glacial-strat-patterns

Fill patterns and sedimentary-structure symbols for glacial / Quaternary stratigraphic columns — the black line-work you pour into a measured section, with a glacial vocabulary the USGS/FGDC lithologic set lacks: diamicton/till families, rhythmites, dropstones that deflect the laminae, cross-bedded outwash, loess, peat.

Two classes, following how measured-section keys are organised:

• Lithologies are tileable fills (till, gravel, sand, mud, loess, …) — what the unit is made of.
• Sedimentary structures (cross-bedding, ripples, folds, …) are overlays drawn over a lithology as bands or lenses that need not fill the interval — what happened to it.

The ornaments are drawn the way a geologist draws them by hand — to show process and form, not just texture. Cross-bed foresets sweep down-current and flatten tangentially into the bounding surface; gravel is imbricated; ripples climb; diamict is poorly sorted with no fabric; varves are wavy and uneven.

Lithologies are generated as parametric, seamless SVG tiles (the source of truth) usable via a matplotlib column helper, standalone SVG/PNG, or an importable Inkscape palette; structures are matplotlib overlays.

Why this exists

The FGDC patterns (Dave Quinn's geologic-patterns repackages them) are the de facto open set, but their glacial/Quaternary coverage is thin and generic. This library fills that gap: patterns designed for glacial sedimentology, keyed to Eyles/Miall-style lithofacies codes (Dmm, Dms, Sr, Fl, …) with plain-name aliases, and parametric so a facies is a family you tune (clast density, lamina spacing, stipple density) rather than one frozen tile.

Lithologies (13 fills)

Tileable fills, keyed to Eyles/Miall-style lithofacies codes with plain-name aliases (mpl.column_fill(ax, "Dmm", …), by code or alias):

Code	Alias	Group
Dmm	till (massive)	diamicton	matrix-supported, massive — lodgement/subglacial till
Dms	stratified diamicton	diamicton	matrix-supported, stratified — melt-out / waterlain
Dcm	clast-rich diamicton	diamicton	clast-supported, massive
Gh	gravel	glaciofluvial	clast-supported, imbricated
Gms	matrix-supported gravel	glaciofluvial	debris-flow / ice-marginal diamict
Sr	ripple-laminated sand	glaciofluvial	climbing-ripple cross-lamination
Sm	massive sand	glaciofluvial	massive / faintly graded
Fl	rhythmite / varve	glaciolacustrine	laminated silt & clay couplets
Fm	massive mud	glaciolacustrine	structureless silt & clay (mud dashes)
Cdm	colluvium (soliflucted)	colluvial	diamict; clasts aligned down-slope
Em	loess	eolian	massive eolian silt
P	peat / organic	organic	bog / fen
R	bedrock	bedrock	undifferentiated substrate

Sedimentary structures (overlays)

A different class from lithologies: structures draw over a lithology fill as a band (x0, x1, y0, y1) that need not fill the interval — the way measured-section keys separate "Lithology" from "Sedimentary Structures". In glacial_patterns.structures:

cross_bedding · trough_cross_bedding · cross_lamination · climbing_ripples (stacked chevrons) · wave_ripples · planar_lamination · folds (convolute / recumbent, a deformed band) · collapse (ice-contact faulting) · dispersed_clasts

Placed features

Point/line elements drawn over a fill:

• mpl.dropstone(ax, x, y, r) — ice-rafted clast deflecting the laminae (over Fl/Fm)
• mpl.boulder_pavement(ax, y, x0, x1) — a clast line marking a subglacial pavement / lag
• mpl.pebble_lag(ax, y, x0, x1, imbricate=…) — a pebble stringer, optionally imbricated
• mpl.erosion_contact(ax, y, x0, x1) — a scalloped erosional / scour surface
• mpl.flame(ax, x, y) — flame / water-escape (dewatering) structure
• mpl.mud_rip_up(ax, x, y) — a mud rip-up clast

Ornament forms are calibrated to published measured-section keys and the USGS FGDC lithologic-pattern standard (mud dashes, sand stipple, tangential cross-bed foresets, festoon troughs, chevron climbing ripples). Flow is left-to-right (one FLOW constant flips it).

Quickstart

pip install -e .        # numpy, matplotlib, pillow

import matplotlib.pyplot as plt
from glacial_patterns import mpl, structures as st

fig, ax = plt.subplots(figsize=(3, 6))
mpl.column_fill(ax, "Dmm", 0, 1, 8, 6)     # till lithology, 8–6 m
mpl.column_fill(ax, "Sm",  0, 1, 6, 4)     # sand lithology, 6–4 m
st.cross_bedding(ax, 0, 1, 6, 4)           # cross-bed structure over the sand
mpl.column_fill(ax, "Fl",  0, 1, 4, 1.5)   # varves, 4–1.5 m
mpl.dropstone(ax, 0.5, 2.4, 0.1)           # ice-rafted clast
ax.set_xlim(0, 1); ax.set_ylim(8, 0)       # depth downward
plt.show()

Address a lithology by code ("Dmm") or alias ("till (massive)"). Full worked sections: examples/example_column.py (lithology fills + structure overlays + dropstones) and examples/example_structures.py.

Raster tiles vs. vector hatches

Two fill routes, same facies keys:

• mpl.column_fill(...) tiles the raster PNG — the faithful ornament.
• hatch.column_fill_hatch(...) uses a matplotlib hatch — vector, resolution-independent, and tiny in PDF/SVG output, at the cost of fidelity (built-in hatches only approximate the tiles). Use it for publication figures you'll zoom or edit downstream:

from glacial_patterns.hatch import column_fill_hatch
column_fill_hatch(ax, "Gh", 0, 1, 4, 3)    # vector gravel fill

See examples/example_hatch.py (writes a true-vector PDF). Each lithology's hatch string is also in metadata/facies.csv.

Inkscape / Illustrator / QGIS

Not using Python? The svg/<code>.svg tiles drop straight in as pattern fills, and metadata/facies.csv is the index. For Inkscape there's a ready-made palette, inkscape/glacial-patterns.svg: open it and copy a swatch, or drop it into your Inkscape user patterns folder to get all 13 lithologies as stock patterns in Object → Fill and Stroke → Pattern. (Structure overlays are matplotlib-only for now.)

Regenerating the assets

python -m glacial_patterns.build   # writes svg/, png/, metadata/, contact_sheet, inkscape/

Regeneration needs the inkscape CLI (used to rasterise SVG → PNG). Editing a generator in glacial_patterns/patterns.py and rebuilding updates every asset.

Design notes

• Lithology vs. structure. Lithologies are fills; structures overlay them. Keeping them separate mirrors measured-section keys and lets any structure sit on any lithology.
• Ornament shows process. Foresets are tangential, gravel imbricated, ripples asymmetric and climbing — directional strokes encode flow and form, the way field-manual ornaments do. Palaeoflow is drawn left to right by convention (one FLOW constant flips it).
• Organic, not CAD-regular. Irregularity is a deterministic function of position, so clast size, lamina thickness, and waviness vary hand-drawn-style while the repeating tile stays seamless.
• SVG-native tiling. Each pattern is an SVG <pattern>; discrete ornaments are wrapped across tile edges (wrap9) so repeats are seamless.
• Deterministic. A seeded PRNG makes tiles reproducible.
• Codes + aliases. Lithofacies code is the primary key; a friendly alias resolves to the same facies.

Roadmap

Optional colour variants per facies, per-facies parameter presets, and further ornaments (striated/faceted-clast markers, till-fabric arrows, iceberg-turbate). Contributions of regional facies schemes welcome.

Licence

• Code (glacial_patterns/, examples/): MIT (LICENSE).
• Pattern assets (svg/, png/): CC-BY-4.0 (LICENSE-patterns.md) — original designs, informed by the public-domain FGDC standard but not copied from it.

See also the sibling repo gsc-of8572-swatches for GSC surficial map-face colours — the complement to these column fills.