# What is the meaning of step in ERA reanalysis data?

The webpage to download ERA data includes an option called "time" which is intuitive, but also an option called "step". I can't find an explanation of the meaning of this. It can't be a time step because it has also 0.

• I suggest editing the question to define the acronym "ERA" Oct 13, 2015 at 0:30

## 2 Answers

The "step" is the number of hours the ERA re-forecast has been run into the future from the "time" of the re-analysis.

For example, asking for ERA data with time = 12:00 and step = 0 means asking for data valid at 12:00 based on a re-analysis for 12:00. But asking for data with time = 12:00 and step = 6 means asking for data valid at 18:00, i.e. for data produced by a 6 h model run based on the 12:00 re-analysis.

To fill the gaps between the analysis times one can use steps from a forward run. Forward runs are needed to get parameters like time accumulated precipitation, which would not be available from an analysis. Sometimes a combination of steps and times might be appropriate for a joint analysis of instantaneous and time averaged parameters, i.e. to stagger them correctly in time.

Data valid 00:00 deviate less from the observations if they come from a 00:00 analysis compared to data from step 12 of the 12:00 analysis. But a series of steps from one initialization time should be more consistent in time compared to a series of independent analyses.

• thank you, it's still somewhat unclear though, all I need is the data at the time it was measured/modelled, what is the purpose of the step, why would someone want to use a step 6 or 12? It would be night instead of day. Oct 15, 2015 at 15:02
• @user4050 e.g. because sometimes it's useful to know the difference between "forecast" and "outturn", and comparing the 00:00 figures with the previous day's 12:00 + 12h forecast is one way to approach that. Oct 16, 2015 at 8:59
• @EnergyNumbers why is it that some variables are not available at 6 and 12 time and not available at step 0? Oct 26, 2015 at 19:13
• I think it could be useful to address the availability of variables in a separate question.
– BHF
Oct 27, 2015 at 6:45
• @user4050 ^ just as BHF says Oct 27, 2015 at 10:55

Some confusion comes from the mixture of analysis data (such as 2m-air temperature) and reanalysis products (such as downward radiation) in the final reanalysis product. The following text directly from ECMWF may help

https://confluence.ecmwf.int/pages/viewpage.action?pageId=56658233

• Your answer is in another castle. Can you edit this and quote the relevant parts here? The text as written now does not seem to answer the question. Dec 5, 2018 at 9:10