From data to target: prioritizing areas with per-lithology statistical analysis

After processing magnetometry, gamma-ray spectrometry and derivatives, the interpreter faces a large volume of information. The objective, however, is always the same: to decide where to invest the next exploration step. This is where statistical analysis comes into play.

The problem of comparing "apples to oranges"

A uranium anomaly that is high for a granite may be perfectly normal for a phosphorite. If you compare absolute values without considering the host rock, you risk chasing false targets and ignoring genuinely significant concentrations.

The solution is to analyze each variable within its own lithological unit. That way, what stands out is the sample that departs from the pattern of that rock — not from the whole set.

Percentile thresholds

A robust way to flag what is anomalous is to use percentiles:

• p90 separates the top 10% of values;
• p95, the top 5%;
• p99, the most extreme 1%.

By applying these thresholds per lithology, the platform automatically highlights the concentrations that depart from the rock's typical behavior — the most promising target candidates.

Z-score: how unusual is this value?

The z-score measures how many standard deviations a sample is above (or below) the mean of its lithology. A high z-score indicates a statistically rare value. It is an intuitive metric for ranking anomalies: the higher it is, the more the point stands out from the local geological "noise".

Cross-referencing layers to gain confidence

No isolated anomaly closes a case. The strength comes from cross-referencing:

• A geochemical anomaly over a structural lineament;
• Potassic alteration coinciding with the edge of a magnetic body;
• Known mineral occurrences nearby.

When these independent lines of evidence overlap, the target stops being a curiosity and starts to deserve field investment.

Less noise, more decisions

Automation's role is not to replace the geologist, but to filter noise. By applying per-lithology statistical thresholds and organizing targets by relevance, the platform gives the professional an objective starting point — shortening the path between data and decision.

How Lithogeo helps

This is exactly the kind of work the Lithogeo platform automates. From an ANM process number or a polygon of your area, it processes and integrates the public data — magnetometry, gamma-ray spectrometry, geology, terrain and mineral occurrences — and delivers the products described here ready for analysis, in minutes and right in the browser.

Educational content. Statistical prioritization is a support tool; final validation must consider the geological model and fieldwork.