From weather to observed climate

Current climate and past weather situations can be described using weather-station measurements, model outputs, reanalyses, or synthetic files. These sources differ in their origin, resolution, and representativeness.

Observing the weather day after day creates the long time series needed to characterise climate, its variability, and its trends.

Weather measurements

A weather station measures air temperature, pressure, wind, humidity, dew point, precipitation, cloud cover, and visibility at a given location. Solar radiation is measured less often, although it is essential for building thermal simulations.

A station’s instruments depend on its purpose: climatology, aviation, agriculture, or road weather. National networks share some observations with the World Meteorological Organization . Stations generally have a WMO number and, in aviation, an ICAO code.

Source: NCEI/NOAA

Source: NCEI/NOAA

Data sources

• Météo-France for observations in France;
• NOAA/NCEI for numerous observations worldwide;
• Meteostat for convenient access to historical hourly or daily observations from different networks.

A station provides a local measurement. It may not represent conditions at a distant site, particularly when relief, urbanisation, or distance from the sea differs.

Explore observed temperatures

The Temperature trends and anomalies tool can be used to explore average temperatures in several French cities and their departure from the 1991–2020 climate normal. It offers two complementary levels of interpretation: an annual view through climate stripes and a daily view for comparing several years.

Forecasting the weather

A weather forecast predicts conditions over the next few hours or days. It starts from the most accurate possible description of the atmosphere: temperature, pressure, humidity, wind, soil conditions, and other variables.

This initial state is built by continuously assimilating observations from weather stations, satellites, radars, aircraft, buoys, and other measurement systems. Small initial errors can grow during a simulation, so predictability rapidly decreases as the forecast range increases.

Forecasting models then solve the equations governing the atmosphere on a grid. They produce a specific chronological evolution, such as the passage of a front or tomorrow’s temperature at a given time. Several simulations with slightly different initial conditions are often combined into an ensemble forecast to estimate uncertainty.

From global to local

Forecasting begins at the global scale because atmospheric conditions over a region depend on phenomena far beyond its boundaries. Finer-grid regional models are then driven by the global model to represent relief, coastlines, land use, and convective phenomena more accurately.

This move towards the local scale is particularly useful when forecasting conditions across a territory, city, or project site. It does not eliminate every uncertainty: some phenomena remain smaller than the model grid or must be represented through parameterisations.

Source: DRIAS

Two main families of methods are used:

• dynamic downscaling, based on a high-resolution physical model, which maintains consistency between variables but is computationally expensive;
• statistical downscaling, which relates large-scale conditions to local observations at a lower computational cost, but may introduce artefacts.

Future climate modelling also uses global models and downscaling. Its purpose is different, however: rather than forecasting the weather on a specific day, it examines statistical climate change over several decades. This approach is presented on the From scenarios to climate projections page.

Reanalyses: reconstructing past climate

A reanalysis combines an atmospheric model with numerous historical observations through data assimilation. It reconstructs the atmosphere on a regular grid over several decades, providing continuous spatial coverage, homogeneous time series, and many variables.

It is not a direct observation: this model result, constrained by measurements, may retain biases, particularly for highly local phenomena.

Source: ECMWF

ERA5

ERA5 is a global reanalysis produced by ECMWF. It provides many atmospheric variables over several decades. ERA5-Land offers a finer reconstruction of land-surface variables.

Hourly weather files

Thermal, energy, and comfort studies often use a full year of hourly data. The file may represent an average climate or an actual year.

Typical Meteorological Year — TMY

A TMY is a synthetic year representative of the climate at a location over several decades. Its months may come from different years selected to reproduce the statistics of the reference period.

TMY files suit energy studies, routine system sizing, and typical comfort assessments. Files are available from Climate.OneBuilding.org , among other sources.

Explore an EPW weather file

The Climate and outdoor comfort analysis tool can load an EPW file from Climate.OneBuilding.org or accept one by drag and drop, select a study period, and explore its main variables: temperature, humidity, solar radiation, and wind. It also provides outdoor comfort analyses, notably using the PET index.

Available variables

Limitations of typical years

Because a TMY represents average climate conditions, it does not accurately reproduce rare events. A severe heatwave may be absent even though it is decisive in a resilience study.

Actual Meteorological Year — AMY

An AMY represents an observed or reconstructed year. Several datasets may be combined when a variable is missing; solar radiation, for example, may be reconstructed using a model or reanalysis.

AMY files can reproduce a historical event, compare simulations with measurements, or study a particularly hot summer. Resimulating a building with the 2003 weather is a common resilience-analysis case.

What about extreme events?

An extreme event is not defined by a maximum value alone: duration, intensity, geographical extent, and cumulative severity also matter. Heatwaves, which are important for comfort and resilience, are covered on the Heatwaves page.