Geological conditioning before covariance: Support-aware grade prediction in a Tanzanian stratiform graphite deposit

Grade models for layered graphite deposits must preserve geological contrasts while testing whether spatial interpolation improves prediction. This distinction matters when choosing the detail justified by exploration drilling. A support-aware evaluation used 3542 nominal two-metre total graphitic carbon (TGC) composites from 100 Tanzanian drillholes. Known lithology, parent-assay grouping, whole-hole and buffered validation, contact comparisons and matched kriging searches were combined with 170,000 synthetic evaluations. Lithology reduced mean absolute error by 5.2% and 4.8% relative to the global mean: paired differences were −0.098 (conditional 95% interval −0.151 to −0.049) and −0.093 (−0.143 to −0.045) TGC percentage points. Separate rock type means improved prediction at two-sided contacts. The empirical short-lag score was −0.012 against a threshold of 0.034, and fitted covariance parameters failed the chosen stability checks. The complete rule detected only 25.24–46.52% of strong twice-spacing synthetic fields and should not determine kriging use. For the tested implementation, wider searches restored all predictions, but regression kriging showed no clear gain over its trend. The contribution is an evaluation sequence linking inherited assay support, geological contrasts and matched prediction comparisons. It supports retaining geological information while testing the added value of residual interpolation where target lithology is known.

Authors

Publication Details

Journal
Applied Earth Science Transactions of the Institutions of Mining and Metallurgy
Published
2026-11-19
DOI
https://doi.org/10.1177/25726838261508344
Primary Topic
Geological Modeling and Analysis
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Geological conditioning before covariance: Support-aware grade prediction in a Tanzanian stratiform graphite deposit

Sudipta Chanda
Applied Earth Science Transactions of the Institutions of Mining and Metallurgy
Geological Modeling and Analysis
article

Geological conditioning before covariance: Support-aware grade prediction in a Tanzanian stratiform graphite deposit

Sudipta Chanda
article en

Abstract

Grade models for layered graphite deposits must preserve geological contrasts while testing whether spatial interpolation improves prediction. This distinction matters when choosing the detail justified by exploration drilling. A support-aware evaluation used 3542 nominal two-metre total graphitic carbon (TGC) composites from 100 Tanzanian drillholes. Known lithology, parent-assay grouping, whole-hole and buffered validation, contact comparisons and matched kriging searches were combined with 170,000 synthetic evaluations. Lithology reduced mean absolute error by 5.2% and 4.8% relative to the global mean: paired differences were −0.098 (conditional 95% interval −0.151 to −0.049) and −0.093 (−0.143 to −0.045) TGC percentage points. Separate rock type means improved prediction at two-sided contacts. The empirical short-lag score was −0.012 against a threshold of 0.034, and fitted covariance parameters failed the chosen stability checks. The complete rule detected only 25.24–46.52% of strong twice-spacing synthetic fields and should not determine kriging use. For the tested implementation, wider searches restored all predictions, but regression kriging showed no clear gain over its trend. The contribution is an evaluation sequence linking inherited assay support, geological contrasts and matched prediction comparisons. It supports retaining geological information while testing the added value of residual interpolation where target lithology is known.

Applied Earth Science Transactions of the Institutions of Mining and Metallurgy
Geological Modeling and Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon