[Met_help] [rt.rap.ucar.edu #64608] History for puzzled about rmse, me and stdev

[Tressa Fowler via RT]

----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

hello!
I have plot the RMSE , ME and STDEV in one drawing(the attached) .
 the value and the trend are analogous between the  RMSE  and the STDEV.
but the guide tell the MSE can be strongly influenced by ME, as shown by
this decomposition(C-8) and the RMSE is simply the squre root of the MSE.


the contradiction is shown in the two situation, how to solve it?

Thanks!

Qiang


----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: puzzled about rmse, me and stdev
From: Tressa Fowler
Time: Thu Dec 12 16:11:46 2013

Hi Qiang,

I see your plots, and I think they both demonstrate that bias (ME) has
an effect on the RMSE values, as mentioned in the users' guide and
many forecast verification text books. I do not understand what your
question is exactly, nor do I see any contradiction.

In one of your plots, the bias is negative and in the other, it is
positive. Both bias values have a positive impact on the MSE and RMSE
because the square of the bias contributes to the MSE, as per the
formula in C-8 of the users' guide. Was this perhaps the contradiction
you mention?

If not, please clarify your question so I can provide better
assistance.

Thanks,

Tressa

On Thu Dec 12 14:25:29 2013, liqiang at ou.edu wrote:
> hello!
> I have plot the RMSE , ME and STDEV in one drawing(the attached) .
>  the value and the trend are analogous between the  RMSE  and the
STDEV.
> but the guide tell the MSE can be strongly influenced by ME, as
shown by
> this decomposition(C-8) and the RMSE is simply the squre root of the
MSE.
>
>
> the contradiction is shown in the two situation, how to solve it?
>
> Thanks!
>
> Qiang



------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #64608] puzzled about rmse, me and stdev
From: Li, Qiang
Time: Mon Dec 16 13:27:09 2013

hello!
I understand the bias values have a positive impact on the MSE and
RMSE.
but as is shown in the page  C-8, the MSE or the RMSE is sumed the
square of  ME and the square of STDEV.
 the guide tell the MSE can be strongly influenced by ME.
but in my plot,  the value and the trend are more analogous between
the  RMSE  and the STDEV than the  RMSE  and the ME.

 the contradiction is shown in the two situation, how to solve it?

Thanks!

Qiang
________________________________________
From: Tressa Fowler via RT [met_help at ucar.edu]
Sent: Thursday, December 12, 2013 5:11 PM
To: Li, Qiang
Subject: [rt.rap.ucar.edu #64608] puzzled about rmse, me and stdev

Hi Qiang,

I see your plots, and I think they both demonstrate that bias (ME) has
an effect on the RMSE values, as mentioned in the users' guide and
many forecast verification text books. I do not understand what your
question is exactly, nor do I see any contradiction.

In one of your plots, the bias is negative and in the other, it is
positive. Both bias values have a positive impact on the MSE and RMSE
because the square of the bias contributes to the MSE, as per the
formula in C-8 of the users' guide. Was this perhaps the contradiction
you mention?

If not, please clarify your question so I can provide better
assistance.

Thanks,

Tressa

On Thu Dec 12 14:25:29 2013, liqiang at ou.edu wrote:
> hello!
> I have plot the RMSE , ME and STDEV in one drawing(the attached) .
>  the value and the trend are analogous between the  RMSE  and the
STDEV.
> but the guide tell the MSE can be strongly influenced by ME, as
shown by
> this decomposition(C-8) and the RMSE is simply the squre root of the
MSE.
>
>
> the contradiction is shown in the two situation, how to solve it?
>
> Thanks!
>
> Qiang





------------------------------------------------