English Español Français

Power-line risk for birds

Relative index (0-100, 99th percentile = 100) per ~500 m cell, species and month, built from OpenStreetMap, GBIF/eBird, Movebank and terrain data. It ranks which pylons, line segments and wind turbines to fix first; it does not certify mortality. Rebuilt on the 2nd of every month.

Project overview · code and method

7regions computed
116,700elements scored

Spain

Cantabria and surroundings (Spain)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

46,418 elements scored: 4,540 power lines, 41,415 pylons, 463 wind turbines

  1. pylon · 570 · Feb · Peregrine Falcon, Golden Eagle, Red Kite · OSM
  2. pylon · 569 · Feb · White Stork, Red Kite, Griffon Vulture · OSM
  3. power line · 486 · Jul · Booted Eagle, European Honey Buzzard, Short-toed Snake Eagle · OSM
  4. pylon · 438 · Feb · Peregrine Falcon, Golden Eagle, Red Kite · OSM
  5. pylon · 438 · Feb · White Stork, Red Kite, Griffon Vulture · OSM

Strait of Gibraltar / Campo de Gibraltar (Cádiz, Spain)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

2,687 elements scored: 414 power lines, 1,757 pylons, 516 wind turbines

  1. pylon · 433 · Nov · Cinereous Vulture, Red Kite, Bonelli’s Eagle · OSM
  2. power line · 351 · Nov · Cinereous Vulture, Bonelli’s Eagle, Peregrine Falcon · OSM
  3. pylon · 333 · Nov · Cinereous Vulture, Red Kite, Bonelli’s Eagle · OSM
  4. power line · 292 · Oct · Black Stork, Bonelli’s Eagle, Short-toed Snake Eagle · OSM
  5. pylon · 260 · Nov · Cinereous Vulture, Red Kite, Bonelli’s Eagle · OSM

France

Camargue (Bouches-du-Rhône and Gard, France)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

5,325 elements scored: 1,272 power lines, 4,053 pylons

  1. power line · 270 · Jul · Eurasian Spoonbill, Purple Heron, Black Kite · Enedis dataset
  2. pylon · 254 · Jun · Short-toed Snake Eagle, Eurasian Spoonbill, White Stork · Enedis dataset
  3. pylon · 231 · Jun · Short-toed Snake Eagle, Eurasian Spoonbill, White Stork · Enedis dataset
  4. power line · 231 · May · Eurasian Spoonbill, Greater Flamingo, Black Kite · Enedis dataset
  5. power line · 214 · May · Eurasian Spoonbill, Black Kite, Greater Flamingo · Enedis dataset

Grands Causses (Lozère and Aveyron, France)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

19,236 elements scored: 3,537 power lines, 15,688 pylons, 11 wind turbines

  1. power line · 464 · Aug · Cinereous Vulture, Bearded Vulture, Griffon Vulture · Enedis dataset
  2. power line · 425 · Aug · Cinereous Vulture, Bearded Vulture, Griffon Vulture · Enedis dataset
  3. power line · 421 · Aug · Bearded Vulture, Short-toed Snake Eagle, Red-billed Chough · Enedis dataset
  4. pylon · 408 · Aug · Cinereous Vulture, Bearded Vulture, Griffon Vulture · Enedis dataset
  5. power line · 406 · Aug · Cinereous Vulture, Bearded Vulture, Griffon Vulture · Enedis dataset

Champagne humide (Lac du Der and Forêt d’Orient, France)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

14,466 elements scored: 3,961 power lines, 10,322 pylons, 183 wind turbines

  1. pylon · 1408 · Sep · Black Stork, European Honey Buzzard, Red Kite · Enedis dataset
  2. pylon · 987 · Nov · White-tailed Eagle, White Stork, Grey Heron · OSM
  3. pylon · 832 · Sep · White Stork, European Honey Buzzard, Red Kite · Enedis dataset
  4. pylon · 821 · Sep · White Stork, European Honey Buzzard, Red Kite · Enedis dataset
  5. pylon · 757 · Sep · Black Stork, European Honey Buzzard, Red Kite · Enedis dataset

La Crau (Bouches-du-Rhône, France)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

7,322 elements scored: 1,952 power lines, 5,359 pylons, 11 wind turbines

  1. power line · 299 · May · Little Bustard, Lesser Kestrel, Black Kite · Enedis dataset
  2. power line · 282 · May · Little Bustard, Lesser Kestrel, Black Kite · Enedis dataset
  3. pylon · 265 · Jul · Egyptian Vulture, Black Kite, Short-toed Snake Eagle · Enedis dataset
  4. pylon · 256 · Jul · Lesser Kestrel, Black Kite, Egyptian Vulture · Enedis dataset
  5. pylon · 255 · Aug · Eurasian Eagle-Owl, Bonelli’s Eagle, Short-toed Snake Eagle · Enedis dataset

Western Pyrenees (Pyrénées-Atlantiques and Hautes-Pyrénées, France)

updated on 2026-10-16

Report · Interactive map · Ranking CSV · GeoJSON · GeoTIFF

21,246 elements scored: 6,852 power lines, 14,394 pylons

  1. power line · 488 · Sep · Black Stork, Peregrine Falcon, Egyptian Vulture · Enedis dataset
  2. pylon · 483 · Jan · Golden Eagle, Peregrine Falcon, Red-billed Chough · Enedis dataset
  3. pylon · 465 · Mar · Egyptian Vulture, Black Kite, Bearded Vulture · Enedis dataset
  4. power line · 440 · Oct · Common Crane, Bearded Vulture, Golden Eagle · Enedis dataset
  5. pylon · 439 · Jan · Golden Eagle, Peregrine Falcon, Red-billed Chough · Enedis dataset

Generated on 2026-10-16.

How it is computed

Spain

  1. Infrastructure. Power lines, pylons and wind turbines are downloaded from OpenStreetMap (Overpass). Lines up to 66 kV are distribution (electrocution risk, the scope of Spanish Royal Decree 1432/2008); above that they only count for collision. Pylons inherit the voltage of their line. Each 0.005° cell gets km of line, distribution pylons and wind turbines. Where OSM has no pylons mapped, 6 per km of distribution line are assumed; the count is capped at 20 pylons per cell (more than that is a substation).
  2. Pylon hazard. Each pylon carries a hazard factor of 0.5-2 derived from its OSM tags (material, insulator type, branch connections, cross-arm design; criteria of Royal Decree 1432/2008), or 1.0 when it has no tags; that factor also scales its position in the ranking.
  3. Presence. Monthly relative frequency = density of records of the species / density of records of all birds (sampling effort), smoothed (σ = 1,500 m with at least 300 records in the month, wider with fewer). Records repeated at the same spot, month and year are weighted by the square root of their count. GBIF data from 2010 onwards, eBird included. Where GPS tracking (Movebank) is available it is blended in as individual-days per cell with a weight of up to 50%, proportional to the number of fixes (full weight from 2,000 on). Where eBird Status & Trends GeoTIFFs are present in data/ebirdst/ they carry 70% against the GBIF frequency.
  4. Terrain. Slope (collision with conductors) and topographic position at 1,500 m (ridges, wind turbines).
  5. Species weights (electrocution / line collision / turbine collision / conservation status) are defined in birdrisk/config.py, drawn from the literature. They are a starting point, not a fit to mortality data. In the aggregate index each species is weighted by status x confidence, with confidence = max(min(1, records/500), min(1, GPS fixes/2000)), or 1 with an eBird model: a species with few records has an unreliable map and must not dominate the ranking.
  6. Index and ranking. Per species, month and cell: electrocution = presence x weight x equivalent pylons; line collision = presence x weight x km of line x (1 + relief); turbine collision = presence x weight x turbines x (1 + ridge). Each index is rescaled so that the 99th percentile of the cells with infrastructure in the region is 100, without clipping the maxima. Each pylon takes the electrocution index of its cell times its hazard factor; each line segment, the mean along its route; each wind turbine, the index of its cell. The peak month and the species that contribute most are kept.

Parameters

Grid cell0.005°
Maximum distribution voltage66 kV
Pylons assumed where none are mapped6 per km
Pylon cap per cell20
Record smoothing (σ)1,500 m with ≥300 records in the month
GBIF recordsfrom 2010, up to 20,000 per species
Minimum records to include a species30
Full confidence500 records or 2,000 GPS fixes
Maximum GPS weight (Movebank)50%
eBird Status & Trends weight70%
Slope at which line collision doubles25°
Topographic position radius1,500 m
Scale99th percentile = 100

Validation

Pending. The ranking has not yet been checked against real mortality: there is no open dataset of electrocuted or collided birds in Spain. The first planned check is against the Basque Government's shapefile of hazardous power lines (region euskadi); once mortality points are available, the species weights will be fitted with a logistic regression.

Limits

France

The French regions use the same model as the Spanish ones; what changes is where the infrastructure comes from, the voltage threshold and the conservation status.

  1. Distribution network: Enedis. Enedis, which runs about 95% of mainland France's distribution network, publishes every overhead medium-voltage (HTA) line (réseau HTA, Licence Ouverte 2.0) and the position of its poles, HTA or low voltage (poteaux HTA et BT, ODbL). They are downloaded for the region's bounding box. The poles file does not say which network each pole belongs to: a pole within 10 m of an overhead HTA line is counted as an HTA pole. The file is incomplete: it holds about half the poles one would expect on the HTA lines, so the pylons mapped in OSM fill the gaps.
  2. Transmission lines and wind turbines: OpenStreetMap. RTE's lines and pylons (63 kV and above) and the wind turbines come from OpenStreetMap (Overpass), as in Spain. Lines up to 50 kV are distribution (electrocution and collision); above that, transmission (collision only). In Spain the threshold is 66 kV (Royal Decree 1432/2008); in France the HTA network ends at 50 kV and 63 kV is already RTE's.
  3. Merging both sources. An OSM distribution line that runs within 25 m of Enedis HTA for at least 50% of its length is the same line and is dropped; the others stay (local distributors such as SICAE or régies, which publish no data). An OSM pylon within 15 m of an Enedis pole is the same pylon: the Enedis pole takes its OSM tags (material, insulators, cross-arm), which set its hazard factor, and the OSM copy is dropped. OSM pylons with no Enedis twin are kept (they are the poles missing from the Enedis file), except those tagged below 1 kV (low voltage). Where a cell has fewer mapped pylons than its line length implies, the model estimates the rest, as in Spain, so nothing is counted twice.
  4. French territory only. Every element outside the French departments that cross the region (IGN Admin Express) is clipped away: in the Pyrenees the bounding box takes in part of Spain, where there is no Enedis data and the count would be lower for that reason alone. The birds are counted over the whole box, since they fly across the border.
  5. Presence, terrain, index and ranking as in Spain: GBIF/eBird records corrected for sampling effort (density of the species' records / density of all bird records, smoothed at σ = 1,500 m), Movebank where available, slope and ridges, then a 0-100 index per species, month and 0.005° cell. Electrocution = presence x weight x pylons; line collision = presence x weight x km of line x (1 + relief); turbine collision = presence x weight x turbines x (1 + ridge).
  6. Conservation status. Each species weighs according to the Red List of the breeding birds of mainland France (UICN France, MNHN, LPO, SEOF, ONCFS, 2016): CR 1, EN 0.9, VU 0.8, NT 0.6, LC 0.4-0.5. Common Crane and White-tailed Eagle are weighted as the wintering birds they are in these regions.
  7. Natura 2000 sites come from the European Environment Agency (SPA = ZPS, SAC = ZSC).

Parameters

Grid cell0.005°
Maximum distribution voltage50 kV
Pylons assumed where none are mapped6 per km
Pylon cap per cell20
Record smoothing (σ)1,500 m with ≥300 records in the month
GBIF recordsfrom 2010, up to 20,000 per species
Minimum records to include a species30
Full confidence500 records or 2,000 GPS fixes
Maximum GPS weight (Movebank)50%
eBird Status & Trends weight70%
Slope at which line collision doubles25°
Topographic position radius1,500 m
Scale99th percentile = 100
Enedis pole counted as HTA≤10 m from an HTA line
OSM pylon and Enedis pole are the same≤15 m apart
OSM line duplicated by Enedis≥50% of its length ≤25 m from HTA

Validation

Pending, as in Spain: there is no open dataset of birds killed by power lines in France to check the ranking against. The LPO and the national action plans keep records that are not open.

What is left out, and why

Limits