How the bell map is made

No one measures how loud church bells are at every address, so this site models it. Here is exactly what the model does, so you can judge it.

1. Which towers ring

Every place of worship and bell tower in OpenStreetMap, cross-checked against the federal building register (class 1272, churches and other religious buildings). Mosques, synagogues, kingdom halls, free-church halls and parish centres are kept on the map but do not ring. Every church gets a confidence: confident, probably, unsure. Chapels count as unsure: many only ring on feast days.

2. How loud a tower is

A sound power level for a full peal by size class, from the footprint area and the building tag: cathedral 128, large church 125, parish church 122, small church 117, chapel 110 dB(A). A single swinging bell (prayer bell, Angelus) is 6 dB less, a clock strike 10 dB less. These are order-of-magnitude values chosen so that a parish church gives about 71 dB(A) a hundred metres away in the open, which matches the few published measurements. They are not measured per church. Tower heights come from OpenStreetMap where tagged (the Bern Minster's 100 m, for example), otherwise from the size class.

3. When bells ring

A default schedule per tradition: Reformed churches ring morning, eleven o'clock and evening prayer bells on weekdays, ring in Sunday on Saturday evening and ring for the Sunday service; Catholic churches ring the Angelus three times a day, a Saturday vigil and two Sunday masses. Both strike the quarters and the hours; by default also at night, since Swiss courts have allowed that and most churches do it, though a good number have switched night strikes off. Real parishes differ a lot, which is why you noticed one village ringing at seven while the next one didn't. Every church can carry a correction, and you will be able to submit one.

4. How sound travels

ISO 9613-2, simplified. Spreading from a point source at the bell chamber (about three quarters of the way up the tower), air absorption of 2.5 dB per kilometre, and diffraction over the highest obstacle on the straight line to a receiver 4 m above ground, as in official noise maps. Obstacles are building footprints from OpenStreetMap (2.8 million) with heights from OpenStreetMap tags, from the floor count in the federal building register (1.7 million matched), or a default by building type, on 30 m Copernicus terrain. Reflections, wind and temperature are not modelled, and the shielding is capped at 20 dB.

5. What the numbers mean

Day (06–22) and night (22–06) are energy averages over a week, like the federal road and rail maps next to them, so you can compare: a Reformed parish church gives about 48 dB(A) by day and 38 at night a hundred metres away in the open. The loudest ring is the peak during a full peal, around 70 dB(A) at a hundred metres: that is the number that wakes you on Sunday morning. Everything below 20 dB(A) is left blank. The grid is 10 m.

What would make it better

Measurements. A phone recording with a position and a time, a few hundred of them, would calibrate the source levels. Per-parish ring schedules would fix the biggest error. Both are planned as contributions with an account. swisstopo's 3D buildings would improve the heights in cities.