Open-Meteo Weather Scraper

Weather Scraper: current conditions and a 16-day forecast, by city name or coordinates

Give it a list of cities and get the weather back as rows. Temperature, what it feels like, humidity, precipitation, wind and a plain-English description instead of a raw numeric code. Ask for a forecast instead and you get one row per day, up to 16 days out, or give it a start and end date and get every day in between, as far back as 1940. It runs on Open-Meteo, which needs no API key, so it needs nothing from you either.

A city name resolves to whichever place Open-Meteo ranks first. That is almost always the one you meant, and for the Springfields and Cambridges of the world it is a coin toss, so send coordinates when the answer has to be exact.

🔍 What Weather Scraper does

Three modes, and they return different things.

Places are looked up by name and resolved to Open-Meteo's top match. If you already have coordinates, leave places empty and fill in latitude and longitude instead. That path takes one point per run.

Set startDate and endDate on a forecast run to pick the days yourself, from 1940-01-01 to 15 days after today. Days before today come from Open-Meteo's historical record, the rest from the forecast.

Units are Open-Meteo's own: Celsius, kilometres per hour, millimetres. There is no setting to change them.

📥 What you give it

{
  "mode": "forecast",
  "places": ["Paris", "Madrid"],
  "forecastDays": 5,
  "maxItems": 100
}

In forecast mode a row is one day, so maxItems has to cover places times days. Five cities at 7 days is 35 rows; leave maxItems at 10 and the last cities get cut off part way down your list. A date window counts the same way, and a window longer than maxItems is cut at its far end.

{
  "mode": "forecast",
  "places": ["Paris"],
  "startDate": "2025-07-01",
  "endDate": "2025-07-31"
}

If you go the coordinates route, type real numbers into both boxes. Sending two empty strings with no places is not the same as leaving them out, and it reads as 0, 0, which is a spot in the Atlantic that will happily report its weather.

📤 What you get back

A real current row from a recent run:

{
  "ok": true,
  "place": "Berlin",
  "latitude": 52.52437,
  "longitude": 13.41053,
  "time": "2026-09-14T07:45",
  "temperature": 14.9,
  "apparentTemperature": 14,
  "humidity": 88,
  "precipitation": 0,
  "windSpeed": 12.3,
  "weatherCode": 3,
  "weather": "Overcast",
  "timezone": "Europe/Berlin"
}

place is what you typed. In geocode mode, name is what Open-Meteo matched it to, and comparing the two is the quickest way to catch a wrong city.

weatherCode is the WMO number and weather is the short text for it, like Overcast or Slight rain. A code with no text mapping reads as Unknown rather than going blank.

🧾 Reading the output

Two kinds of row land in your dataset.

When some places resolve and others do not, the run keeps going. The ones that worked come back as rows and are charged; the ones that did not are named in the run log, not in the dataset. Compare your row count against the number of places you sent.

Coordinates outside the legal range can come back as a row with the weather fields empty rather than as an error, so keep latitude within 90 and longitude within 180.

A forecast day that comes back completely empty, often the 16th, is left out and not charged, so a 16-day forecast can return 15 rows for a place. In a date window, each row is also checked against the window by its own date.

▶️ How to run it

1. Open Weather Scraper and click Try for free. 2. Pick a Mode. Current weather is the default. 3. Type your cities into Places, or leave it empty and fill in Latitude and Longitude. 4. Set Forecast days if you chose the forecast, then click Start. 5. Download the dataset as JSON, CSV or Excel, or read it from the Apify API.

💰 How much does it cost?

$0.20 per 1,000 rows. Flat on every Apify plan, no volume tiers.

You pay per row delivered. In forecast mode a row is one day, so a 7-day forecast for 10 cities is 70 rows, and the whole of July for 3 cities is 93. In current and geocode mode a row is one place. Diagnostic rows and empty forecast days are not charged.

💡 What people use it for

• A morning run over a list of depots or sites, with the rows going straight into a sheet.
• Checking the 16-day outlook for a set of event locations before committing to a date.
• Resolving a column of city names to coordinates and timezones in geocode mode, then using those

coordinates everywhere else.

• Building a rain history by scheduling current hourly and keeping every row.

🚧 What it does not do

• No hourly data. Current conditions, or daily forecast rows. Nothing in between.
• Past days are estimates, not station readings. Open-Meteo's historical record rebuilds each day

from observations and weather models on a grid, so a row describes the area around your coordinates. Nothing before 1940-01-01.

• No unit choice. Celsius, kilometres per hour and millimetres, always.
• One coordinate pair per run. A list of exact points means a run each, or place names instead.
• Top match only when resolving a name, and no list of the alternatives it rejected.
• A forecast is a forecast. Day 14 is a model output, not a promise.
• No alerts, warnings, air quality or marine products, even though Open-Meteo publishes some of

those separately.

🧭 Which location scraper do you need?

❓ Questions people ask

Do I need a weather API key? No. Nothing to apply for and no quota to watch.

How far ahead can it forecast? Sixteen days, set with forecastDays.

Can I get past weather? Yes: forecast mode with startDate and endDate, back to 1940-01-01, one row per day per place.

I asked for Springfield and got the wrong one. Add the country, like Springfield, US, or send the coordinates. The name lookup takes the top match and does not ask.

Can I get hourly readings? Not from this one. Run current on a schedule instead and keep the rows.

Can I schedule it? Yes, like any Apify actor. A saved list of places makes a daily run trivial.

Where does the data come from? Open-Meteo, for both the weather and the place lookup. Read their terms for attribution and fair use before building on it at volume.

🆘 If something breaks

Open the Issues tab on the actor page. Send the mode, the places or coordinates you used, and the run ID. The errorCode on the diagnostic row usually names the problem on its own.