Road safety dataset.
Open statistics 2015–2026.
Municipal collision severity, vehicle fleet normalization, and wildlife impact across 307 Finnish municipalities. Based on 39,594 verified injury collisions and 11 years of Open-Meteo & FMI climate time series.
Open Data Package & Replication
Full 307 municipality master dataset precomputed with Koneisto Core Go Turbine.
@dataset{pelkonen_2026_road_weather,
author = {Pelkonen, Tommi},
title = {{Finnish Road Weather and Traffic Safety Index 2015--2026: Municipal Collision Severity, Vehicle Fleet Normalization, and Wildlife Impact Dataset}},
year = 2026,
month = sep,
publisher = {Zenodo},
version = {1.0.0},
url = {https://rengasrepo.fi/tutkimus},
note = {DOI pending (Zenodo) / Open Science Dataset (CC BY 4.0)}
}# Suomen tiesää- ja onnettomuusdatan toistettavuusanalyysi (Python)
import pandas as pd
import json, urllib.request
# 1. Haetaan 307 kunnan master-aineisto
url = "https://rengasrepo.fi/data/rengaskeli_kuntatilastot_master.json"
req = urllib.request.urlopen(url)
data = json.loads(req.read().decode('utf-8'))
# 2. Muunnetaan kunnat dataframeksi
kunnat = pd.DataFrame.from_dict(data['kunnat'], orient='index')
# 3. Tarkastellaan kolaritiheyttä suhteutettuna 1 000 rekisteröityyn autoon
karkikunnat = kunnat.sort_values(by='kolarit_per_1k_autoa', ascending=False)
print("Top 10 kuntaa - henkilövahinkokolarit per 1 000 autoa (11 v keskiarvo):")
print(karkikunnat[['kunta', 'henkiloautot_liikenteessa', 'kolarit_per_1k_autoa', 'kolarit_per_10k_asukasta', 'kuolemanriski_pct']].head(10))
# 4. Koko maan keliolosuhteet
print("\nKoko maan kelijakauma henkilövahinkokolareissa:")
for keli, pct in data['koko_maa_yhteenveto']['keli_jakauma'].items():
print(f" {keli}: {pct} %")
Five Municipal Risk Archetypes in Finland
When collision figures are normalized per 1,000 cars and per capita, municipal traffic risks diverge sharply.
1. Transit Highway Hubs
Representative municipalities: Heinola, Joutsa, Ii, Lapua
Traffic Risk Anatomy
Heavy transit logistics, highway speeds and narrow passing zones relative to small local vehicle fleet.
Statistical Precaution & Fitting Week
High-speed directional stability and a 10-meter shorter stopping distance save lives. Never enter highway autumn on under 4 mm tread.
National Road Weather & Road Type Evidence (11 Years)
All 39,594 injury collisions recorded by police in Finland (2015–2025) categorized by surface condition and road classification.
Road Surface Conditions (All Finland)
| Condition | Accidents | Share |
|---|---|---|
| Bare, dry | 23 605 | 59,6 % |
| Bare, wet (rain) | 7 417 | 18,7 % |
| Icy (black ice / pack) | 4 253 | 10,7 % |
| Snowy | 2 374 | 6,0 % |
| Slushy (slush-planing) | 1 049 | 2,6 % |
| Winter ruts bare | 663 | 1,7 % |
Icy roads cause almost twice as many injury collisions as snowy roads (10.7% vs 6.0%).
Road Classification & Fatality Rate
| Road Class | Injuries | Deaths | Fatality % |
|---|---|---|---|
| Private / Lake cottage roads | 1 887 | 179 | 9,5 % |
| State highways | 18 972 | 1 436 | 7,6 % |
| Urban municipal streets | 13 835 | 357 | 2,6 % |
⚠️ Private rural roads carry nearly four times higher fatality probability per collision than urban streets.
Oct–Nov Early Winter Collisions: -33.3%
Transitioning from fixed calendar enforcement (1 Dec) to weather-dependent requirement (1 Nov) coincided with modern vehicle safety tech (ADAS, AEB) dropping autumn injuries from 769.0 to 512.6/yr. However, peak day collision spikes on the exact first icy morning have sharpened.
Jyväskylä August Collisions (2018–2024)
When Rally Finland was cancelled in August 2020 due to Covid, Jyväskylä recorded a record-low 6 injury accidents (vs. 15 in August 2018). Normal rally years see an 80–114% surge in August collisions compared to July due to massive regional spectator traffic.
307 Municipalities Safety Ranking 2026
Search any Finnish municipality, explore top lists of safest towns, highway hubs, and cottage road fatality risks.
Thematic Evidence Articles
Detailed investigations on first snow physics, lake cottage gravel dangers, and driver risk homeostasis.