record findings, limitations, reviewer decisions and the exact state that was evaluated.

Evaluation follows the earliest failing design layer. A domain-task failure is not repaired with colour; a data-abstraction failure is not repaired with animation; a state failure is not repaired with explanatory prose. Corrections preserve the original finding and supersede the affected derivative.

Evidence and quality gates

  • Every visible mark maps to a stable object, value, support, status and version.
  • Every transform, aggregation, classification, projection and smoothing step is declared.
  • Legends expose units, classes, missingness, uncertainty cues, filters and applicable scale.
  • Selection, filtering, scenario, camera, clipping and hidden-object state remain inspectable.
  • Consequential information is not conveyed by colour, hover, gesture or perspective alone.
  • Static export preserves the analytical state, evidence references and limitations.
  • Unresolved hard failures block the affected conclusion without suppressing source evidence.

Capstone evidence interface

The capstone is a linked plan–section–3D–table evidence interface with a drillhole or core component, scenario comparison, uncertainty inspection, evidence cards and a static fallback. It uses only SYN-VIS evidence and generic roles. The learner submits a task model, visual grammar, state schema, portrayal manifest, accessibility matrix, adversarial fixtures and review report beside the interface.

Acceptance requires deterministic replay, correct shared identity, visible population counts, reproducible cameras and sections, source traversal, preserved alternatives, no silent repair, equivalent core tasks across required input modes and a static package from which another reviewer can reconstruct the conclusion.

Core sources