From the historical weather API of Open-Meteo. It combines reanalysis models: ERA5 (about 25 km grid), ERA5-Land (about 11 km) and, from 2017, ECMWF IFS (about 9 km). A reanalysis is a weather model fed with observations from weather stations, satellites, aircraft, buoys and radar. The result is a complete hourly record on a grid. These are not the measurements of one weather station.

Grid data describes the average conditions in a grid cell of roughly 9 to 25 km. It is good for flat land away from the coast. It is less reliable when:

• the data point is far from your site, for example at sea, where the tool falls back to the nearest land point;
• your site is on the coast, where sea and land conditions mix;
• the terrain is hilly or mountainous, where wind and temperature change over short distances.

Under step 1, the tool shows the data point it used and its distance to your site. Use the points above to judge whether the data represents your site. The tool does not correct the data.

For each day the tool looks at the hours between 06:00 and 18:00 local time. It takes the average of every block of 4 consecutive hours and keeps the worst block for each variable. If that worst block exceeds any of your limits, the whole day counts as unworkable.

• Rain: millimetres per hour, averaged over the worst 4 hours. A limit of 1.9 mm/h stops work when it rains more than 7.6 mm in 4 hours.
• Wind: mean wind speed in metres per second at 10 m height.
• Gusts: maximum gust speed in metres per second at 10 m height.
• Temperature: degrees Celsius at 2 m height, averaged over the warmest or coldest 4 hours of the working day.

Wind at a crane hook or on a roof is higher than at 10 m. Set your wind limits with that in mind.

The table gives starting values for common outdoor activities, in the terms of this tool: the worst 4 hours between 06:00 and 18:00. NA means no limit.

¹ Coating datasheets usually also limit the relative humidity and require the surface to be at least 3 °C above the dew point. This tool does not include humidity; check the coating datasheet.

More years give more examples of each month, but the climate is changing. Older years describe a climate that is further from today. Five years is a balance between enough examples and recent conditions. Choose more years when you want a wider view of bad years.

The local time of your site, determined from the coordinates. The working window of 06:00 to 18:00 is therefore local daylight time, wherever the site is.

Missing hours are rare. When a value is missing, the tool fills it in:

• gaps of up to 6 hours are filled by linear interpolation between the neighbouring hours;
• longer gaps are filled with the average of the same hour on the same dates (plus or minus 3 days) in the other years.

The filled column in the CSV shows which values were filled. A rain value of 0.0 is a real measurement of no rain, not a gap. If a variable is not available at all for your site, the tool says so and switches off that limit.

The averages include the weather of the years you selected, nothing more. A storm week or frost period worse than anything in that record is not in the data. Treat such events as separate risks in your risk register.

The tool is free and uses a free weather data service. To keep the use of that service modest, each browser can fetch weather data 5 times in 7 days. That is enough for 5 new sites per week. Changing the limits, reading the chart and downloading the CSV do not count.

Your browser sends the coordinates directly to Open-Meteo to fetch the data. RvA Planning does not receive or store them.

The tool is free to use, as it is. It is built and tested for my own work. If you find an error or have a suggestion, use the contact form below and mention your coordinates. I read every message, but I do not promise a response or a fix within a set time. If you need guaranteed support, a commercial tool is a better choice.