DATAKILDE · MODELS
Models (26)
Where no register has the answer, we compute it: noise from roads, rail and air where the noise maps stop, settlement under the building from satellite measurements, how far above storm-surge level the house stands, the snow load where it stands, and more. Every model states its method and version, its inputs, what it is checked against — and what it is not.
Use the models in the API
Address models are ordinary sections of the address API: the same key, grants and lookup count as any other section. Three calls cover most uses:
# Which models exist, and how they are made
curl https://api.approach.no/api/v1/addresses/models
# Every model for one address
curl -H "X-API-Key: $DATAKILDE_KEY" "https://api.approach.no/api/v1/addresses/{id}?package=models"
# One model
curl -H "X-API-Key: $DATAKILDE_KEY" "https://api.approach.no/api/v1/addresses/{id}?include=noise_model"Every record has method, method_version, inputs and validation. A validation value not yet measured is null with status not_yet_computed — we never state a figure we have not measured. Addresses the model has not reached yet answer not_found with a reason, not empty. In the MCP server the tool is datakilde_models.
Try an active model with your account or selected API key
Run a model on any point
The sections are precomputed per address. Models released in the Model API can also be run on demand for a point, with a fixed version and release — the same key, but its own models access. The cards below show which sections are also available there.
curl https://api.approach.no/api/v1/predictions/catalog
curl -X POST -H "X-API-Key: $DATAKILDE_KEY" -H "Content-Type: application/json" \
-d '{"model":"noise_exposure","subject":{"address_id":"{id}"}}' https://api.approach.no/api/v1/predictions/runNo model has an active release in the Model API yet. A model is released only after its validation is approved.
The Model API reference — its catalogue lists only models with an active release.
Noise and traffic
- Modelled noise exposure by source (screening calculation)noise_modelHow much noise reaches the address from roads, rail and air, and which source dominates?Physical calculation · Inputs: NVDB ÅDT, heavy share and speed limits, Entur rail traffic, Avinor movements, BRREG activity, Kartverket DSM line of sightChecked against: official END bands and T-1442 zones, shown beside the result (agreement_with_mapped)Also in the Model API: noise_exposure (awaiting an approved release)Read first: exposure_index, best_estimate_lden_db, lden_db, lnight_db
- Road traffic around the address in 100 m, 500 m and 1 km discstraffic_exposureHow much traffic passes within 100 m, 500 m and 1 km?Computed from terrain and surface models · Inputs: NVDB ÅDT, Trafikkdata sensorsChecked against: none: an exposure measure, not a predictionRead first: max_adt_100m, traffic_load_100m, adt_segments_over_5000_100m, adt_segments_over_15000_100m
Environmental nuisance
- Modelled ship traffic and at-berth emissions near a waterfront addressship_exposureHow much ship traffic and berth emission is near a waterfront address?Index of documented components · Inputs: Kystverket AIS, ship register, port callsChecked against: none yetRead first: vessel_hours_500m, vessel_hours_1km, cruise_calls_365d, berth_nox_index_kg
- Nuisance composite: noise, air quality, radon, contaminated ground, power lines and ship traffic, each kept separatenuisance_indexHow much combined environmental nuisance (noise, air, radon, contaminated ground, power lines, ships) is there at the address?Index of documented components · Inputs: noise_model, noise zones, air-quality forecast, radon, contaminated sites, power lines, ship_exposureChecked against: none: each component is kept separate and has its own sourceRead first: nuisance_index, noise_score, air_quality_score, radon_score
Ground
- Ground-integrity screening scoreground_integrityHow sound is the ground under the address?Index of documented components · Inputs: løsmasser, marine clay, quick-clay and flood zones, NADAG boreholes, landslide eventsChecked against: none: a screening scoreRead first: score
- Building-level ground subsidence from InSAR (vertical velocity and differential settlement)ground_subsidenceIs the building settling, and unevenly?Statistical model · Inputs: InSAR Norway (NGU), building footprints, marine clay, løsmasserChecked against: settlement on marine clay and fill versus elsewhere; known settlement areasRead first: vertical_mm_per_year, settlement_p90_mm_per_year, differential_mm_per_year_per_10m, subsidence_class
Climate and natural hazards
- Modelled surface-water indicatorswater_riskWhere does surface water collect and run at the address?Computed from terrain and surface models · Inputs: DTM10 (Kartverket)Checked against: none yet: not a flood modelRead first: depression_score, hand_score, flow_accumulation_score, slope_score
- Climate and natural hazard exposure indexhazard_exposureHow exposed is the address to climate and natural hazards overall, and which hazard drives it?Index of documented components · Inputs: NVE hazard zones, NGU, Kartverket storm surge and sea level, Klimaservicesenteret projectionsChecked against: none: the worst component plus documented adjustmentsAlso in the Model API: hazard_exposure (awaiting an approved release)Read first: hazard_index, hazard_level, dominant_hazard
- Modelled wildfire exposure (wildland-urban interface)wildfire_exposureHow exposed is the address to wildfire?Index of documented components · Inputs: MET fire weather index, NIBIO SR16 forest, AR5/AR50 land coverChecked against: none yetRead first: wildfire_exposure_index, high_fwi_day_share, flammable_share_100m, flammable_share_500m
- Storm-surge freeboard at the address (model)storm_surge_freeboardHow many metres above storm-surge levels is the address, now and in 2050 and 2100?Physical calculation · Inputs: Kartverket tide API return levels and land uplift, DTM1 ground height, sea-level projectionsChecked against: Kartverket storm-surge polygons (inside polygon ⇔ freeboard below 0)Read first: freeboard_20y_present_m, freeboard_200y_present_m, freeboard_1000y_present_m, freeboard_200y_2050_m
- Modelled 50-year ground snow load for the address's 1 km cell (seNorge)snow_loadWhat is the 50-year ground snow load where the building stands?Statistical model · Inputs: seNorge snow water equivalent (NVE/MET), 1957 onwardsChecked against: out-of-sample exceedance of the fitted 10-year valueRead first: ground_snow_load_50y_kn_m2, trend_30y_kn_m2_per_decade
Energy
- Bedrock heat-pump feasibility: depth to bedrock, nearby energy wells, drilling constraints and indicative drilling lengthheat_pump_feasibilityCan a bedrock heat pump be drilled here, and how deep is the rock?Statistical model · Inputs: GRANADA energy wells (NGU), quick-clay zones, contaminated sites, building heat demandChecked against: leave-one-out depth to bedrock against GRANADA wellsRead first: depth_to_bedrock_m, nearby_energy_wells_200m, drilling_constraints, indicative_drilling_m
- Rooftop solar potential of the main building (model)solar_potentialHow much solar power can the roof produce, and what is it worth?Computed from terrain and surface models · Inputs: NVE solar irradiance, DOM1 roof planes, sun_hours horizon, electricity price and grid tariffChecked against: PVGIS modelled yieldRead first: usable_roof_area_m2, system_kwp, annual_yield_kwh, kwh_per_kwp_best_plane
- Estimated energy label of the main building (model, not an energy certificate)energy_label_estimateWhich energy label is a building without a certificate likely to have?Machine learning · Inputs: Enova energy certificates, Matrikkel building facts, GRANADA energy wells, district-heating areasChecked against: grouped cross-validation by kommune against Enova certificatesRead first: estimated_label, probabilities, confidence
Terrain, sun and building
- Elevation, slope, aspect and relief of the ground at the addressterrainWhat is the terrain at the address: height, slope, aspect and position?Computed from terrain and surface models · Inputs: DTM10 (Kartverket)Checked against: none: derived directly from the terrain modelRead first: elevation_m, slope_deg, slope_pct, aspect_deg
- Trees and vegetation near the buildingvegetationHow close are trees, and how much canopy is around the building?Computed from terrain and surface models · Inputs: DOM1 and DTM1 (Kartverket)Checked against: none yetRead first: nearest_tree_crown_m, trees_within_10m, trees_within_30m, tallest_tree_within_10m_m
- Modelled sun hours on the plotsun_hoursHow many hours of direct sun does the address get through the year?Computed from terrain and surface models · Inputs: DSM and DTM (Kartverket)Checked against: none yet: clear-sky horizon modelRead first: sun_hours_jun21, sun_hours_mar21, sun_hours_dec21
- Modelled view from the plotviewshedHow open is the view, and is there water in it?Computed from terrain and surface models · Inputs: DTM10 (Kartverket)Checked against: none: terrain onlyRead first: visible_share, farthest_visible_m
- Building height above ground and estimated storeys from the national surface modelbuilding_heightHow tall is the building, and how many storeys does that suggest?Computed from terrain and surface models · Inputs: DOM1 and DTM1 (Kartverket), building footprintsChecked against: Matrikkel storeys where registeredRead first: height_p95_m, height_median_m, estimated_storeys
Safety
- Fire and burglary exposure indices for the address, with every factor shownfire_and_burglary_indicatorsHow do fire and burglary risk factors stack up at the address?Index of documented components · Inputs: Matrikkel, SEFRAK, NVDB, BRREG, SSB, police log, KOSTRAChecked against: none: indices, not probabilitiesRead first: fire_index, burglary_index, fire_factors, burglary_factors
- Building fire risk relative to the national rate (model from area fire statistics)fire_riskHow does the building-fire rate for buildings like this one compare with the national average?Machine learning · Inputs: DSB BRIS building fires per postnummer, Matrikkel building year, type, storeys, energy source, SEFRAK, SSB population and income, fire_response_timeChecked against: held-out years of DSB BRIS building fires per postnummer: deviance against the national rate, calibration by decile, rank correlationAlso in the Model API: building_fire_risk (awaiting an approved release)Read first: relative_risk, risk_band, area_expected_fires_per_1000_buildings_per_year
- Modelled magnetic field from overhead power lines, against DSA's 0.4 µT investigation levelpower_line_fieldWhat magnetic field could nearby power lines give at the address?Physical calculation · Inputs: NVE power lines and voltageChecked against: no field measurements available; current is assumedRead first: field_typical_ut, field_high_load_ut, above_investigation_level_typical, above_investigation_level_high_load
- Estimated fire-service response time from the nearest staffed fire station, against the § 22 requirementfire_response_timeHow long would the fire service take to arrive?Physical calculation · Inputs: DSB fire stations and staffing, road network routingChecked against: plausibility against BRIS mission statisticsRead first: estimated_response_minutes, within_requirement, requirement_minutes, station_name
Access and travel
- Walkability (15-minute city) scorewalk_scoreHow much is within walking distance?Index of documented components · Inputs: OpenStreetMap, BRREG, schools, kindergartens, public-transport stopsChecked against: none: a weighted countRead first: score
- Public-transport travel time at 07:30 on a weekday to the kommune centre and the nearest regional hubcommute_timeHow long is the trip by public transport to the kommune centre and the nearest regional centre?Physical calculation · Inputs: Entur journey plannerChecked against: none: no observed travel timesRead first: kommune_centre_minutes, hub_minutes, hub_name, kommune_centre_name
Economy
What a model is not
An estimate is never the official figure it approximates: the noise estimate is not a T-1442 noise mapping, the energy label is not an energy certificate, and the snow load is not the building’s design load. The first caveat of each section says what the estimate is not.