At the Met Office, observations are the cornerstone of our forecasting capability. They provide a real-time snapshot of atmospheric conditions, feeding directly into our systems to ensure our forecasts remain accurate and reliable.
In this blog, we explore the breadth and depth of the observations we use, how they are processed, and the innovations shaping the future of atmospheric data processing.
The role of observations in data assimilation
Data assimilation is the process of combining observations with computer model output to produce the best possible estimate of the current state of the atmosphere. This estimate is then used to initialise weather forecasts. Observations are an essential ingredient in this process, anchoring our models to reality and helping to correct any deviations.
Every six hours, tens of millions of observations are assimilated into our global atmospheric model, with tens of thousands more feeding into our higher-resolution regional models every hour. These observations come in two forms: direct measurements, such as temperature readings from thermometers, and indirect measurements, which work out atmospheric variables from other data, such as satellite imagery.
Before assimilation, all observations undergo rigorous processing and quality control. This ensures that only reliable data are used, and that each observation is appropriately weighted based on its accuracy and relevance. Some observations, like those from weather balloons, are considered highly valuable and serve as a benchmark for calibrating other data sources.
READ MORE: Waters surrounding UK experiencing significant marine heatwave
A diverse range of observations
We assimilate a wide variety of observation types, each offering unique strengths and contributing to a comprehensive picture of the atmosphere. These include:
-
Surface-Based Observations
Surface-based systems include weather stations, ships, buoys, aircraft, and radar. These provide a rich array of data, including near-surface temperature, pressure, wind, humidity, and visibility, as well as vertical profiles of temperature, wind, and humidity throughout the atmosphere. Their diversity and reliability make them a vital component of our observational network.
-
Satellite Observations
Satellites contribute the majority of the data assimilated into our models. They offer unparalleled global coverage, especially over oceans and remote regions where surface observations are sparse. Weather satellites operate in two main orbits:
-
Geostationary satellites orbit at approximately 36,000 km above the equator, remaining fixed over one location. They provide continuous observations of the same region, ideal for monitoring rapidly evolving weather systems.
-
Polar-orbiting satellites travel from pole to pole at altitudes between 200 and 1,000 km, completing an orbit roughly every 90 minutes. Multiple orbits are needed to achieve full global coverage.
Satellite instruments typically measure electromagnetic radiation. This radiation has either been emitted by the Earth and atmosphere or sent by the satellite and been reflected back. These radiance measurements are crucial for constraining atmospheric temperature and humidity.
We also use radiance information from satellites to provide other useful atmospheric information. For example, by analysing satellite images that show cloud patterns, we can estimate important atmospheric conditions like wind speed, cloud type, and humidity. These are called 'derived products' because they’re calculated from the satellite’s raw data, and they help improve the accuracy of weather forecasts.
Another notable observation technique is radio occultation, which uses signals from Global Navigation Satellite Systems (GNSS) to deduce atmospheric temperature and humidity by measuring how signals bend as they pass through the atmosphere.
READ MORE: How will AI weather forecasts make maximum impacts for users?
Evolving with innovation
Despite the vast quantity and variety of observations we already use, we are continually evolving to improve their assimilation. This involves two key strategies:
-
Enhancing the use of existing data through ever-evolving quality control, better uncertainty estimation, and more sophisticated observations. Emerging techniques, including machine learning, are being explored to extend the range of conditions under which data can be assimilated, such as visible-spectrum radiances and cloud-affected observations.
-
Incorporating new datasets to expand our observational coverage. The upcoming third generation of Meteosat satellites promises to enhance our ability to monitor severe weather events in near real-time. Since 2023, the Met Office has been using data over Europe from thousands of aircraft each day to improve weather forecasts. This data, known as Mode-S EHS, includes information such as wind speed, temperature, and aircraft position, which is collected during flights. Originally used for air traffic control, this data has proven valuable for meteorology, especially in areas where traditional weather observations are limited. The integration of aircraft data into forecasting models has enhanced the accuracy of short-term weather predictions, particularly for wind and temperature at cruising altitudes. This is especially beneficial for aviation safety and efficiency, but it also improves general weather forecasts for the public. Recently the coverage of Mode-S data has been expanded to include observations from around the globe. This has significantly improved forecast accuracy and the data are becoming a permanent part of the Met Office’s forecasting toolkit.
A new satellite is scheduled to launch today in the latest phase of a multi-year project to transform weather forecasting and climate monitoring.
— Met Office (@metoffice) July 1, 2025
So what exactly is happening and why does it matter? Read this short thread, or check the story below 👇#MTGS1
🧵
These advancements bring new challenges, particularly in terms of data volume and processing requirements. Our next-generation supercomputing infrastructure will be critical in enabling us to fully exploit these opportunities.
READ MORE: Better forecasts ahead as Met Office transitions to a supercomputer in Azure cloud
Observations are the lifeblood of weather forecasting. They ensure our models remain grounded in reality and enable us to deliver accurate, timely forecasts that help protect lives and property. As we continue to innovate and expand our observational capabilities, we remain committed to maintaining the highest standards in data assimilation.
Keep up to date with weather warnings, and you can find the latest forecast on our website, on YouTube, by following us on X and Facebook, as well as on our mobile app which is available for iPhone from the App store and for Android from the Google Play store.