Vaisala Xweather Energy

Vaisala Energy Support

Which Wind Time Series dataset should I choose?

Vaisala offers three datasets in our Wind Time Series Tool. Two of these, ERA5 and MERRA2, are global in coverage and considered classic reanalysis datasets, in that great care was taken to ensure consistency over their long periods of record. The third, HRRR (High Resolution Rapid Refresh), is a forecast dataset that covers only the continental United States. HRRR is available for a shorter period of time but at much finer resolution.

The reanalysis models, ERA5 and MERRA2, are fairly coarse in spatial resolution at roughly 30 km and 90 km respectively. To overcome some of the limitations of that coarseness, the ERA5 and MERRA2 data offered by Vaisala have been downscaled to match the long-term annual average value of our 5km global wind dataset as described here

The HRRR data is at a spatial resolution of roughly 3 km and comes from a weather forecast model run every hour, assimilating all the latest observational data from ground, radar, and satellite sources. As a forecast model, HRRR is not expected to be consistent over time so we only offer data since January 2021, after the last major model upgrade. Furthermore, as HRRR is at a finer spatial resolution (3 km) than the Vaisala 5 km dataset, it is provided with no further spatial enhancement or mean matching.

So which version to choose? Where all three sources are available, this question is addressed in detail in Davidson and Millstein, 2022 and Millstein, 2023 (references below). In brief, the most accurate data available is from HRRR though it is only for the USA and the period since January 2021. The ERA5 dataset is considered by many to be the gold standard for long-term global reanalysis data and while it is at coarser resolution than HRRR has a much longer duration, making it more useful as a long-term reference. Although it has the coarsest spatial resolution, MERRA2 also has a long history and can be more accurate at some locations. This makes it useful as a second opinion when considering long-term variability of ERA5 (or vice-versa).

A paper comparing all three datasets against wind energy production data in the USA can be found here (Davidson and Millstein, 2022) and here (Millstein, 2023):

More details about the HRRR model can be found here

More details about ERA5 can be found here

More details about MERRA2 can be found here

If you didn't find what you were looking for, please contact 3TIER Support for additional assistance.