Geochemical mapping is rapidly evolving from classical interpolation toward integrated workflows combining robust multivariate analysis, geostatistics, and artificial intelligence (AI). Modern environmental and exploration geochemistry relies on high-dimensional, multi-source datasets requiring advanced methods to unravel complex spatial patterns, quantify uncertainty, and support decision-making. Traditional approaches often struggle with non-linear relationships, compositional constraints, and heterogeneous data integration. This session highlights the shift toward next-generation analytical frameworks capable of handling contemporary geochemical challenges.

We welcome methodological and applied contributions, including:

- Advanced multivariate statistics: PCA, factor analysis, robust/fuzzy clustering, compositional data analysis (CoDA), ilr transformations, and handling censored data.
- Geostatistical modeling: Variography, kriging variants (ordinary, universal, co-kriging, indicator), sequential Gaussian simulation, multiple-point statistics, and uncertainty assessment.
- Machine & Deep Learning: Random forests, gradient boosting, SVMs, neural networks (CNNs, GNNs) for anomaly detection, source apportionment, and predictive mapping.
- Hybrid approaches: Integration of geological knowledge, process-based understanding, GIS, and AI-driven models to improve interpretability and physical consistency.
- Data infrastructures: Science- and AI-ready data standards, reproducible workflows, and open-source tools (R, Python, GIS).

We particularly encourage studies that:
a) Demonstrate novel combinations of multivariate statistics, geostatistics, and AI in environmental, hydrogeochemical, or mineral exploration contexts;
b) Address challenges such as spatial heterogeneity, imbalanced training data, high dimensionality, model interpretability, and uncertainty propagation;
c) Present validated case studies in contamination assessment, radiological risk, baseline definition, or mineral resource targeting;
d) Explore emerging topics, including deep learning for irregular spatial patterns, AI-assisted source identification, transfer learning, and domain-knowledge constraints;
e) Develop practical tools bridging innovation and end-user applications in academia, industry, and regulatory agencies.