Frequently asked questions (FAQ):

• Why is the forecast for my location wrong?

  • Reasons for wrong forecasts

    There are several factors that can lead to inaccurate forecasts for a specific location. Some of those factors are inherent in weather forecasting:

    • Highly localized weather conditions such as thunderstorms with hail and showers. For example, showers may be predicted for a location but occur elsewhere in the vicinity. Another example includes frost that occurs very locally in places where cold air can accumulate.
    • Weather conditions that depend on some subtle physical equilibrium. Examples include whether hail stones develop in a cloud, the forming of fog, or the onset of drizzle.
    • Small-scale variations in the mountains or along coastlines that have a strong influence on weather cannot be adequately captured in weather models.

    Other factors are related to the data we use to locally adjust forecasts or to the data that “feed” the numerical weather models in the first place:

    • Lack of nearby observations that would allow us to make local adjustments to a forecast
    • Errors in nearby observations that lead to wrong local adjustments of a forecast
    • In some regions of the world only very few weather stations and other observations are available. This lack of data leads to higher uncertainty in the weather forecast for those regions.

    Finally, imprecise weather calculations may very occasionally be found even in the best numerical weather models, such as the ones operated by meteoblue. They need to be detected and improved by careful analysis with the help of repeated and detailed observations.

  • Tools and features of forecast error prevention

    On some days, forecasts may be quite uncertain but still better than no forecast at all – as long as the user is aware of the uncertainty. meteoblue therefore always indicates predictability on the 7- and 14-day forecasts, provides uncertainty ranges (e.g. in the 14-day forecast), diagrams to quickly assess where weather models agree or diverge, and also a short-term verification of the forecast. These provide valuable information for users of the meteoblue forecasts, and it allows us to monitor our forecast system for errors.

    • Predictability: This index indicates how certain the forecast for the day in question is. You can find it on our 7-day forecasts as target icons and on our 14-day forecasts as colored bars for each day.
    • MultiModel: This diagram shows how weather models agree or differ at a location. A forecast is more certain where the models predict similar weather conditions and where the values are closer to each other. Where the graphs show outliers or drift apart, the forecast is more uncertain. Note that the weather models are colour-coded to show which forecast comes from which model.
    • Short-term verification: We continuously verify our forecasts with measurement data gathered worldwide. For short-term verification, we use data from the weather station closest to the selected location (the distance is displayed at the top) and compare our temperature and wind forecasts (thin black lines) with measurement data (fat black lines and rhomb symbols). Forecast error statistics are given underneath each diagram. Where we detect unexpected differences between our forecast and a nearby station, we check the station for possible errors.

• What is the meaning of the various symbols used on the web site?

  • Pictograms: Weather conditions are graphically summarised by pictograms. All Pictograms and their descriptions are listed here.
  • rainSPOT displays precipitation 15 km around the selected location as a pattern of blue or green boxes on a target symbol background. The target symbol is centred on the selected location. Light blue boxes represent light rain, while dark blue boxes indicate heavy rain. rainSPOT appears on the daily and (3-)hourly forecasts and on the current weather page.
  • Predictability bar: The bar in the daily overview below the sunshine hours indicates predictability, a measure of how likely the forecast is to be accurate.

• How does location search work?

  meteoblue provides detailed weather forecast and historical data for every point on Earth. Locations can be found through our location search bar at the top of the meteoblue website. You can type in names, coordinates, or postal codes to find a location. Locations can also be selected by clicking directly on the map that opens when clicking on the location search bar. If you click on the locator symbol to the right, location search detects your current location, provided you allow the meteoblue website to do so. meteoblue uses a chosen location, until you pick a different location. You can find out more on our help page for location search.

• How is the weather data calculated?

  meteoblue operates a large number of numerical weather models and in addition uses data from models operated by weather services around the world. meteoblue analyses available model data automatically and creates a so-called consensus forecast. This meteoblue multimodel approach is applied to all locations and to most of the weather variables. We also use observations and current measurements, if available, to adjust our forecast for those variables.

  Forecasts of temperature, wind and dewpoint are further improved with artificial intelligence (AI) in the meteoblue Learning MultiModel (mLM). The AI component is trained with available historical and current measurement data. When a user requests a weather forecast, mLM analyses the availability of weather models for the desired weather variable and the desired location, for example Basel (Switzerland). mLM uses the different temperature outputs of the weather models available and weights those data based on what it “learned” from historical and predicted weather data. In addition, current measurements and observations are taken into account if available. If temperature data for Zurich are needed, mLM again checks weather model availability and again calculates the weighting of the variables from the different weather models. However, the weighting here can be (and probably is) completely different than in Basel. And if the requests the same data later, the weighting can be completely different again.

  We provide extensive information about the meteoblue Learning MultiModel and the verification of our forecasts. An overview of the available weather models and their spatial resolution can be found here.

• What weather data can I obtain from meteoblue, and how?

  You can explore our offer in six dimensions:

  • Global coverage – Weather data for any place on Earth
  • Top precision – Simulations with maximum proven accuracy
  • Unique time range – Gapless hourly data repository from 1984 onwards
  • Easy access – High speed delivery of data, maps, images etc.
  • Multiple data sources – Easy choice of best data source
  • Sector specific – Customized solutions for different sectors such as agriculture, energy, logistics etc.

For detailed information on meteoblue products and services, see our help pages and our full FAQ there.