Marktstammdatenregister plotter

Animated choropleth maps of installed wind & solar capacity in Germany, driven by data scraped from the public Marktstammdatenregister (MaStR).

Wind	PV (all sizes)

Installed wind (left) and PV (right, full registry ~4.86 M plants, all sizes) capacity per Kreis, 2005 → May 2026. 22 yearly frames plus a final May 2026 YTD snapshot from the live registry. Fixed Jenks bins so colors stay comparable across years. Available as .gif (universal autoplay) and .mp4 (LinkedIn-native, ~30 % smaller, sharper) — see fig/.

Warning

Research-quality code — designed to run once. Most of it is AI-generated and occasionally very verbose. Nothing has been cleaned up. Good luck.

---

What it does

The MaStR registry contains every grid-connected electricity-generating unit in Germany — millions of solar panels, tens of thousands of wind turbines, plus hydro, biomass, gas and more. Each record carries an install date, a capacity in kW, geographic coordinates, and a pile of enum-encoded metadata.

This repo:

1. loads the registry — primary path is the daily-fresh open-mastr bulk SQLite snapshot (pixi run db-mastr-core → data/mastr/open-mastr.db, ~6 GB, gitignored); fallback paths are the Zenodo parquet dump under BNetzA_MaStR/ and the legacy JSON-API scrape (parser.py),
2. joins turbines to German county polygons extracted from OSM,
3. renders one choropleth PNG per month from the year 2000 to today, and
4. assembles the frames into an animated GIF with ffmpeg.

Data sources

Source	Refresh cadence	Coverage	Pixi task
open-mastr SQLite (default)	daily, full registry	6.1 M PV + 42 k wind + 2.5 M storage	pixi run db-mastr-core
Zenodo parquet	snapshot @ 2025-02-09	full registry at cutoff	manual download to BNetzA_MaStR/
MaStR JSON API	ad-hoc, top-N only	top 200 k by power	pixi run scrape-non-pv, scrape-bess, etc.

Loaders auto-detect the strongest source available; explicit selection via mastr_plot.load_from_bulk(tech, source="sqlite" \| "zenodo" \| "auto").

Pipeline at a glance

Module architecture

Frame loop

---

Interactive notebooks

Two marimo reactive notebooks ship with the repo, plus their pre-rendered HTML exports in docs/:

Notebook	Source	Static HTML	Controls
PV explorer	pv.py	docs/pv.html	date · installation type · bin count · colormap
Wind explorer	wind.py	docs/wind.html	date · onshore/offshore · bin count · colormap

python -m marimo edit pv.py            # reactive editor
python -m marimo run wind.py           # read-only app
python -m marimo export html pv.py -o docs/pv.html

When no data-*.json or germany_kreise.gpkg are present, both notebooks fall back to a synthetic demo dataset so they always render.

Sample renders (real MaStR data)

Generated from the open-MaStR Zenodo full registry dump (cutoff 2025-02-09) merged with the MaStR API delta through May 2026. Wind: full registry (~42 k turbines, every Kreis covered). PV: full registry (~4.86 M plants, including all rooftop and balcony solar — no power threshold). BESS: top 200 k storage units by installed power (covers Pumpspeicher down to ~10 kW home batteries).

2026-05-01 snapshot · real data: 32 107 wind turbines active, 79.7 GW total installed. PV: full registry, ~4.86 M plants. Offshore: 1 732 turbines · 10.4 GW (Nordsee + Ostsee). BESS: 193 599 batteries active · 6.49 GW / 9.4 GWh.

Per-unit size-bin distribution

Where the GW + GWh actually sit when you sort by single-unit nameplate power. Same bins used for all three technologies — wind almost entirely 1-10 MW; PV spread across 10 kW – 100 MW; BESS bimodal (residential 10-100 kW + utility 100-1000 MW).

Wind	PV

Installed capacity by Bundesland

Niedersachsen leads onshore wind (12.7 GW). Bayern leads utility-scale PV (12.0 GW). Schleswig-Holstein and Brandenburg both top 8 GW wind. The five biggest Bundesländer carry ~70 % of national capacity.

Who owns Germany's grid?

Top 30 by combined wind + PV. Offshore project vehicles dominate the very top (DanTysk Sandbank, EnBW Hohe See, Baltic Eagle, Borkum Riffgrund 2…) because every farm is its own LLC. Consolidating by parent would put RWE, Iberdrola and Ørsted at the top.

Offshore wind detail

1 732 turbines / 10.4 GW offshore — 8.6 GW Nordsee + 1.8 GW Ostsee. MaStR anonymises offshore coordinates, so this chart groups by operator instead of mapping individual turbines.

Energy-type mix

All registered electricity-generating units in this scrape, by Energietraeger. Coal (Braunkohle + Steinkohle, 31 GW) is now decisively below utility-scale PV (73 GW). Erdgas (36 GW) carries most of the peaking load.

PV orientation

South-facing dominates at 62 % of installed PV; east-west flat-mount (utility parks) is the strong second at 14 %. Trackers stay rare — only 0.5 GW. Most plants tilt 0–19° (flat utility footprints).

Current fleet — orientation polar rose (May 2026)

Polar rose of total installed capacity at the snapshot date. Radial distance = GW. South (bottom) dominates at 54 GW — the elongated red lobe. SW and SO shoulders account for another ~22 GW combined. North-facing slivers are barely visible at under 1 GW each. Ost-West dual-pitch panels split 50/50 to the Ost and West lobes. Trackers (0.24 GW) excluded — no fixed azimuth.

Orientation × commissioning year — polar heatmap

Each ring = one commissioning year (2000 → 2025, inner → outer). Color = GW commissioned that year in that compass direction. South (bottom) is 54 GW cumulative — roughly 70× more than North. The outer rings are darkest: the post-2020 build-out dominated by large ground-mount arrays, which still skew strongly south-facing. East-West split panels contribute to the Ost and West sectors (capacity divided 50/50). Ost-West tracking (nachgeführt) excluded.

Wind fleet age + repowering

Decommission wave from 2019 onwards: first EEG cohort reaching its 20-year boundary. Mean per-turbine MW has grown 15× since the late 90s (0.5 → 7 MW), driven by the new 15 MW offshore class.

Battery storage (BESS) — batteries only

MaStR carries 2.5 M storage records under Energieträger=2496. Below covers batteries only (Stromspeichertechnologie = "Batterie"). Pumped-hydro storage gets its own section because it's a fundamentally different technology and reported separately everywhere serious.

Snapshot 2026-05-01 (batteries, active): 193 599 units · 6.49 GW / 9.4 GWh. Pipeline (+ planned-and-permitted): + ~ 8 GW LSS Li-ion by ~ 2028.

Three-sector split (battery-charts.de / BVES / EASE convention)

Sector	Threshold	Batteries active 2026-05-01
HSS Heimspeicher	< 30 kWh	176 411 units · 2.47 GW · 2.54 GWh
CSS Gewerblich	30 kWh – 1 MWh	16 655 units · 0.68 GW · 1.20 GWh
LSS Großspeicher	≥ 1 MWh	526 units · 3.34 GW · 5.65 GWh

Same thresholds as battery-charts.de (RWTH Aachen · Figgener et al.), the BVES market monitor, the EASE European Storage Market Monitor, and the EU SET-Plan flexibility scenarios. Useful for direct cross-reference.

HSS is the headcount story (176k households with PV+battery). LSS is the GW + GWh story (grid-scale Li-ion at 3.3 GW + 8 GW more in the pipeline).

Pumped-hydro storage (PSH)

41 sites · 6.48 GW · 927.5 GWh active at 2026-05-01 — roughly 100× the energy density of Li-ion batteries because PSH was built for multi-hour duration. Median duration ~ 8 h vs ~ 1 h for residential Li-ion.

Per-state breakdown + duration	Top 15 sites

Goldisthal (Thüringen) alone — 4 PSS units × 265 MW × 9.64 GWh — accounts for 1.06 GW + 38.5 GWh.

Capacity density (MW per km²)

Absolute capacity makes large rural Kreise look more impressive than they are. Normalising by area reshuffles the ranking — small coastal Kreise on the wind side, small city-Kreise hosting one or two utility parks on the PV side.

Wind density	PV density

Top per-km² Kreise: Dithmarschen 1.6 MW/km² wind, Straubing Stadt 1.5 MW/km² PV.

Capacity added during 2024

Where Germany actually installed new wind + PV during 2024 — 13.4 GW in this scrape. Leipziger Land alone added 565 MW (utility solar in the former-lignite belt).

Build-out per Bundesland

Cumulative installed capacity per state since 2000. Bayern leads at 22 GW (driven by PV), Niedersachsen second at 19 GW (driven by wind), Brandenburg third at 18 GW (mixed).

PV by installation type

16 000 free-standing utility parks carry ~ 43 GW. 184 000 building-mounted commercial rooftops carry ~ 30 GW. Residential and balcony plants (millions of units, ~ 30 GW) are included in the full-registry scrape.

---

Quickstart

# 1. Install deps (pixi/conda recommended)
pixi add geopandas pandas numpy matplotlib mapclassify pyogrio shapely

# 2. Scrape registry rows (writes data-1.json ... data-7.json)
seq 7 | xargs -P 4 -I{} curl --get \
  'https://www.marktstammdatenregister.de/MaStR/Einheit/EinheitJson/GetErweiterteOeffentlicheEinheitStromerzeugung' \
  --data-urlencode 'sort=' \
  --data-urlencode 'page={}' \
  --data-urlencode 'pageSize=25000' \
  --data-urlencode 'group=' \
  --data-urlencode 'filter=Energieträger~neq~\'2495\'~and~Energieträger~neq~\'2496\'' \
  --data-urlencode 'forExport=true' -o data-{}.json

# 3. Build county polygons (one-time)
osmfilter germany-latest.o5m \
  --keep-nodes="boundary=administrative and ( admin_level=6 or admin_level=4 )" \
  --keep-ways="boundary=administrative and ( admin_level=6 or admin_level=4 )" \
  --keep-relations="boundary=administrative and ( admin_level=6 or admin_level=4 )" \
  --drop-version --drop-author \
  -o=germany_admin_levels_4_6.osm
ogr2ogr -f GPKG germany_kreise.gpkg germany_admin_levels_4_6.osm \
  -sql "SELECT name, admin_level, boundary FROM multipolygons \
        WHERE boundary = 'administrative' \
        AND (admin_level = '6' \
             OR (admin_level = '4' AND name IN ('Berlin','Hamburg','Bremen')))" \
  -nlt MULTIPOLYGON -overwrite -nln multipolygons

# 4. Render frames
jupyter nbconvert --to notebook --execute wind.ipynb --output wind.executed.ipynb

# 5. Assemble GIF — see "Animation" section below

---

Getting Marktstammdatenregister data

Yes, the registry offers a full XML export. But the API filters server-side and returns JSON instead of XML, so this repo scrapes that instead:

seq 7 | xargs -P 4 -I{} curl --get \
  'https://www.marktstammdatenregister.de/MaStR/Einheit/EinheitJson/GetErweiterteOeffentlicheEinheitStromerzeugung' \
  --data-urlencode 'sort=' \
  --data-urlencode 'page={}' \
  --data-urlencode 'pageSize=25000' \
  --data-urlencode 'group=' \
  --data-urlencode 'filter=Energieträger~neq~\'2495\'~and~Energieträger~neq~\'2496\'' \
  --data-urlencode 'forExport=true' -o data-{}.json

The filter excludes Energieträger codes 2495 (Solar / PV) and 2496 (Speicher / storage), so the default scrape covers wind, biomass, hydro, gas, etc. ~169 k rows. Rate-limit yourself; the API gets slow under heavy load. For PV use pixi run scrape-pv-top (or scrape-bess for storage) — see below.

Scraping PV separately

The PV slice of the registry holds ~6 M plants (rooftop, balcony, ground). The API returns 25 000 rows per page, so a full pull is ~245 pages and ≈24 GB of JSON — usually not what you want. The repo defaults to the open-MaStR Zenodo full registry parquet (~4.86 M PV plants) merged with an API delta for the live tail. For a quick top-by-capacity JSON slice use the pixi task:

pixi run scrape-pv-top   # 200 k largest plants (8 pages, ~800 MB)

This is Bruttoleistung-desc sorted — captures every plant ≥ 49 kW.

Filter quirk — MaStR silently drops the second clause when ANDed with a different field, e.g. Energieträger~eq~'2495'~and~ArtDerSolaranlageId~eq~'852' returns the same row count as Energieträger~eq~'2495' alone. Filter client-side after loading if you need a narrower slice.

Decoding the rows

MaStR fields are numeric enum codes (e.g. 698 for "Süd-Ost", 853 for "building"). parser.py decodes the six enums that matter for plotting:

Unknown codes become None. The registry adds codes over time, so check parser.PowerPlant.from_json after large data refreshes.

---

Getting map data

County boundaries come from a Germany OSM extract (e.g. geofabrik.de):

osmfilter germany-latest.o5m \
  --keep-nodes="boundary=administrative and ( admin_level=6 or admin_level=4 )" \
  --keep-ways="boundary=administrative and ( admin_level=6 or admin_level=4 )" \
  --keep-relations="boundary=administrative and ( admin_level=6 or admin_level=4 )" \
  --drop-version --drop-author \
  -o=germany_admin_levels_4_6.osm

ogr2ogr -f GPKG germany_kreise.gpkg germany_admin_levels_4_6.osm \
  -sql "SELECT name, admin_level, boundary FROM multipolygons \
        WHERE boundary = 'administrative' \
        AND (admin_level = '6' \
             OR (admin_level = '4' AND name IN ('Berlin','Hamburg','Bremen')))" \
  -nlt MULTIPOLYGON -overwrite -nln multipolygons

admin_level=6 is Kreis / Landkreis. The three city-states (Berlin, Hamburg, Bremen) are admin_level=4, so they get pulled in by name.

Gotcha — Hamburg's MultiPolygon contains "Nationalpark Hamburgisches Wattenmeer" as part-id 2. The notebook strips it explicitly. If you regenerate the GPKG from a newer OSM extract, double-check the part-id has not shifted.

---

Turning results into GIFs

It is surprisingly hard to turn a folder of PNGs with names like wind-2007-04.png into a video with ffmpeg, which insists on frame%03d.png. The block below renames everything to a tmpdir first, then runs palettegen + paletteuse for a clean palette:

set -l file wind
set -l frames_to_repeat 120
mktemp -d | read -l temp_dir
and cp "$file"-*.png $temp_dir
and begin
    set -l i 1
    set -l last_frame_path ""
    for f in (ls "$temp_dir/$file"*.png | sort)
        mv $f (printf "%s/frame%03d.png" $temp_dir $i)
        set last_frame_path (printf "%s/frame%03d.png" $temp_dir $i)
        set i (math $i + 1)
    end
    set -l current_duplicate_index $i
    for j in (seq 1 $frames_to_repeat)
        cp "$last_frame_path" (printf "%s/frame%03d.png" $temp_dir $current_duplicate_index)
        set current_duplicate_index (math $current_duplicate_index + 1)
    end
end
and ffmpeg -framerate 30 -i "$temp_dir/frame%03d.png" \
    -vf "scale=-1:1200:flags=lanczos,split[s0][s1];[s0]palettegen[p];[s1][p]paletteuse=dither=none" \
    -loop 0 -y "$file".gif
and rm -rf "$temp_dir"

The final-frame repetition keeps the last month on screen for ~4 seconds before the GIF loops. Drop -loop 0 if you want a one-shot.

Fun fact: This whole renaming dance exists because PNGs have no timestamp metadata. JPGs with EXIF would just work with ffmpeg's glob pattern.

---

Layout

File / dir	Purpose
parser.py	PowerPlant dataclass + JSON-to-record decoder
mastr_plot.py	Shared helpers (load, aggregate, choropleth) + synthetic demo data
pixi.toml	Reproducible env: pixi install then pixi run pv-edit / wind-edit / docs-build
pv.py	Marimo notebook — interactive PV explorer
wind.py	Marimo notebook — interactive wind explorer
wind.ipynb	Original Jupyter notebook: load → join → plot → save frames
fig/	Rendered PNG/GIF outputs (gitignored), plus pipeline + sample SVGs
docs/	Read-the-Docs–style site published at maykthewessen.github.io/marktstammdatenplotter
CLAUDE.md	Conventions for Claude Code agents
CITATION.cff	Citation metadata for the software + upstream datasets
tests/	pytest suite covering parser.py enum decoders (52 tests)
scripts/	CI helpers: render_samples.py, render_wind_gif.py, build_kreise_json.py, build_downloads.py
docs/data/	Bulk downloads — mastr-snapshot.parquet + CSV.gz mirrors
CHANGELOG.md	Notable changes per version (Keep-a-Changelog format)
CONTRIBUTING.md	How to add a chart / parser enum, style + commit rules

---

What this method does not track

The choropleths and animations cover the full PV registry (all sizes) and the complete wind fleet, but a few categories are systematically excluded — listed explicitly so nobody draws the wrong conclusion:

Excluded	Why	Approx scale
Records with NaT install_date	Snapshot filter is install_date ≤ snap	~ 15 k rows / 71 GW (mostly legacy conventional)
Records with NaN Laengengrad / Breitengrad	Spatial join drops them	~ 76 k rows / 0.8 GW
Offshore turbines in choropleths	No Kreis covers open sea	1 732 turbines / 10.4 GW (reported separately)
Battery storage (Speicher)	Energieträger codes 2495 + 2496 filtered out	several GW BESS
Heat-only plants (Wärme)	Outside Wind+PV scope	~ 2 GW thermal
Plants approved but not commissioned	No InbetriebnahmeDatum yet	several GW pipeline
Behind-the-meter / unregistered Balkonkraftwerke	Registration was optional pre-2017	~ 0.5 GW
Plants decommissioned before snapshot	Removed by removal_date > snap filter	3 GW cumulative since 2013

Offshore lat/lon: contrary to the original notebook's assumption, MaStR does carry real coordinates for offshore plants. The StandortAnonymisiert field is only a sea label ("Nordsee…" / "Ostsee…"). Drop the synthetic-sea workaround if you want point-accurate offshore positions.

Credits

Forked from emmericp/marktstammdatenplotter. Data © Marktstammdatenregister, Bundesnetzagentur. OSM © OpenStreetMap contributors.