[NARCCAP-discuss] Consistency between tas, tasmax and tasmin.

David Bronaugh bronaugh at uvic.ca
Tue Feb 7 14:07:50 MST 2012


Hi Gauvin,

I think you have a methodological problem with your plotting. If you want
to display tasmax/tasmin for a day, the lines for tasmin and tasmax should
always be flat, but yours are on angles. Can you replot using independent
line segments for each day for each of tasmax/tasmin? I believe this will
probably "resolve" the second problem. The first one does sound like a real
data issue; I can't speak to it.

David

On Tue, Feb 7, 2012 at 11:59 AM, Gauvin St-Denis, Blaise <
GauvinSt-Denis.Blaise at ouranos.ca> wrote:

> Hi,
>
> There seems to be inconsistent behaviour between tas, tasmax and tasmin in
> some narccap simulations.
>
> To frame the problem, we have 3-hourly temperatures (tas) and daily
> minimum and maximum temperatures (tasmin,tasmax). One would expect that:
> tasmin <= tas <= tasmax
> during a given day. Now, some models might have their 'day' start at a
> different hour and compute tasmin/tasmax with respect to the previous 24h,
> but still, this relationship should be valid within at most a 24h offset.
>
>
> Here are the problems I have:
>
> WRFG-CCSM: Offset of multiple days between tas and (tasmin,tasmax). Could
> this be linked to the missing timesteps problems?
> example in attached figure (tasmm_wrfg_ccsm.png): we can see that there is
> an offset of more than 48h.
>
> Note that I did not see this problem in the WRFG-CGCM3 run.
>
>
>
> RCM3: 3-hourly temperatures often overshoot/undershoot the (tasmin,tasmax)
> envelope.
> example in attached figure (tasmm_rcm3_gfdl.png): a few points where the
> 3-hourly temperature is well below/over 3 deg. C of tasmin/tasmax
>
> Note that this is present in both RCM3-GFDL and RCM3-CGCM3.
>
>
>
> Any confirmation/comment on those issues would be appreciated,
>
> Thanks,
>
> Blaise Gauvin St-Denis
> Groupe Scénarios Climatiques
> Ouranos
>
> st-denis.blaise at ouranos.ca
> Tel. 514-282-6464 #302


More information about the narccap-discuss mailing list