Weather forecast

How the forecast is made

  1. Observation satellites and monitoring stations around the world collect weather data.
  2. The data is sent to weather centers.
  3. Meteorologists add the data to predictive models and have it processed by computers.

Each predictive model takes into account the specifics of atmospheric circulation, water evaporation and condensation, the Earth's heat, and the influence of many other factors. After the data is processed, the meteorologists get a forecast of temperature, precipitation, wind speed and direction, and other weather events.

None of the predictive models is absolutely accurate, and each of them has its own advantages and disadvantages. Therefore, Yandex.Weather uses the Meteum technology to make its forecasts. It combines predictive modeling with machine learning.

What is Meteum

Meteum is Yandex's proprietary weather forecasting technology. Powered by machine learning, Meteum compares data from different predictive models and then makes its own forecast.

Meteum produces the forecast in two stages.

Stage 1. Making its own forecast
  1. Yandex.Weather receives data from weather stations and satellites. To make the forecast more accurate, the service uses detailed data on the type of underlying surface from Yandex.Maps.
  2. Yandex.Weather makes its own forecast using a predictive model. Unlike most meteorological companies, the Yandex.Weather model makes an accurate forecast not only for regions and cities, but also for particular city districts, metro stations, suburban settlements, and other locations.
Stage 2. Producing the final version of the forecast
  1. The service gets information about forecasts made using other predictive models including US, Canadian, Japanese, and European ones.
  2. Yandex.Weather compares the forecasts and actual weather conditions using machine learning algorithms. The algorithms identify the model with the most accurate forecast for particular weather conditions and region. If no model provides a highly accurate forecast, Meteum combines data from the most accurate forecasts to make its own one.

Yandex.Weather uses Meteum to forecast weather for regions and cities around the world.

The forecast takes into account the user's exact location.



How often weather data is updated

Yandex.Weather provides current weather data and data for upcoming hours and days.

Weather data is updated at least once every three hours. The frequency of weather data updates for a particular city depends on when the data is updated at the nearest weather station.

In the regions for which a precipitation map is made, precipitation data is updated once every 10 minutes.