OpenWeatherMap APIs Overview

Learn more about the OpenWeatherMap APIs, their coverage, their weather prediction model, and their performance metrics.

Types of APIs

OpenWeatherMap provides a number of weather APIs that provide up-to-date and highly accurate weather forecasts and historical weather data. It also provides APIs to retrieve air pollution data and utility APIs that let users retrieve unique identifiers and geographical coordinates for various locations. This information is helpful because most of the APIs we cover in this course require the geographical coordinates of a location rather than its name as a parameter.

The following illustration explains the flow of the OpenWeatherMap APIs we follow in this course.

Salient features of the OpenWeatherMap APIs

OpenWeather ensures the delivery of accurate data to its users by using its tools to observe accuracy and quality metrics. Let's look at OpenWeatherMap's coverage, its weather prediction models, and the different metrics it employs to evaluate the accuracy of the generated data.

Coverage

A total of 371 cities are used to evaluate weather data. These include national capitals along with many other major cities across the world.

Prediction model

OpenWeatherMap APIs provide weather data using its Numerical Weather Prediction (NWP) model. This model gathers data from the data sources listed below:

  • Global NWP models: National Oceanic and Atmospheric Administration Global Forecast System (NOAA GFS) 0.25 and 0.5 grid sizes, National Oceanic and Atmospheric Administration Climate Forecast System (NOAA CFS), (European Centre for Medium-Range Weather Forecasts ECMWF Re-Analysis (ECMWF ERA)

  • Weather stations: Meteorological Terminal Air Report (METAR) stations, users' stations, companies' stations

  • Weather radar data

  • Satellite data

OpenWeather uses its array of algorithms to process the data gathered from these sources. This processing is done in real-time to improve the accuracy and quality of metrics. Moreover, it provides the latest current weather data and weather forecasts.

The NWP model provides high accuracy readings between 90% and 100%, with an inaccuracy of only about 1%. Its data sources include 82,000 weather stations across the globe, national meteorological agencies (NOAA, Environment Canada, Met Office, and so on), radars, and weather satellites.

Metrics

OpenWeather considers several reliable sources to compare forecasts, including multiple weather stations run by meteorological agencies. Moreover, weather radar sources are used for comparing precipitation forecasts.

The quality evaluation metrics can be divided into the following three groups:

  1. Common scores that show the accuracy of forecasts, intended for forecast users.

  2. Metrics that are used to compare various raw data sources, and algorithms that are used to choose between them after processing.

  3. Diagnostic metrics that are used to improve forecasts by limiting certain types of errors in them.

These metrics are used to evaluate the quality of weather forecasts for common and special purposes.