For our 50th episode of GeOCHemISTea, we’re doing something a little different... a panel discussion that brings together two very different perspectives on one deceptively simple problem: levelling geochemical data.

Joined by a government survey geochemist, Pedro Acosta-Gongora, and a mathematical geologist, Tom Carmichael, this conversation digs into what levelling really means in practice, why it’s become standard in regional datasets, and where it can quietly distort the geological signal. From legacy data and batch effects to the tension between clean maps and real-world noise, this episode explores the intersection of applied geochemistry and data science.

If you’ve ever worked with multiple datasets and wondered “can I actually trust this?...” this panel is for you.

For this episode we read:

Exploratory data analysis of geochemical exploration data (Arne and Garrett, 2026)

Data Leveling of Multi-Map Geochemical Exploration Data Using Compositional Data Analysis: A Case Study from the Baiyinchagan-Maodeng Area, Inner Mongolia, China (Tang et al., 2025)

Levelling of multi-generational and spatialyy isolated geochemical surveys (Main and Champion, 2022)

G-BASE data conditioning procedures fro stream sediment and soil chemical analyses (Lister and Johnson, 2005)