Elaheh, in the ECMWF catalogue, "time" signifies the epochs of the actual analysis, where the system makes use of all available recent observations to produce a realistic state estimate of the atmosphere. These epochs usually are at 0, 6, 12, and 18 UTC. From those analysis times, short forecasts are issued that project the atmospheric state to the future and provide valuable background information to the system for the next analysis cycle. These forecasts mostly rely on the model, not the observations. The "steps" that you see are snapshots of these forecasts at 0, 3, 6, 9, and 12 hours distance from the initialization epoch (= analysis times). So if you download fields for "time 6" + "step 3", you will get forecasted data at 9 UTC, but be aware that the quality of forecasted fields is generally lower than in the case of analyses. Regards, Michael
Elaheh, in the ECMWF catalogue, "time" signifies the epochs of the actual analysis, where the system makes use of all available recent observations to produce a realistic state estimate of the atmosphere. These epochs usually are at 0, 6, 12, and 18 UTC. From those analysis times, short forecasts are issued that project the atmospheric state to the future and provide valuable background information to the system for the next analysis cycle. These forecasts mostly rely on the model, not the observations. The "steps" that you see are snapshots of these forecasts at 0, 3, 6, 9, and 12 hours distance from the initialization epoch (= analysis times). So if you download fields for "time 6" + "step 3", you will get forecasted data at 9 UTC, but be aware that the quality of forecasted fields is generally lower than in the case of analyses. Regards, Michael
Well, the 3-hour forecast may actually be better for some variables than the analysis... because the model has re-adjusted to its own physics and can be more consistent.
... That is for sure the case with at least some of the wave parameters; check here:
Ole Johan Aarnes, Saleh Abdalla, Jean-Raymond Bidlot, and Øyvind Breivik, 2015: Marine Wind and Wave Height Trends at Different ERA-Interim Forecast Ranges. J. Climate, 28, 819–837.doi: http://dx.doi.org/10.1175/JCLI-D-14-00470.1
I need to develop a climatology of low level cloud cover, convetive precipitation and other modelled variables. for +1 UTC I suppose the best would be to download output using time=12UTC and step= 6 to give time to the model to develop precipitation and get the output for 18 UTC. Is it OK?
Dear Andari reanalysis data is blended with forecast, any type of data that you get from a model is somehow blended with forecast, the difference is that in some data-sets reanalysis is exactly the forecast one, and in other reanalysis is forecast+observation,
Andari, actually not all the Interim datasets include forecast+observation, many of them are just foretasted data, (I don't know exactly which one) we do not have observation of all parameters globally, and on the other hand these observations are not just measurements taken in meteorological stations, some are remote sensing data (satellite, radar...).