Key takeaways
- An EPW file is text: 8 header lines followed by hourly weather records, 8,760 of them for a full non-leap year.
- Each hourly record has 35 comma-separated fields covering temperature, humidity, pressure, solar radiation, illuminance, wind, sky cover and precipitation, all in SI units.
- Downloadable EPW files, such as those on Climate.OneBuilding.Org, are usually typical years assembled from months of different real years, not the weather of one calendar year.
- Climate.OneBuilding.Org, run by two of the format's creators, publishes free EPW files for more than 17,000 locations.
- An EPW describes one weather station or grid cell, so check its location, elevation and period of record before relying on it for your site.
What is an EPW file?
EPW stands for EnergyPlus Weather. Drury Crawley, Jon Hand and Linda Lawrie described a new, generalised weather format for two simulation programs, EnergyPlus and ESP-r, in 1999; it later became known as EPW. It is based on the data in the older TMY2 format, rearranged into simple comma-separated text in SI units so that people and spreadsheets can read it as easily as software.
Because the format is shared, one file can feed an EnergyPlus energy model, a Grasshopper script in Ladybug Tools or a web tool such as the CBE Clima Tool. By 2015 two of its authors, Drury Crawley and Linda Lawrie, counted more than 20 building simulation programs that read EPW.
How is an EPW file structured?
The file opens with eight header lines, each starting with a keyword. Then come the data records, one line per time step. Most weather data is hourly, so a full year has 8,760 data lines (365 days × 24 hours). The format also allows several records per hour.
- LOCATION: city, region, country, data source, WMO station number, latitude, longitude, time zone (hours from GMT) and elevation in metres.
- DESIGN CONDITIONS: heating and cooling design conditions matched to the station's WMO number, usually taken from the ASHRAE Handbook of Fundamentals.
- TYPICAL/EXTREME PERIODS: typical and extreme weather periods found in the data by the EnergyPlus Weather Converter.
- GROUND TEMPERATURES: monthly undisturbed ground temperatures at several depths, calculated from the air temperature record.
- HOLIDAYS/DAYLIGHT SAVING: whether 29 February is observed, daylight saving dates and any holidays.
- COMMENTS 1 and COMMENTS 2: free text that may describe the data source and how the file was made.
- DATA PERIODS: the number of records per hour and the start and end dates of the data.
Which weather variables does each record contain?
Each data line has 35 fields. The first five give the date and time (year, month, day, hour and minute), the sixth holds data-source and uncertainty flags, and the rest are weather values:
- Temperature and humidity: dry-bulb and dew-point temperature (°C), relative humidity (%) and station pressure (Pa).
- Radiation: extraterrestrial, global horizontal, direct normal and diffuse horizontal solar radiation, plus infrared radiation from the sky (all in Wh/m²).
- Daylight: global, direct normal and diffuse horizontal illuminance (lux) and zenith luminance (cd/m²).
- Wind: direction in degrees, with north at 0 and east at 90, and speed in m/s.
- Sky and visibility: total and opaque sky cover in tenths, visibility, ceiling height and present-weather codes.
- Water and ground: precipitable water, aerosol optical depth, snow depth, days since last snowfall, albedo, and liquid precipitation depth (mm) with its accumulation period.
Where do EPW files come from?
An EPW is only a container; the source of the data inside matters more. The file name and the source field in the LOCATION line usually tell you where it came from.
- Typical years from weather stations. Climate.OneBuilding.Org, run by Linda Lawrie and Drury Crawley, two of the format's creators, publishes free TMYx files for 17,315 locations, built from NOAA's Integrated Surface Database with TMY and ISO 15927-4 methods. In the 2026 release their solar radiation comes from the ERA5 reanalysis. The site also hosts national sets such as TMY3 for the United States and CWEC for Canada.
- Typical years from satellite and reanalysis data. PVGIS, from the European Commission's Joint Research Centre, generates a typical year in EPW format for a chosen location, using satellite or reanalysis solar radiation and ERA reanalysis data for the other variables.
- Experimental national products. NOAA's National Centers for Environmental Information publishes an experimental typical-year product for 1,972 stations, based on 1998–2022 data, in CSV and EPW formats.
- Actual and future years. The same format can hold one real calendar year, for example to test a design against a recorded heatwave, or a typical year adjusted to a future climate.
How to open, check and use an EPW file
Any text editor opens an EPW, and the eight header lines are readable as they are. A spreadsheet can import the file as comma-separated text; skip the first eight lines to get one row per hour and one column per field. The EnergyPlus Weather Converter can also turn an EPW into a CSV version and produce a statistics report.
Two conventions catch people out. Hours run from 1 to 24, with hour 1 covering 00:01 to 01:00. And radiation values are totals over the period before the timestamp, while temperatures are values at the timestamp. Missing data is flagged with codes such as 99.9 for temperature and 9999 for radiation, and not every field is used by every program: EnergyPlus ignores the global horizontal radiation field and works from direct normal and diffuse horizontal radiation.
Every chart and simulation inherits the file's assumptions, so choosing it is a design decision. Before using one, check:
- Location and elevation: is the station close to the site and at a similar altitude, and does it share the site's exposure, for example coastal or inland?
- Source and period: a typical year built from 2011–2025 data can describe a noticeably different climate from one built from 1976–2005 data.
- The Year column: in a typical year it changes from month to month, because each month comes from a different source year.
- Purpose: a typical year leaves out extremes by design, so it is not the right file on its own for overheating checks or peak loads.
How C4B uses EPW files
C4B does not ask you to find a weather file. For every search it loads the nearest Climate.OneBuilding.Org weather file for the location you pick, usually a TMYx file, on every plan including Free. In projects on paid plans it also compiles a typical year from reanalysis data at the site's exact coordinates, so you can compare the station with the site.
Downloading EPW files from C4B, whether a custom site file or a future-climate file, is available on the Technical and Enterprise plans. The files use the standard format, so they work in EnergyPlus-based tools.
Frequently asked questions
Is an EPW file the same as a TMY?
No. EPW is a file format; a TMY (typical meteorological year) is a way of assembling a representative year of weather. Repositories such as Climate.OneBuilding.Org publish typical years as EPW files, but the same format can hold a real calendar year or a year adjusted to a future climate.
How many rows does an EPW file have?
A full year of hourly data has 8 header lines followed by 8,760 hourly records, so 8,768 lines in total for a non-leap year. Files with sub-hourly data or partial years have a different count, given in the DATA PERIODS header line.
Can I open an EPW file in Excel?
Yes. EPW is comma-separated text, so a spreadsheet can import it. Skip the first eight header lines to get one row per hour with 35 columns, and work on a copy, because spreadsheets can reformat numbers and dates when saving.
Where can I download EPW files for free?
Climate.OneBuilding.Org publishes free typical-year EPW files for more than 17,000 locations worldwide, and PVGIS generates free typical-year EPW files for a chosen location from satellite and reanalysis data. Check the source, the period of record and the distance to your site before using one.
Why does my EPW file show different years in different months?
Because it is a typical year. Each calendar month was picked from the year whose weather for that month best matched the long-term record, so January might come from 2014 and February from 2019. The EnergyPlus documentation notes that the Year field is not really used in EnergyPlus.
See it for your site
Site-specific weather data
A typical-year weather file compiled from reanalysis at the project's coordinates, alongside the nearest station.
Future climate scenarios
Future weather files for any year up to 2100 under five SSP scenarios, overlaid on the project's analyses.
Site climate analysis
A full climate study for any location in the world, run in the browser from a single address or map pin.
See your site's climate before design begins
Pin any location and get climate, comfort and preliminary energy insight your whole team can read. Start with a 7-day free trial, or book a 30-minute demo.
References
- EnergyPlus Auxiliary Programs: EnergyPlus Weather File (EPW) Data Dictionary
- EnergyPlus Auxiliary Programs: Source Weather Data Formats
- Crawley, Hand and Lawrie (1999), Improving the weather information available to simulation programs
- Climate.OneBuilding.Org: repository of building simulation climate data
- European Commission JRC: PVGIS typical meteorological year generator
- NOAA NCEI: Typical Meteorological Year (experimental)
- EnergyPlus: Weather Data for Simulation
- European Commission JRC: Photovoltaic Geographical Information System (PVGIS)
- Ladybug Tools primer: Import EPW component
- CBE Clima Tool documentation: overview
Related guides
Site climate analysis for architecture
A step-by-step method for the climate layer of an architectural site analysis: where to get the data, what to read, what to put on the sheet and how each finding becomes a design move.
Typical meteorological year (TMY)
How typical meteorological years are selected, how TMY, TMYx and ISO 15927-4 reference years differ, and when to use actual or extreme years instead.
Reanalysis data and ERA5
How reanalysis datasets such as ERA5, ERA5-Land, CERRA and NORA3 work, what their resolution means for a building site, and how they compare with station data.
Future weather files and morphing
How morphing combines an observed weather year with climate model projections, how the methods differ and what future weather files can and cannot tell you.
GHI, DNI and DHI
What GHI, DNI and DHI mean, how they relate, their units and data sources, and how they feed façade, shading and PV design.
How to read a wind rose
Petals, frequency rings, speed bands and calms explained, with the checks and design uses that matter on a building project.