TECHNOLOGIES AND SERVICES

OUR ANALYSIS METHODS:

Random Forest — an algorithm for forecasting mineralization zones based on geophysical, spectral, and geochemical data.
Bayesian inversion — creation of models taking into account uncertainty and probabilistic scenarios.
Deep learning (neural networks) — for recognition of geophysical domains and structural boundaries.
GPT modules — automatic interpretation of drilling logs, conversion of text into classifications.
Drillhole Re-Logger — automatic re-logging of drill core based on comments and data.

WHAT THE CLIENT RECEIVES:

• Mineralization forecast prior to drilling
• Reduction of risks of “empty” drill holes
• Analysis of large datasets in hours, not weeks

KUMBA INNOVATION:

• Integration of AI analysis directly into the geologist’s workflow
• AI operates in real time during drilling and interpretation

WHAT WE DO:

• Processing of Sentinel-2 images (VNIR, SWIR)
• 16 mineral spectral endmembers
• Comparison with the USGS library
• Identification of alteration zones: clay minerals, carbonates, oxides, sulfides
• Construction of 2D and 3D anomaly maps

WHAT THE CLIENT RECEIVES:

• A clear map of possible mineralized zones
• The possibility of planning exploration without preliminary drilling
• A digital atlas of the area with priority anomalies

KUMBA INNOVATION:

• Correlation of satellite data with geochemistry and ground-penetrating radar
• A unique automated methodology using AI for noise filtering

The Exploration AI Platform represents an integrated geospatial system based on artificial intelligence, intended for the identification and forecasting of zones of deep mineralization (up to 400 meters).
The operation of the platform is based on a combination of various data sources and advanced modeling methods, including:
• Sentinel-2 satellite imagery in the VNIR and SWIR ranges (multispectral and hyperspectral data)
• Geochemical data — survey results and historical exploration results
• Machine-learning methods and neural networks
(Random Forest, XGBoost, Stacking Classifiers, Bayesian Inversion, Convolutional Neural Networks, DBSCAN/HDBSCAN, K-Means, Gradient Boosting)
• Verified spectral database of mineral and gas “fingerprints”

CORE FUNCTIONS:

The system performs automatic pattern recognition, depth data extrapolation, and anomaly clustering, taking into account the geological context — stratigraphy, lithology, tectonics, and pathways of hydrothermal fluid movement.

WHAT THE CLIENT RECEIVES:

• Identification of hidden disseminated sulfide mineralization
• Structural map of the deposit
• Confirmation of spectral and GPR data

WHAT IS INCLUDED:

• IP survey: 2D/3D inversion of chargeability and resistivity
• Drone-based magnetic survey: aeromagnetic map, anomalies, and gradients
• Geophysical interpretation with AI support
• Construction of sections and 3D models of magnetic and polarization anomalies

KUMBA INNOVATION:

• Combination of GPR
• + IP + magnetics
• + spectral models
• Rapid data acquisition in hard-to-access areas thanks to unmanned (UAV) technologies

WHAT THE CLIENT RECEIVES:

• Volumetric model of the exploration area and structures
• Clear representation of ore bodies
• Tool for decision-making on drilling and resource evaluation
• Construction of block models, isosurfaces, mineralization domains

GEOSTATISTICAL AND MACHINE METHODS:

• Kriging, Inverse Distance Weighting (IDW), Principal Component Analysis (PCA), Multivariate Regression
• Clustering (K-means, DBSCAN), Support Vector Machines (SVM)
• Integration into a single model of data from GPR, IP, magnetic survey, spectral analysis, and drilling
• Visualization, animation, and analysis of models in Micromine, Leapfrog, Geoscience Analyst

REPORT PREPARATION

• Preparation of digital models compliant with the methodological requirements of the JORC and NI 43-101 standards

WHAT THE CLIENT RECEIVES:

• Conducting a preliminary justification of resource potential
• Preparation of projects for external evaluation (audit-ready) and technical / investment analysis

KUMBA INNOVATION:

• Work within the logic of international approaches: classification, reliability, transparency