Drawing the Line: Resource Classification Boundaries on Indonesian Projects

I’ve seen the same conversation play out in every resource estimate review I’ve sat through. The geologist points to a map and says “Measured here, Indicated here, Inferred out there.” The auditor asks why. The geologist says “drill spacing.” The auditor asks what else. Silence.

Classification is the single most consequential output of a resource estimate. It determines what can be converted to reserves, what can be financed, and what shows up in the press release. Get it wrong and you’re either leaving money on the table (under-classifying) or exposing yourself to liability (over-classifying). Both happen on Indonesian projects, and both are preventable.

The three categories — what they actually mean

Before drawing boundaries, let’s be precise about what each category represents:

Measured: Tonnage, density, shape, physical characteristics, grade, and mineral content are estimated with sufficient confidence to allow the application of modifying factors. The geological and grade continuity is established beyond reasonable doubt. You can make mine planning decisions at this level.

Indicated: The confidence level is sufficient to allow the application of modifying factors in a mine design, but the geological and grade continuity is not established beyond reasonable doubt. Mine planning can proceed but with more flexibility built in.

Inferred: Geological and grade continuity is assumed but not verified. The category is useful for exploration strategy and preliminary economic assessment, but not for mine design or reserves conversion.

Both JORC 2012 and KCMI 2017 use the same category names and the same underlying philosophy. The difference is in disclosure requirements and which Competent Person signs off.

What classification is NOT

This is where most projects go wrong. Classification is not:

1. A drill spacing template. “40m spacing = Measured, 80m = Indicated, 160m = Inferred” is a starting heuristic, not a rule. I’ve seen 40m spacing on a structurally complex epithermal vein that barely supported Indicated, and 80m spacing on a sedimentary-hosted coal that comfortably supported Measured.

2. A tonnage target. The Indicated boundary is not drawn to hit the number the deal team needs. This is the most common classification failure I encounter — and it’s one of the five resource estimation mistakes that cost companies real money.

3. A ring around the drill pattern. Drawing classification as concentric buffers around the drillhole collar map ignores geology, data quality, and estimation confidence entirely.

4. Irreversible. Classification can and should be updated as new data arrives. A Measured resource that fails reconciliation should be downgraded, not defended.

The four inputs to classification

Classification boundaries should be drawn using four inputs, weighted by deposit type:

1. Drillhole spacing (necessary, not sufficient)

These ranges shift with geological complexity. A simple, laterally continuous sedimentary bed can achieve Measured at wider spacing than the table suggests. A nuggetty epithermal vein with structural offsets might need tighter spacing than the minimum listed.

2. Geological continuity

This is the one that separates good classification from template-driven classification. Questions to ask:

• Can I trace the ore body between drillholes on cross-sections without ambiguity?
• Does the lithological model agree with the geophysical model?
• Are structural offsets mapped and accounted for?
• Is the weathering boundary consistent across the deposit?

If a 40m-spaced drill pattern crosses a fault that offsets the ore body by 15m, and that fault isn’t modeled, the spacing is irrelevant. The geological continuity is broken, and classification drops.

Indonesian context: Laterite nickel deposits in Halmahera and Sulawesi often have highly variable base-of-ore profiles. The ore body is continuous in area but thickness and grade vary over short distances. On these deposits, geological continuity in the horizontal plane is high, but vertical continuity (ore thickness) is lower. Classification should reflect this — Measured requires tighter spacing than the ore area continuity alone would suggest.

3. Data quality

Classification cannot exceed the quality of the data feeding the estimate. If your assay QA/QC shows 12% failure rate on standards, your classification ceiling is Indicated regardless of drill spacing. If your survey data has uncorrected deviation issues, the same applies.

Check:

• Assay QC pass rate (standards, blanks, duplicates) ≥ 95%
• Survey data validated (downhole deviation within tolerance)
• Density measurements: sufficient count per domain (minimum 30 per domain, ideally 50+)
• No material data gaps or exclusions that are undisclosed

4. Estimation confidence (kriging variance / kriging efficiency)

This is the quantitative input that most templates skip. Kriging variance and kriging efficiency provide a mathematical measure of how well the estimate is informed by data at any given block.

• Kriging variance ≤ 0.3 × sill: typically supports Measured (on deposits with good geological continuity)
• Kriging variance 0.3-0.6 × sill: typically supports Indicated
• Kriging variance > 0.6 × sill: Inferred only

Kriging efficiency (the ratio of the kriged estimate’s variance reduction relative to the naive variance) is another useful metric. A kriging efficiency above 0.7 generally supports higher classification; below 0.4 is Inferred territory.

Important: These thresholds are deposit-specific and should be calibrated against reconciliation data when available. Don’t apply them blindly.

Drawing the boundary — a worked example

Let’s walk through a real example. A Sumatran epithermal gold project, 60 holes on variable spacing (25m in the core, 50m in the halo, 100m on the periphery). Vein orientation N-S, dipping 70°E. One post-mineral fault offsets the vein by 8m at the southern end.

Step 1: Plot drillhole spacing on the plan. Identify the 25m, 50m, and 100m zones.

Step 2: Overlay the geological model. The vein is continuous in the 25m and 50m zones. The fault offset is modeled in the south. South of the fault, continuity is assumed but not verified between the 50m-spaced holes.

Step 3: Check data quality. QA/QC pass rate 97%. Survey validated. Density: 42 measurements in the vein domain, 18 in the halo (halo is under-measured but acceptable for Inferred).

Step 4: Review kriging variance map. In the 25m core, KV/sill = 0.22. In the 50m zone, KV/sill = 0.38-0.52. South of the fault, KV/sill = 0.65-0.80.

Classification decision:

The fault is the boundary between Indicated and Inferred, not the drill spacing. This is the kind of nuance that a spacing template misses.

Common mistakes on Indonesian projects

1. Using KCMI default spacing without adjustment. KCMI 2017 provides reference spacing tables, but these are starting points, not rules. A deposit with complex structural history needs tighter spacing than the KCMI default to achieve the same classification.

2. Ignoring weathering boundaries. On laterite nickel deposits, the transition from limonite to saprolite is a geological boundary that affects both grade and density. If the boundary is poorly constrained, classification should reflect that uncertainty.

3. Classification drawn by the deal team. I’ve seen a presentation where the Measured boundary was adjusted the night before a roadshow to include an area that was clearly Indicated. The CP signed it. When the project was later audited, the CP lost their registration. The classification is the CP’s call, full stop.

4. No classification map in the report. A table of tonnages by category without a map showing where each category is spatially is not a classification. The auditor needs to see the boundaries on a plan and on sections.

5. Upgrading classification after infill without re-estimation. Infill drilling improves data density, but if the estimate isn’t re-run with the new data, the classification hasn’t actually changed. “We drilled infill so now it’s Measured” is not valid without re-estimation.

How Orebit GeoSuite helps

The classification workflow in Orebit Resource (Phase 03) is built to prevent the common failures:

• Classification criteria are inputs, not afterthoughts. You type in the spacing thresholds, kriging variance limits, and minimum sample counts before classification runs. The tool won’t classify without rules on paper.
• Kriging variance map is generated alongside the block model. You see the spatial distribution of estimation confidence before drawing boundaries.
• Classification map export (PDF + PNG) is included by default. The plan and section views are ready for the report.
• Boundary editing is done on the kriging variance map, not on a blank canvas. You’re drawing boundaries with the data in front of you.
• JORC/KCMI Table 1 Section 3 is auto-populated with the classification criteria you used. No transcription errors.

Try Core free → · Get the bundle — Rp 99K → (single modules Rp 49K · one-time · yours forever)

Bottom line

Classification boundaries are the interface between your technical work and the financial world. They need to be defensible, documented, and drawn by someone who understands the geology, the data, and the estimation methodology — not by someone looking at a tonnage target.

The four-input approach (spacing + continuity + data quality + kriging variance) gives you a framework that holds up under audit. The spacing template gives you a starting point. The rest is professional judgment, and that’s what the CP signature means.

Reviewing your classification criteria before an audit? Email hello@orebit.id with the spacing table and kriging variance summary. I’ll tell you what an auditor would ask first.