How Open-Meteo ERA5 Weather Data Is Used in Solar Performance Analysis

ERA5 is ECMWF's fifth-generation atmospheric reanalysis dataset, providing hourly solar irradiance at 9 km global grid resolution from 1940 to the present. Open-Meteo is an open-source API that delivers ERA5 data without cost for non-commercial use. Together, they provide the actual historical weather record that physics-based solar models need to answer what your system should have produced in any given month.

Definition and How It Works

ERA5 is not a forecast and not a simple weather station record. It is a reanalysis: ECMWF's atmospheric model is run over historical periods, assimilating observations from satellites, weather balloons, ground stations, ocean buoys, and aircraft to produce the most physically consistent reconstruction of past atmospheric conditions possible at every grid point globally.

For solar modeling, ERA5 provides three irradiance variables at every grid point for every hour since 1940: GHI (global horizontal irradiance), DNI (direct normal irradiance), and DHI (diffuse horizontal irradiance). It also provides 2-meter air temperature and 10-meter wind speed, which are needed to calculate panel operating temperature and the resulting efficiency losses.

Open-Meteo exposes ERA5 and other numerical weather prediction model outputs through a clean API. This makes the full ERA5 record accessible to solar modeling applications without requiring each user to download hundreds of gigabytes of raw ECMWF data files. See how these data inputs flow into the physics model at /resources/methodology.

ERA5 Data Attributes

Attribute	ERA5 Value
Spatial resolution	~9 km global grid
Temporal resolution	1 hour
Historical record	1940 to present
Irradiance variables	GHI, DNI, DHI
Other variables used in solar modeling	2m air temperature, 10m wind speed
Primary use in solar	Physics-based expected production modeling

Why ERA5 Outperforms TMY for Performance Benchmarking

TMY (Typical Meteorological Year) data represents a statistically constructed average year, assembled from typical months across a 30-year historical record. PVWatts uses TMY as its weather input, which is appropriate for pre-installation production estimates. The problem is that TMY data does not correspond to any actual year.

A homeowner whose system was installed in 2021 cannot compare their monthly production to a TMY-based model, because 2021's weather was not the TMY average. If January 2022 was 30% cloudier than the TMY January, a TMY-based model will predict far more production than the actual weather warranted, making the system look like it is underperforming when it is not.

ERA5 provides the actual hourly irradiance that reached a location in each specific month. A physics-based model using ERA5 answers the right question: given what the weather actually was, how much should this system have produced?

Resolution and Accuracy

ERA5's 9 km grid spacing means each grid point represents roughly 5.6 miles. For most suburban and rural residential locations, the nearest ERA5 grid point provides irradiance data that matches ground-truth pyranometer measurements within a mean bias error of approximately 2-5% globally. In clear-sky desert regions (Phoenix, Las Vegas, inland California), ERA5 accuracy is at the high end of that range. In complex terrain (mountain valleys, coastal fog zones) or at high northern latitudes in winter, accuracy may be toward the lower end.

Dense urban canyons, hillside locations with significant horizon obstruction, and sites near large water bodies may see slightly higher error because ERA5 cannot resolve sub-9 km topographic features. For the vast majority of US residential rooftop solar locations, this is not a meaningful limitation.

Frequently Asked Questions

How is ERA5 different from a weather station near my house? A weather station measures conditions at one point. ERA5 provides a physically consistent field that covers every location on earth, with no gaps. Many regions lack weather stations with solar irradiance sensors (pyranometers), and those that exist may have data gaps or calibration issues. ERA5 is more consistent and more complete, even if a nearby high-quality pyranometer might be slightly more accurate at that exact point.

Does ERA5 data go up to the current month? ERA5 has a data latency of approximately 5 days to 3 months depending on whether you are using the preliminary near-real-time ERA5T dataset or the final ERA5 release. For practical solar monitoring purposes, ERA5T data available through Open-Meteo covers recent months with accuracy nearly equivalent to the final release.

Can ERA5 account for local shading from trees or buildings? No. ERA5 provides the irradiance arriving at the top of the local atmospheric column, before any ground-level obstructions. Shading from trees, chimneys, neighboring structures, or panel-to-panel self-shading must be modeled separately using the system's layout geometry. ERA5 handles the sky; the physics model handles the site.

---

Data: pvlib physics modeling + Open-Meteo ERA5 weather data | Last updated: 2026-04-08 | Solar Benchmark