Our hosting provider is having an outage, so some energy-data.io services may not work normally right now, including the API and AI chat. Your data and licence allowances are not affected. We're working on it and will update this notice once everything is back. Questions: info@energy-data.io

Start 7-day trial

Weather data sources for degree days and M&V.

Degree days and weather normalization are only as trustworthy as the weather behind them. Here is where Energy-Data.io gets historical observations, reanalysis fills, and typical-year / long-term reference data — and how each is used.

Two jobs, two kinds of weather

Energy analysis needs both what actually happened (Actual Meteorological Year / AMY) and what is typical (TMY or a long-term average). We use observed and reanalysis weather for actual periods, and separate TMY / long-term inputs for normalization, simulation, and benchmarking.

Historical (AMY)

Station networks plus reanalysis, interpolated to your site coordinates for baseline and reporting periods.

Typical / long-term

EPW TMY files from climate.onebuilding.org for simulation, and coordinate-precise long-term averages (up to 30 years) for degree-day normalization.

Station networks behind actual-year series

Historical weather for degree days and M&V periods is built on government and national observation databases, quality-checked and interpolated to the requested location.

ISD (NOAA)

Integrated Surface Database — global sub-hourly and hourly observations from airports and surface stations.

MADIS (NOAA)

METAR/ASOS/AWOS and mesonet feeds, including maritime and specialty networks, for denser near-surface coverage.

DWD

Deutscher Wetterdienst station observations across Germany and Central Europe.

GHCN-Daily / GHCNh (NOAA)

NOAA Global Historical Climate Network daily summaries and quality-controlled hourly station records.

ERA5 and MERRA-2 fill the gaps

Where station density is low or records have gaps, reanalysis provides a continuous global hourly series. We combine ECMWF ERA5 (Copernicus Climate Change Service) and NASA MERRA-2 — the same streams cited in our API documentation when describing continuous hourly or daily weather for any location.

ERA5 (ECMWF / Copernicus)

Hourly global reanalysis from 1940 to near-present — used to complete and extend series where station coverage is thin.

MERRA-2 (NASA)

Global reanalysis from 1980 onward — a second independent stream for continuity and cross-checks.

Where typical-year data comes from

A Typical Meteorological Year is a synthetic year assembled from multi-year history (Sandia / ISO 15927-4 month selection). Energy-Data.io uses two complementary approaches:

Coordinate-precise long-term averages

For degree-day normalization we do not snap to a fixed station TMY. We calculate long-term mean temperature series at your exact coordinates for a user-defined window of up to 30 years — auditable for contracts, ISO 50001, and WMO-style climate normals.

In strict simulation language, “TMY” means a month-selected EPW file. In normalization workflows we often say “long-term average” for the coordinate-precise reference series — same role, clearer terminology.

How the pieces fit together

  1. Actual periods use historical station + reanalysis weather at your coordinates (AMY).
  2. Reference weather for normalization uses your chosen long-term window, or an EPW TMY when the workflow is simulation-led.
  3. Degree days are computed from those series at your base temperature, then used in regression and M&V reporting.

Why this matters Transparent sources let auditors and clients reproduce the weather basis — not just accept a black-box degree-day number.

Common questions

  • Historical (actual-year) weather combines NOAA ISD and MADIS station networks, DWD observations, and GHCN daily and hourly records with ERA5 and MERRA-2 reanalysis. Observations are quality-checked and interpolated to the coordinates you request.

  • Station data are measured observations from airports and surface networks. Reanalysis (ERA5, MERRA-2) fills gaps by combining observations with a weather model, so remote or station-sparse locations still get a continuous hourly series.

  • For building-simulation EPW files we use the Repository of Building Simulation Climate Data at climate.onebuilding.org. For degree-day normalization we calculate coordinate-precise long-term averages over a user-defined reference period of up to 30 years — not a fixed nearest-station TMY file.

  • Nearest-station snaps can misrepresent coastal gradients, terrain, and urban heat islands. Coordinate-precise series reduce that proximity error and make the reference period auditable for M&V and ISO 50001 reporting.

Next step

Try the weather on your site.

Run heating and cooling degree days for any location, base temperature, and date range — free, no account required.