20CR: ensemble mean wind-speeds appear wrong

Created by Neil.Swart on - Updated on 08/09/2016 11:28

I suspect that the 20CR.v2 ensemble mean wind speeds provided publicly were calculated incorrectly, and I will briefly explain.

As confirmed in the attached plots, the ensemble mean speeds provided for 20CR were computed as:

wspd = [<u>^2 +<v>^2]0.5

where <> indicates the ensemble mean. The correct calculation would be

wspd = < [u2 + v2]0.5 >

As you can see in the attached, the provided ensemble mean speeds look nothing like the speeds in the individual ensemble members, and appear  wrong.

After downloading all the individual ensemble members and recalculating the speeds using the correct formula above, I get an ensemble mean which is perfectly

consistent with the individual members (i.e. falls in the middle of them), as would be expected.

The ensemble mean wind speeds provided also look physically unrealistic (e.g. too low at extratropical latitudes prior to the 1990's), and exhibit unrealistically

large trends. When using the individual ensemble members, or the correctly re-calculated ensemble mean speed, these issues are corrected.

I suspect many users could be affected by this, and I hope that this info might be useful.

-Neil Swart


(top) 20CR.v2 wind-speeds at sig0995 averaged over 40 to 60 S and smoothed with a 5-year wide boxcar (bottom) 20CR.v2 wind-speed climatology over 1871 to 1899. Wind speeds are shown for the provided ensemble mean monthly speeds (from ESRL), and as calculated from the daily u and v components of the individual ensemble members (downloaded from NERSC). Also shown are two different computation of the ensemble mean derived from the individual members, computed as per the formula in the figure key.























Note in the time-series plot attached the data are all smoothed with a 5-year wide boxcar. 


Mon, 11/25/2013 - 10:47

I also have a question:

Can you confirm whether this issue might affect the wind-stress fields too?

I.e.: Calculating the ensemble mean stress as

tauu = ^2

instead of

tauu = < u^2 >

Or are the stess fields computed online in the model from the instantaneous U and V and therefore okay?


Mon, 11/25/2013 - 10:39

Hi Gil,

In these calculations using all the individual ensemble memebers I was only considering the sigma .995 level winds. I have also looked in some detail
at the 10 m winds - at least at the ensemble mean 10m winds downloaded from ESRL. My guess is that the same issue affects the 10 m wind
speeds, because they also exhibit the (unrealistically) large trends in time, as seen in the sigma .995 winds above.

That said, I cannot confirm this by performing the same calculation with the 10m winds, because the individual ensemble members for the 10 m winds
are not available on the nersc.gov site, as far as I can see.


Neil, 10 m winds are at the portal.nersc.gov site in netCDF4. They are under the "first_guess" directories. For example, for the 10 m zonal wind http://portal.nersc.gov/pydap/20C_Reanalysis_ensemble/first_guess/u10m/ These are also OPeNDAP enabled. The "derived" directories at http://portal.nersc.gov/pydap/20C_Reanalysis_ensemble/ have the daily and monthly averages for each member. Additional selected variables are available from the NERSC Tape Science Gateway in GRIB format: http://portal.nersc.gov/archive/home/projects/incite11/www/20C_Reanalysis/everymember_grib_indi_fg_variables and 3 dimensional and surface grids in GRIB format for most years at http://portal.nersc.gov/archive/home/projects/incite11/www/20C_Reanalysis/everymember_full_analysis_fields Note that the first guess is not directly incremented by the observations. It is the product of the numerical weather prediction model forecast initialized from the analyzed state. Please let me know if I can be of more help. best wishes, gil


Fri, 11/22/2013 - 14:50

Hi Cathy,
Thats good to know, thanks. The resulting wind-speed errors are not small, but in fact are very large (>3m/s or 50% in some cases).
You can see that very clearly in my figure - but I'm having trouble getting that to show up in the post. SOrry but I'm a noob at editing these pages!


Fri, 11/22/2013 - 14:43

We will look at calculating the values from the individual ensemble members. You are correct the wind speed from the ensemble means would be lower than that of the average wind speed of the ensemble members and it would be most different where the ensemble members were most different from each other. We don't have the wind speed file listed in our docs but it is in /Datasets/ I'll remove the files and document this. I'm not sure when we can generate the re-computed files.

Add new comment

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.