Most nickel laterite resource statements in Indonesia still rest on a single method: Inverse Distance Weighting (IDW). The reasons are sound. It is fast, available in every mining package, and easy to explain to a reviewer. The problem is not the accuracy of IDW. The problem is what it cannot give you: a measure of how confident you should be in the tonnes and grade it reports.
This article looks at why laterite is the deposit type that needs that measure most, and what recent research offers to fill the gap.
Why laterite is hard to estimate
Nickel laterite forms by weathering of ultramafic rock. The profile is layered: limonite at the top (iron-rich, low to moderate Ni), saprolite below it (higher Ni, high MgO and SiO2), then bedrock. Three things make it difficult.

- The contacts undulate. The limonite-saprolite and saprolite-bedrock contacts follow old topography and fracturing, not flat planes. Between two holes 25 to 50 m apart, a contact can rise or fall by several metres, and layer thickness itself changes with slope (Swara et al., 2023).
- Grade changes quickly. Ni and Fe vary sharply between holes, even within one block. A Central Sulawesi study on 25 m spacing still found heterogeneity that had to be managed through grade control and more selective mining boundaries (Stephen and Haifan, 2026).
- Three layers are often too simple. Domaining from visual core description alone can misclassify material. Ni, Fe, MgO, SiO2 and Al2O3 all help define domains, and data-driven clustering has been shown to lower the variance within each domain (Sherafim et al., 2026).

What IDW gives you, and what it does not
On tonnes alone, IDW is not far off. In the Kolaka Block, North Konawe, IDW returned 4.61 Mt of laterite resource against 4.66 Mt from ordinary kriging, a difference of about 1%. The gap only opened at reserve level: 1.75 Mt (IDW) versus 1.66 Mt (kriging), roughly 5% (Putra et al., 2024). Another study concluded that ordinary kriging was not superior for laterite grade estimation; IDW suited the limonite zone and nearest neighbour the saprolite (Bargawa, 2022). None of this contradicts the case for kriging in formal reporting: on laterite, kriging earns its place through declustering, anisotropy and a kriging variance that can support classification, not through a different tonnage.
So if the tonnes are similar, what is the problem? IDW cannot answer three questions:
- What is a reasonable range for the tonnes? IDW gives one number. With no measure of uncertainty, Measured, Indicated and Inferred end up resting on drill spacing alone.
- Where are the contacts least certain? A block far from any hole and a block next to one are treated the same, yet their dilution risk is very different.
- Where should the infill holes go? Without an uncertainty map, infill planning falls back to judgement.
Kriging helps in part through the kriging variance, and several Indonesian studies use it for classification via RKSD (Nawir et al., 2023). But the kriging variance depends only on the sample configuration, not on the grades, so it still does not answer the question about the range of tonnes. The basic differences between kriging, IDW and machine learning are covered in our article Kriging vs IDW vs Machine Learning: Which Grade Estimation Method Is More Accurate?
What conditional simulation adds
Recent research moves from one deterministic model to many equally probable realisations. For an Indonesian laterite there is now a complete workflow: unfolding to flatten the undulating profile, plurigaussian simulation for lithology, imputation of incomplete assays, a multivariate transform (PPMT) so that Ni, Fe, MgO and SiO2 stay consistent with one another, and grade simulation by turning bands. The result reproduces lithology proportions, spatial continuity and the multivariate relationships, and gives a basis for resource classification and drillhole spacing analysis (Rolo et al., 2026).

The benefit shows up directly as operational risk. At a laterite deposit in Brazil, ten plurigaussian realisations were used to map the probability of each layer. The maps separated areas where the deterministic model is well supported from sectors with uncertain contacts, and local dilution risk in the limonite horizon reached 90,000 t. The same maps set priorities for infill drilling and geological control (Sepulveda and Marques, 2026). In Central Halmahera, conditional simulation has been compared directly with the specific-area method for classifying laterite resources (Renaldy et al., 2025).
What we recommend
- Domain first, estimate second. Separate limonite and saprolite, and test the domains statistically (histograms, CV, Ni-Fe-MgO relationships), not from core description alone.
- Keep IDW as a check estimate, but not as the sole basis for classification. Add ordinary kriging with kriging variance or RKSD as the first step; the whole workflow runs natively in GEOVIA Surpac.
- For deposits that will be reported or transacted, run conditional simulation on the main domains and report tonnes and grade as P10/P50/P90. That range is what financiers and due-diligence reviewers now ask for.
- Use the uncertainty map to plan infill drilling. Drill where the model is least certain, not simply on a tighter grid.
IDW is not wrong. It answers one question: how much. On laterite, the next question matters just as much: how sure. If you would like a second opinion on an existing laterite block model, or a classification and infill plan built on simulation, our resource estimation team works on exactly this.
Ranges like these are now expected at every study stage, from scoping to feasibility. How the three stages differ, and how often their numbers miss, is the subject of Scoping, PFS and FS: What Each Mining Study Answers.
References
Bargawa, W. S., 2022. The performance of estimation techniques for nickel laterite resource modeling. Jurnal Teknologi. https://doi.org/10.11113/jurnalteknologi.v84.17560
Nawir, A. et al., 2023. Resources estimation of laterite nickel using ordinary kriging method at PT Mahkota Semesta Nikelindo, Morowali. International Journal of Applied Sciences and Smart Technologies.
Putra, I. K. et al., 2024. Estimation of nickel laterite resources and reserves using ordinary kriging and inverse distance weighting (IDW) methods: A case study from the Kolaka Block, North Konawe. Journal of Earth and Marine Technology. https://doi.org/10.31284/j.jemt.2024.v5i1.6190
Renaldy, R. et al., 2025. A comparative study of conditional simulation and specific area methods for nickel laterite mineral resource classification: Insights from Central Halmahera, North Maluku. Mining, Metallurgy & Exploration. https://doi.org/10.1007/s42461-025-01407-w
Rolo, R. et al., 2026. Probabilistic modeling of lateritic nickel mineral resources. Minerals. https://doi.org/10.3390/min16050551
Sepulveda, G. and Marques, D., 2026. Geological uncertainty and internal dilution risk in a nickel laterite deposit: A plurigaussian simulation approach. Mining Technology. https://doi.org/10.1177/25726668261459979
Sherafim, J. M. et al., 2026. Optimizing lateritic nickel resource estimation through K-means and implicit modelling. IOP Conference Series: Earth and Environmental Science.
Stephen, S. and Haifan, M., 2026. Estimation of lateritic nickel resources using the inverse distance weighting (IDW) method at PT Five Star Indonesia, Central Sulawesi. Priviet Social Sciences Journal. https://doi.org/10.55942/pssj.v6i2.1594
Swara, H. R. et al., 2023. Analysis of empirical Bayesian kriging methods for optimization of measured resources estimation of laterite nickel. IOP Conference Series: Earth and Environmental Science. https://doi.org/10.1088/1755-1315/1134/1/012037


