Kortagerð (Mapmaking) — Iceland Maps

Generate high-quality maps of Iceland using cached LMI geodata. Supports interactive HTML (Leaflet) and static PNG/SVG (geopandas + matplotlib).

Quick Start

# Ensure geodata is cached
uv run python scripts/lmi.py download

# Interactive map (full-screen, layer toggles, popups)
uv run python scripts/kortagerð.py html -o reports/map.html

# Static map (publication-quality PNG)
uv run python scripts/kortagerð.py static -o reports/map.png

# Zoomed to capital area with Reykjavík highlighted
uv run python scripts/kortagerð.py static --bounds capital --highlight "Reykjavíkurborg" -o reports/rvk.png

# Overlay custom points from CSV
uv run python scripts/kortagerð.py static --points data/my_locations.csv -o reports/custom.png

Bounding Box Presets

Name	Area	Bounds [W, S, E, N]
iceland	Full country (default)	-24.7, 63.2, -13.1, 66.6
capital / reykjavik	Greater Reykjavík	-22.1, 63.95, -21.3, 64.25
southwest	Reykjanes to Hekla	-22.5, 63.6, -19.5, 64.5
north	Skagafjörður to Húsavík	-20.0, 65.2, -15.0, 66.6
east	Eastfjords	-16.0, 64.2, -13.3, 65.8
westfjords	Westfjords	-24.7, 65.0, -21.0, 66.6
south	South coast	-21.0, 63.2, -17.5, 64.3
akureyri	Akureyri area	-18.4, 65.5, -17.7, 65.8

Cached Layers (in data/geodata/)

File	What	When to use
Landmask.geojson	Iceland land polygon	Always — base fill for any map
CoastalLine.geojson	Coastline	Border outline
AdministrativeUnit_level2.geojson	128 municipalities	Choropleth, boundaries, --highlight
RoadLines.geojson	Road network	Infrastructure, transport maps
WatercourseLine.geojson	Rivers (28k features)	Hydrology (skip for perf if not needed)
Lake_Reservoir.geojson	Lakes	Hydrology, geography
LandIceArea.geojson	Glaciers	Terrain, geography
BuiltupAreaPoints.geojson	98 settlements + pop	City labels, settlement maps
NatureParkArea.geojson	National parks	Environmental, tourism maps
IslandArea.geojson	Islands	Offshore geography
Airport_Airfield_points.geojson	Airports	Transport maps
Port.geojson	Harbors	Maritime, transport maps

Caching strategy (for raster-overlay maps)

Vector layers are pre-cached at Tier 1 (data/geodata/). For maps that overlay rasters (e.g. Copernicus HRL Grassland), three more tiers exist under data/cache/ to avoid re-downloading and re-projecting on every render.

Tier	Path	Built by	What	Speedup
1	data/geodata/*.geojson	scripts/lmi.py download	LMI WFS vectors (~50 MB)	—
2	data/raw/lmi_hrl/*.tif	scripts/lmi_hrl.py fetch grassland	Source HRL GeoTIFFs (~860 MB)	—
3	data/cache/rasters/*.tif	scripts/build_cache.py rasters	LZW + ISN93-projected GeoTIFFs (~9 MB each — 98× smaller)	skip 30 s reproject per render
4	data/cache/constants.json	scripts/build_cache.py constants	Iceland total area + 4-CRS bbox + per-source SHA-256 + grassland area	skip 1.5 s polygon area + 826 MB scan per render
5	data/cache/arrays/*.npy	written automatically on first map render	Decoded probability arrays (e.g. GRAVPI)	skip 5 s RGB-decode + reproject

Render scripts read these tiers via scripts/utils/cache.py:

from scripts.utils.cache import iceland_constants, cached_raster, CacheMissingError

try:
    K = iceland_constants()
    iceland_km2 = K["iceland_total_area_km2"]
except CacheMissingError as e:
    print(e.hint)         # tells you the build_cache.py command to run

Tier 3+4 are explicit (build_cache.py all); Tier 5 is opportunistic (populated on first render). Sources are SHA-256 fingerprinted in constants.json so scripts/build_cache.py status flags stale entries when an upstream raster changes.

Benchmark

# Cold (full re-download): ~120 s for grassland
# Warm-raw (rebuild Tier 3+4): ~10 s for grassland
# Warm (steady state): ~3-6 s
uv run python scripts/bench_maps.py run --mode warm
uv run python scripts/bench_maps.py history

Results land in data/cache/benchmarks.json with timestamps and a delta vs. the last warm baseline.

Tests

# Cache-source consistency + map-output smoke tests
uv run pytest tests/test_cache_consistency.py tests/test_maps_render.py -v

Troubleshooting

• "Missing data/cache/constants.json" → run scripts/build_cache.py all
• "STALE — source changed" in status → run scripts/build_cache.py rasters
• Render is slow despite cache → check scripts/build_cache.py status; the tier-5 .npy file may be missing for grassland_probability_heatmap.py (auto-rebuilds on next run).

Python Code Templates

Load a layer with geopandas

import geopandas as gpd
from pathlib import Path

GEODATA = Path("data/geodata")
land = gpd.read_file(GEODATA / "Landmask.geojson")
municipalities = gpd.read_file(GEODATA / "AdministrativeUnit_level2.geojson")

Simple choropleth map

import geopandas as gpd
import matplotlib.pyplot as plt
from pathlib import Path

GEODATA = Path("data/geodata")
land = gpd.read_file(GEODATA / "Landmask.geojson")
munic = gpd.read_file(GEODATA / "AdministrativeUnit_level2.geojson")

# Merge your data onto municipalities
# munic = munic.merge(your_data, left_on="namn", right_on="municipality")

fig, ax = plt.subplots(figsize=(12, 8), facecolor="#f0f4f8")
ax.set_facecolor("#c8d6e5")
land.plot(ax=ax, color="#f5f0e6", edgecolor="#2d3436", linewidth=0.6)
munic.plot(ax=ax, column="your_metric", cmap="YlOrRd", legend=True,
           edgecolor="#636e72", linewidth=0.3, alpha=0.8)
ax.set_xlim(-24.7, -13.1)
ax.set_ylim(63.2, 66.6)
ax.set_aspect(1 / 0.42)  # latitude correction at 65°N
ax.set_title("Your Title")
plt.tight_layout()
fig.savefig("reports/choropleth.png", dpi=200, bbox_inches="tight")

Overlay custom points

import polars as pl
import geopandas as gpd
import matplotlib.pyplot as plt

GEODATA = Path("data/geodata")
land = gpd.read_file(GEODATA / "Landmask.geojson")

# Your point data
pts = pl.read_csv("data/processed/your_points.csv")

fig, ax = plt.subplots(figsize=(12, 8))
land.plot(ax=ax, color="#f5f0e6", edgecolor="#2d3436", linewidth=0.6)
ax.scatter(pts["lon"], pts["lat"], c="#e74c3c", s=20, zorder=10)

Interactive HTML with custom data overlay

import json
from pathlib import Path

GEODATA = Path("data/geodata")

# Load base layers
with open(GEODATA / "Landmask.geojson") as f:
    landmask = json.load(f)

# Build Leaflet HTML with your data embedded as JSON
# Follow the pattern in scripts/kortagerð.py cmd_html()

R Code Templates

Load layers with sf

library(sf)
library(ggplot2)

geodata <- "data/geodata"
land <- st_read(file.path(geodata, "Landmask.geojson"))
munic <- st_read(file.path(geodata, "AdministrativeUnit_level2.geojson"))
roads <- st_read(file.path(geodata, "RoadLines.geojson"))
glaciers <- st_read(file.path(geodata, "LandIceArea.geojson"))
lakes <- st_read(file.path(geodata, "Lake_Reservoir.geojson"))

Basic ggplot2 map

ggplot() +
  geom_sf(data = land, fill = "#f5f0e6", color = "#2d3436", linewidth = 0.3) +
  geom_sf(data = glaciers, fill = "#dfe6e9", color = "#b2bec3", linewidth = 0.2) +
  geom_sf(data = lakes, fill = "#74b9ff", color = "#0984e3", linewidth = 0.2) +
  geom_sf(data = roads, color = "#e17055", linewidth = 0.3) +
  coord_sf(xlim = c(-24.7, -13.1), ylim = c(63.2, 66.6)) +
  theme_minimal() +
  labs(title = "Iceland")

Choropleth with municipality data

# Merge your data onto municipalities
munic_data <- munic %>%
  left_join(your_data, by = c("namn" = "municipality"))

ggplot() +
  geom_sf(data = land, fill = "#f5f0e6", color = "#2d3436", linewidth = 0.3) +
  geom_sf(data = munic_data, aes(fill = your_metric), color = "#636e72", linewidth = 0.2) +
  scale_fill_viridis_c() +
  coord_sf(xlim = c(-24.7, -13.1), ylim = c(63.2, 66.6)) +
  theme_minimal()

Color Palette

Standard colors used by kortagerð.py:

Element	Fill	Stroke
Ocean	#c8d6e5	—
Land	#f5f0e6	#2d3436
Glaciers	#dfe6e9	#b2bec3
Lakes	#74b9ff	#0984e3
Rivers	#0984e3	—
Major roads	#d63031	—
Minor roads	#e17055	—
Nature parks	#a8e6cf	#00b894
Municipalities	#b2bec3 (dashed)	—
Highlight	#ffeaa7	#fdcb6e
Settlements	#2d3436	#fff

Tips

• Latitude correction: At 65°N, set ax.set_aspect(1/0.42) in matplotlib or the map will appear stretched
• WatercourseLine has 28k features — skip for faster rendering on zoomed maps
• Municipality names are in the namn column of AdministrativeUnit_level2.geojson
• Settlement population is in the ppl column of BuiltupAreaPoints.geojson
• Road classification: filter by rtt field (1-3 = highways, 4-10 = regional, >10 = local)
• For extra layers not pre-cached: uv run python scripts/lmi.py fetch ERM:WetlandArea