Intro to Depth Registration

🤔 What is Depth Registration?

Depth Registration is the process Datarock Core uses to assign a downhole depth to every piece of rock identified in your core trays. This step is essential because all downstream products and outputs will use these registered depths.

Think of Depth Registration as the system that brings together all available depth information, evaluates the visible core, and produces the most accurate depth for each pixel of rock.

💻 How does Depth Registration work?

To determine the correct depth position for each piece of core, Datarock Core combines several inputs:

1. Identifying Rock in the Image

The system first classifies each row in a core tray into:

• Coherent rock – cylindrical, intact pieces of core

• Incoherent rock – broken or displaced fragments of core

• Core blocks – non-rock labeled markers from the drillers

• Empty tray – visible tray floor or small gaps between core

This creates a full map of “what is rock” and “what is not rock” along a sampling line through the middle of each row.

2. Collecting Depth Information

Datarock Core uses all available depth references (“tie points”), including:

• Depths provided by customers in filenames or CSVs, and

• Depths automatically detected using OCR (Optical Character Recognition)

OCR reads markings on core and core blocks, and keeps only the values that make sense relative to known depths.

3. Accounting for Core Loss

When intervals of core were not recovered during drilling, or there's a NAVI, depths can not be accuratly calculated. By providing core loss data this helps ensure accurate interpolation.

If core loss falls within a piece of rock or across two images, Datarock may split or shift that segment to correctly incorporate the loss.

4. Handling Broken Material Through Compaction

Smaller fragments or rubble may spread out in the tray and appear longer than they were when in-situ. A compaction factor is applied by Datarock Core to compress these broken intervals to better represent their true downhole length.

5. Combining Everything into Final Depths

All depth tie points and rock measurements are assembled, and the system interpolates the correct depths between fixed markers. Compaction and core loss are applied where needed, producing a final depth for every pixel of the core.

✅ Depth Health Checks

The goal of Depth Health Checks (DHC) is to avoid the need to review everything that has been uploaded. Instead, DHC will flag problematic areas where core length doesn't match particular depth intervals or here might be missing depth information.

▪️Interval length check

Intervals that are too long can lead to uncertainty, such as mask missing core loss, NAVI or other factors leading to inaccuracies.

▪️Rock quantity check

This check compares the median of the metres for each interval, pointing out that there might not be enough rock, or there might be too much rock. 

▪️Dipole checks

When there's a mistake somewhere, this might impact other intervals, where there's a whole sequence pushing errors from one interval onto the next.

⚙️ Depth Settings

▪️Dynamic row sampling

Moves the sampling line if the core is not perfectly centred in the tray.

▪️Minimum width of coherent rock

Reclassifies coherent rock as incoherent when the visible width is too narrow.

▪️Optical Character Recognition (OCR) Behaviour

Controls whether OCR should consider depth markings on core and/or core blocks.

To request these settings to be modified, please lodge a support ticket.

▪️Minimum prediction confidence

Defines the confidence threshold for model predictions.