Well analysis

Goal. Read wells the way an interpreter does — a correlation view across bores, a net-cutoff (NetSettings) A/B sweep, and tidy zone tables in lithostratigraphic order.

This tutorial uses petekIO's well surface (loaded in the data-ingest tutorial) and the viewer's Wells tab.

Runnable notebook

Mirrored by the executed notebook petekIO — well analysis.

The correlation view

Attaching bores to a model (or opening a wells payload directly) lights up the Wells tab: N wells side-by-side on a shared inverted depth axis, each with its log tracks, tops as cross-track lines, and zones shaded in their identity colour.

# On the model path — attach bores when you build:
model = grid.model(props, con, wells=proj.wells)
model.wells_bundle()          # the wells_logs payload the Wells tab renders (or None)
model.view()                  # open it; the Wells tab is populated

Two hanging modes: TVD (absolute depth) and flatten-on-pick — choose a horizon and every well shifts so that pick aligns at Δ = 0 (the transform is viewer-side). A well with no top for the chosen pick is parked at absolute TVD with a dashed frame. Curve colour is identity by track, so PHIE reads one colour across every well.

Each bore carries the raw logs (PHIE with its net-cutoff line + fill, NTG, SW, a FACIES net strip) and the upscaled cube curves (PHIE_UP etc., the blocky cell value sampled back along the bore) as separate tracks — so you see exactly what the model kept.

NetSettings — a net-cutoff A/B sweep

Net pay hangs on a cutoff decision. NetSettings captures that decision as a value, so comparing two cutoffs is one loop over the per-zone net statistics:

import petekio

for cutoff in (0.08, 0.12):                      # A vs B
    net = petekio.NetSettings(porosity_cutoff=cutoff)   # discover exact params via help()
    table = w.zone_table("PHIE", net=net, stats=("mean", "p50", "net_fraction"))
    print(f"cutoff={cutoff}:")
    print(table)

The sweep shows how net fraction and net-conditioned averages move with the cutoff — the interpreter's sensitivity, made a value you can version and diff.

The net cutoff, end to end

The same net-cutoff decision flows into the model: ps.Prop("PORO", net_only=True, net_cutoff=...) conditions the porosity cube on net cells, and the viewer's Wells tab draws the PHIE net cutoff line at that value. One decision, one identity, all the way through.

Zone tables in stratigraphic order

Zone tables come back in true lithostratigraphic order — merged field-wide across every well in the tops file — not measured-depth order:

w.zone_table("PHIE", stats=("mean", "p50"))     # pandas DataFrame, zone in strat order
bore.zone_stats("PHIE")                          # [(zone, Stats), ...]
bore.zone_stats("PHIE", "Top A").mean            # one zone directly (None if absent)
geo.strat_order                                  # the field-wide column

A marker that pinches out (zero thickness) in one bore is still ordered correctly by a bore that develops it, because load_well_tops merges every well's relative ordering into one column.

Tie residuals

When a model ties surfaces to well picks, the per-horizon residuals are readable — the Map tab shows a tie-quality glyph per well, and the numbers are one call away:

model.well_tie_residuals()
# [{well, horizon, measured_depth_m, model_depth_m, residual_m}, ...]

Next: see the visualization tutorial for the full Wells tab, or synthetic data & testing for how these wells are generated.