roadnoise

Transparent, screening-level road traffic noise in Python for environmental-noise consultants, planners and researchers working in the context of the EU Environmental Noise Directive (END, Directive 2002/49/EC).

• CNOSSOS-EU road source model – Directive 2002/49/EC Annex II as introduced by Commission Directive (EU) 2015/996 and amended by Commission Delegated Directive (EU) 2021/1226: vehicle categories 1, 2, 3, 4a, 4b (5 reserved), rolling and propulsion noise per octave band 63 Hz – 8 kHz, speed laws, road-surface, studded-tyre, temperature, gradient and acceleration corrections, line-source emission per metre.
• Propagation (screening) – geometric divergence, ISO 9613-1 atmospheric absorption at any temperature / humidity, flat-ground ground effect (CNOSSOS-EU homogeneous + favourable, or ISO 9613-2 general method), single-edge diffraction (CNOSSOS-EU Δdif with ground terms, ISO 9613-2 D_z, or Maekawa), optional +3 dB facade reflection.
• Receivers – segmentation of road lines into point sources, regular grids, facade receivers.
• Indicators – L_day, L_evening, L_night, L_den (5/10 dB penalties, 12/4/8 h or Member-State variants) and END exposure statistics (people per 5 dB band, optional rounding to hundreds).
• CRTN – the UK Calculation of Road Traffic Noise (1988) as an independent quick check.
• I/O – GeoJSON in, ESRI ASCII grid / GeoJSON out; GeoTIFF with the raster extra; geopandas helpers with the geo extra. Only numpy is required.

Scope. roadnoise is a screening tool. It is not a certified or conformance-tested CNOSSOS-EU implementation and must not be used on its own for strategic noise maps reported under the END. See Differences to full CNOSSOS-EU.

Installation

pip install roadnoise              # numpy only
pip install "roadnoise[raster]"    # + rasterio for GeoTIFF output
pip install "roadnoise[geo]"       # + geopandas/shapely helpers

Python 3.9 – 3.13.

Quick start

import roadnoise as rn

# Emission per metre of 1104 cars/h + 96 heavy vehicles/h at 50 km/h
lw = rn.line_sound_power({"1": 1104, "3": 96}, 50)      # dB re 1 pW/m, 8 bands
print(rn.a_weighted(lw))                                 # ~84.1 dB(A)

# A road from AADT, a barrier, a building, facade receivers
road = rn.Road.from_aadt([(0, 0), (400, 0)], aadt=24000, speed=50, heavy=0.08)
barrier = rn.Barrier([(50, 12), (350, 12)], height=3.0)
building = rn.Building([(150, 40), (250, 40), (250, 60), (150, 60)], height=12.0)
pts, _ = rn.facade_receivers(building.footprint, spacing=5.0)

res = rn.compute_levels([road], pts, receiver_height=4.0,
                        barriers=[barrier], buildings=[building])
res.lden, res.lnight          # dB(A) per receiver

grid = rn.compute_grid([road], cell=10.0, margin=100.0, barriers=[barrier])
grid.levels["lden"]           # (nrows, ncols), north-up

table = rn.exposure_statistics(res.lden, population=3.0, round_to=100)
print(table.to_text("People per Lden band"))

See examples/quickstart.py for a complete script.

Command line

roadnoise emission --flow 1200 --speed 50 --heavy 0.08
roadnoise emission --flow 1200 --speed 90 --heavy 0.15 --surface "2-layer ZOAB" --json
roadnoise grid roads.geojson --cell 10 -o lden.asc            # ESRI ASCII grid
roadnoise grid roads.geojson --cell 10 -o lden.tif            # needs roadnoise[raster]
roadnoise grid roads.geojson --cell 10 -o map.asc --indicator all --barriers walls.geojson
roadnoise receivers roads.geojson dwellings.geojson -o out.geojson \
          --population-field population --round-to 100
roadnoise crtn --flow 20000 --speed 75 --heavy-pct 10 --distance 20
roadnoise surfaces

Common propagation options: --method cnossos|iso9613, --ground G, --temperature, --humidity, --max-distance, --segment, --facade, --buildings, --barriers, --assume-projected.

GeoJSON input

Coordinates must be in a projected CRS in metres. Files without a legacy crs member whose coordinates look like longitude/latitude are rejected (override with --assume-projected). Road properties (case-insensitive):

Barriers: LineStrings with height (m). Buildings: Polygons with height (m). Receivers: Points with optional height (m) or a Z coordinate.

What is implemented, and how it was checked

Coefficient tables (all shipped, none are placeholders)

Note that the 2021 Table F-1 differs from the 2015 one (e.g. category 1 A_R at 1 kHz is 100.1 dB, not 97.3 dB) – older tools may still use the 2015 numbers. Category 5 has no coefficients in either directive: it raises an error unless you supply coefficients with load_coefficients_csv() (format: category,coefficient,63,…,8000 with rows AR, BR, AP, BP; values are range-checked). National road surfaces can be loaded with load_surface_csv() (name,category,63,…,8000,beta[,v_min,v_max]).

Emission formulas (Annex II §2.2)

(2.2.1) line source, (2.2.2)–(2.2.3) total power and the 20 km/h floor, (2.2.4)–(2.2.5) rolling noise, (2.2.6)–(2.2.9) studded tyres, (2.2.10) temperature, (2.2.11)–(2.2.12) propulsion, (2.2.13)–(2.2.16) gradient, (2.2.17)–(2.2.18) acceleration near junctions, (2.2.19)–(2.2.20) road surface (with a warning outside the validity speed range).

Propagation

• A_div = 20 lg d + 11 (2.5.12); A_atm = α d / 1000 (2.5.13).
• α from the ISO 9613-1 pure-tone formulae at exact band mid-frequencies. The implementation reproduces ISO 9613-2 Table 2 (e.g. 20 °C / 70 %: 0.1, 0.3, 1.1, 2.8, 5.0, 9.0, 22.9, 76.6 dB/km) at all six tabulated temperature/humidity pairs to the printed precision (unit-tested).
• CNOSSOS-EU ground effect (2.5.14)–(2.5.20) for homogeneous and favourable conditions with G_s = 0 for the road platform; long-term level (2.5.9) with an occurrence p per period.
• CNOSSOS-EU diffraction (2.5.21) with C″ = 1 (single edge), Δ_ground(S,O) and Δ_ground(O,R) as amended in 2021, curved-ray path differences in favourable conditions (2.5.24)–(2.5.27), and the 2021 diffraction criterion (δ > −λ/20 and the Rayleigh criterion δ > λ/4 − δ*).
• ISO 9613-2: general ground method (Table 3), alternative A-weighted method (eq. 10), D_z with K_met (eq. 14, 18) capped at 20 dB, A_bar = D_z − A_gr ≥ 0, optional C_met.
• Maekawa barrier attenuation in the Kurze–Anderson closed form.

The 2015/2021 Official Journal prints the ground-effect and diffraction equations as images. They were implemented from the published formula structure (term names, bounds, special cases quoted in the text) together with the standard CNOSSOS-EU formulation, and are unit-tested against independent hand calculations of those equations – but they could not be compared character-by-character with the OJ text.

CRTN (UK, 1988)

Basic noise level (18 h and 1 h), speed and % heavy correction, gradient (+0.3 dB/%), road surface, distance, ground cover, angle of view and +2.5 dB facade. Not implemented: the barrier chart, the gradient speed reduction, the low-flow correction and conversions from L_A10 to L_den. CRTN predicts L_A10,18h, not an END indicator; use it as a cross-check only. The Nordic prediction methods are not included.

Validation in the test suite

ISO 9613-2 Table 2 absorption values; hand calculations of every emission correction and of the ground and diffraction equations; energetic summation (60 ⊕ 60 = 63.01 dB, ten equal sources +10 dB); L_den (equal levels give L + 6.39 dB, Member-State period variants); flow doubling +3.01 dB; a long, finely segmented road over hard ground matches the analytic line-source integral L = L_W′ − 10 lg(4d) + 3 within 0.05 dB; symmetry and monotonic decay with distance; higher barriers shield more; CRTN reference point (75 km/h, 0 % heavy gives a 0 dB speed correction; 20 000 veh/18 h gives a basic noise level of 72.1 dB(A)). Coverage is above 95 %.

Differences to full CNOSSOS-EU

roadnoise deliberately simplifies the propagation part. Results can differ significantly from a conformant CNOSSOS-EU implementation, in particular in built-up areas:

• Flat ground only. No digital terrain model, no mean ground plane per path, no equivalent heights, no embankments or cuttings; ground factor G is uniform.
• No path finding. One straight vertical propagation plane per source–receiver pair; no "rubber-band" convex hull and no multiple diffraction: only the single most effective edge (largest path difference) is used, so thick buildings and double barriers are under-estimated.
• No lateral (vertical-edge) diffraction and no reflections from facades or barriers (image sources). The optional +3 dB facade term is a flat correction.
• Barriers and building walls are thin edges at their top height; absorption of barrier surfaces is ignored.
• The favourable-condition occurrences p (defaults day 0.5, evening 0.75, night 1.0) are screening assumptions, not directional meteorological statistics.
• The AADT → hourly-flow split (default 70 % day / 20 % evening / 10 % night) and the heavy-vehicle split (default all category 3) are placeholder assumptions; provide q_day/q_evening/q_night, medium_share and counts where available.
• No statistical urban approach, no population/dwelling assignment rules of Annex II §2.8 beyond the facade receiver helper and the per-band exposure table.
• Sound power is computed once per road and per END period; there is no hourly traffic model.

Citation

If you use roadnoise, please cite it via CITATION.cff. Key references:

• Directive 2002/49/EC of the European Parliament and of the Council relating to the assessment and management of environmental noise (OJ L 189, 18.7.2002).
• Commission Directive (EU) 2015/996 establishing common noise assessment methods (OJ L 168, 1.7.2015).
• Commission Delegated Directive (EU) 2021/1226 amending, for the purposes of adapting to scientific and technical progress, Annex II to Directive 2002/49/EC as regards common noise assessment methods (OJ L 269, 28.7.2021).
• Kephalopoulos S., Paviotti M., Anfosso-Lédée F. (2012). Common noise assessment methods in Europe (CNOSSOS-EU). JRC Reference Report EUR 25379 EN.
• ISO 9613-1:1993 Acoustics – Attenuation of sound during propagation outdoors – Part 1: Calculation of the absorption of sound by the atmosphere.
• ISO 9613-2:1996 Part 2: General method of calculation.
• Maekawa Z. (1968). Noise reduction by screens. Applied Acoustics 1(3), 157–173; Kurze U. J., Anderson G. S. (1971). Sound attenuation by barriers. Applied Acoustics 4(1), 35–53.
• Department of Transport and Welsh Office (1988). Calculation of Road Traffic Noise. HMSO, London.

License

MIT © Yusuf Eminoğlu