C2 · Publication Volume 12

Mineral Exploration Lifecycle, Target Generation and Competing Hypotheses

Treats exploration as staged uncertainty reduction supported by explicit hypotheses and decision gates.

Purpose and boundary of this book

Mineral exploration is a sequence of decisions made with incomplete geological knowledge. The purpose is not to collect every possible dataset or to defend a favourite deposit model. It is to reduce decision-relevant uncertainty in stages, preserve viable alternatives, and choose the next observation that can most clearly distinguish among them. A target is therefore a testable proposition about a mineral system at a stated scale, not a coloured anomaly or a promise of discovery.

This book follows the current concept map from regional screening through project generation, target concepts, source auditing, competing hypotheses, survey and sampling design, target ranking, decision gates, bias control and whole-program review. Geological reasoning remains central, but every interpretation is connected to observation support, data provenance, uncertainty, practical constraints and a predeclared decision rule.

The tutorial does not make investment recommendations, property valuations, resource or reserve statements, permitting advice, safety certifications or predictions of discovery. A high-ranked synthetic target is only the best candidate under an explicitly limited teaching model. It may be downgraded when new evidence arrives, when a necessary process is absent, when data quality is insufficient or when safe and lawful work is not possible.

General and institution-neutral scope

This is a general, institution-neutral tutorial. It has no affiliation with, sponsorship by, endorsement from or curriculum relationship to any company or individual. It is not written for a named operator, investor, consultancy, university, government programme, software product, exploration licence, mine, property or private dataset. Every locality, target, score, cost unit and observation in an unnamed example is synthetic teaching material.

People, public agencies, standards organisations, journals and repositories named in source notes identify technical sources only. A citation does not make any person or organisation the author, publisher, sponsor, partner, endorser, scientific authority or curriculum subject of this tutorial. The website that hosts the material is only a delivery surface. It is not presented as the tutorial's author, publisher, sponsor, provider, scientific authority or curriculum subject.

The methods are deliberately transferable. Learners may substitute the applicable tenure system, environmental rules, reporting code, coordinate reference system and data licence for their jurisdiction, but must never infer those details from a generic example. Commercial product names are unnecessary: the record is defined by inputs, transformations, assumptions and outputs rather than by software branding.

Exploration as an evidence-and-decision lifecycle

An exploration lifecycle can be represented as six linked states rather than as a rigid corporate stage chart:

  1. Question and search space: define the commodity or material question, geological system, spatial scale, exclusions and decision owner.
  2. Regional screening: identify permissive geological environments, important data absences and access constraints without pretending that permissivity is prospectivity.
  3. Target concept: express source, driver, pathway, trap, preservation and observable footprints as necessary, supporting, ambiguous or refuting criteria.
  4. Target test: acquire observations at a support and resolution capable of changing the hypothesis comparison.
  5. Decision gate: continue, modify, pause, drop or advance to a more intrusive test under a rule written before results are interpreted.
  6. Learning and portfolio update: preserve negative results, revise probabilities and criteria, and transfer lessons without converting hindsight into prior knowledge.

The states may repeat. Regional understanding can change after a detailed survey; a target test can reveal a new regional control; a failed hole can discriminate between geometry and system absence. Progress means improved decisions, not simply more expenditure or more data. A dataset has value only if it is fit for the question and could alter an action.

Learning outcomes

After completing this book, you should be able to:

  • frame exploration as staged reduction of geological and decision uncertainty;
  • distinguish a search-space hypothesis, play, lead, target concept, target and tested target without treating the terms as universal legal classes;
  • build a regional screening record that separates geological permissivity, evidence maturity, access, tenure, infrastructure and data availability;
  • translate a mineral-system model into scale-aware necessary, supporting, ambiguous and refuting criteria;
  • compile an inventory with source, licence, coordinate system, date, resolution, lineage, limitations and known gaps;
  • state at least two viable competing hypotheses and derive observations that discriminate among them;
  • design orientation work, survey geometry and sampling support around purpose, heterogeneity and expected footprint scale;
  • rank targets transparently while accounting for evidence dependence, confidence, missingness and model sensitivity;
  • define decision gates using actions, minimum evidence, stopping conditions, information gain and non-technical controls;
  • detect confirmation bias, survivorship bias, target leakage, spatial leakage and multiple-testing risk; and
  • review an exploration program as a coupled technical, data, safety, environmental, permitting and communication system.

Prerequisites and notation

The book assumes the geological foundations developed in the preceding Earth-materials, mapping, sedimentology, petrology, structural geology, regolith, geochemistry and mineral-systems volumes. Learners should be comfortable with conditional probability, proportions, logarithms, weighted averages, coordinate systems, sampling support and the distinction between observation and inference.

For a hypothesis H and evidence E, the conditional-probability update is

$P(H\mid E)=\frac{P(E\mid H)P(H)}{P(E)}.$

Probabilities express a declared state of knowledge under a model; they are not frequencies of discovery promised by the tutorial. A likelihood ratio P(E\mid H_1)/P(E\mid H_2) describes how evidence discriminates between two hypotheses. A score is not automatically a probability. Utility may be expressed in synthetic decision units so that actions can be compared without implying money, investment merit or economic value.

Expected value of perfect information is the difference between the expected utility of acting after learning the true state and the best expected utility available now. Sample information is imperfect and has an acquisition burden, so its net value must account for accuracy, action change, cost, disturbance, safety and delay. The equations are decision aids, not substitutes for judgement or legal authority.

How to use the diagrams and synthetic cases

Every lesson contains one purpose-built schematic. The diagrams are conceptual and not maps of real districts. Shapes, colours, contour fields and target identifiers are instructional. An arrow means a declared relationship or workflow transition, not proof of causation. Relative distance and size carry only the meaning stated in the caption.

All unreferenced datasets are synthetic. They are intentionally small enough to audit by hand and are not benchmarks for expected grade, anomaly magnitude, survey spacing, drilling success or project value. The correct output from an exercise has four parts: calculation, geological interpretation, decision implication and unresolved uncertainty.

Work through examples with an evidence ledger. Keep measured observations separate from transformations and interpretations. Record which information was available when each decision was made. When reviewing a result, do not silently use later information to judge an earlier choice. That temporal separation is essential for detecting leakage and for learning honestly from negative outcomes.

Reproducible exploration records

A defensible record links each decision to the versioned evidence that supported it. At minimum, retain object identifiers, acquisition method, location and coordinate reference system, observation support, time, operator role rather than personal identity, instrument or laboratory method, qualifiers, quality-control state, licence, transformation lineage, uncertainty and supersession history.

The interpretation layer should contain a hypothesis register, prediction matrix, target criteria, ranking method, dependence notes, sensitivity tests and dissenting interpretations. A decision record states the available actions, chosen action, reason, gate criteria, residual risk, authorising role and review trigger. Names of individuals are unnecessary in the teaching model; a real governance system may record accountable roles as required by applicable rules.

Negative observations are first-class evidence only when the method had sufficient detection power. A blank map cell may mean not surveyed, surveyed below detection, inaccessible, excluded, lost, censored or genuinely absent. Those states must not share one null value. Superseded files remain traceable so that later reinterpretation can reconstruct what changed.

Assessment and completion standard

Completion requires a target-portfolio dossier for a fictional region. The dossier must include:

  • a bounded search question and scale hierarchy;
  • a source-audited data inventory and gap map;
  • at least two competing system or geometry hypotheses for every advanced target;
  • criteria classified as necessary, supporting, ambiguous or refuting;
  • an orientation and sampling plan with support, spacing, direction and quality controls;
  • a ranking model with dependence audit and sensitivity analysis;
  • gate rules for continue, modify, pause, drop and intrusive testing;
  • a bias and leakage review that preserves the chronology of decisions;
  • safety, environmental, permitting and communication constraints; and
  • a final recommendation whose confidence and unresolved tests are explicit.

A satisfactory submission lets an independent reader reproduce the ranking from the same inputs, alter a contested weight, and see whether the decision changes. It also explains what a negative result would mean under each hypothesis. It does not claim discovery, legal rights, economic value or regulatory compliance.

Core sources