Lithology: rock classification and logs

The first interpretation step is turning resistivity numbers into named rock units. pycsamt.interp.lithology provides a RockDatabase that classifies a resistivity into a lithology, and a StratigraphicLog that stacks those classifications into a borehole-style column. This example classifies the synthetic section and draws single-station and multi-station (fence) logs.

The rock database

RockDatabase.default ships a ready table of rock types with resistivity ranges. classify maps a value to its most likely rock.

from pycsamt.interp import RockDatabase

# Use the fence diagram (2nd figure) as the card thumbnail.

db = RockDatabase.default()
for rho in (8.0, 40.0, 200.0, 1500.0):
    print(f"{rho:7.0f} ohm-m  ->  {db.classify(rho).name}")

   8 ohm-m  ->  Alluvium (wet)
  40 ohm-m  ->  Aquifer
 200 ohm-m  ->  Granite (weathered)
1500 ohm-m  ->  Limestone

From model to stratigraphic logs

The quickest way to get per-station logs is the hydro interpreter, which classifies every sounding and exposes the resulting StratigraphicLog objects on .logs. (The next example covers the hydrogeology it adds on top.)

from _interp_data import demo_model

from pycsamt.interp import HydroInterpreter

rm = demo_model()
hydro = HydroInterpreter(
    water_table_depth=20.0,
    aquifer_range=(30.0, 300.0),
    clay_max=20.0,
    min_zone_thickness=8.0,
).fit(rm)
logs = hydro.logs
_mid = logs[len(logs) // 2]
print(
    f"{len(logs)} logs, e.g. {_mid.station_name!r} with "
    f"{len(_mid.layers)} layers"
)

11 logs, e.g. 'S05' with 19 layers

A single stratigraphic log

PlotStratigraphicLog draws the classic two-track column: hatched lithology on the left, the resistivity curve on the right. This is the figure you would hand a hydrogeologist.

from pycsamt.interp.plot import (
    PlotFenceDiagram,
    PlotStratigraphicLog,
)

PlotStratigraphicLog(logs[len(logs) // 2]).plot()

<Figure size 800x1000 with 2 Axes>

A fence diagram across the line

PlotFenceDiagram places every station’s log side by side at its true distance, so laterally-continuous units (here the aquifer and clay) line up into correlatable layers — the standard cross-section deliverable.

PlotFenceDiagram(logs).plot()

/opt/build/repo/pycsamt/interp/plot.py:308: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.
  fig.tight_layout()

<Figure size 2200x1000 with 11 Axes>

Reading it. The fence shows the four-unit sequence holding laterally with a gently undulating aquifer base, exactly the structure built into the model. Where a log’s colours break the trend, that station’s sounding is worth revisiting before trusting the correlation.