Gedday,
Dicko pointed out something
interesting at this morning's VRO chart
discussion:
If you go to the NMOC MSLP Forecast Charts comparison page
http://web.bom.gov.au/nob/nmoc/srco/chart_discussion/chart/index.htm
and have a look at the 12
Z progs for +5 days (i.e verifying at 12 Z 1
June), you'll see that over
southwest WA, GASP has placed
a deep-ish low
with the 1008 line going
through Perth. The EC, JMA,
UK progs actually
have a ridge extending over
southwest WA, with the pressure at Perth
around 1020 to 1024 hPa.
So: Here is the question:
We are running an Ensemble Prediction System
for GASP: One would
hope that this path in prog space taken by UK, EC and
JMA would be one of the
paths encompassed by the ensemble of GASP progs.
Is this the case?
I don't know yet I'll have to look through the GASP
ensemble outputs?
Whether the answer is yes or no, what are the
implications for use and
interpretation of the GASP ensemble system?
Cheers
John McB
David
Thomas
Hi John,
if you look at the spaghetti
diagram for the ensemble 500hPa you will see I
high degree of uncertainty
in the IO very soon into the model run. It is not
surprising that the WA situation
becomes a "replace high by low" scenario,
even from day 4.
cheers
Dave
Mike Naughton
Hi John et al.
gasp_eps has moved from sx5
to sx6, but the web pages are still
catching up. At this
moment they are at
http://gale.ho.bom.gov.au/bm/internal/daas/ensemble2/index.html
(which is the link under Trials on the NMOC home page)
The other usual links e.g.
from the NMOC Chart Discussion page
should be up and running
again in next few days. :-)
Looking at results there,
there are number of members with ridge
SW of WA at day 5: p1, p4,
p5, p7, p12, p13, p15, p17, p19, p21,
p24; which together add
up to a significant proportion of the
members.
The day 4 tubing picks up
two tubes of significant outliers
headed by p7 and p21 respectively:
- p7 tube picks up
p12, p24;
- p21 tube picks up
p2, p4, p13.
Looking at these nearly sends
me blind; hopefully we can do
something soon to improve
the quality of the graphics.
As Dave commented, ther spaghetti
shows lots of variation in this
area, but it's not really
easy to see whether there are some
members without the low
(as the stamps and tubes show) or if they
different members simply
have different phasing of the system.
Cheers,
Mike
John
McBride
On Fri, 28 May 2004, Mike
Naughton wrote:
> Looking at results there,
there are number of members with ridge
> SW of WA at day 5: p1,
p4, p5, p7, p12, p13, p15, p17, p19, p21,
> p24; which together add
up to a significant proportion of the
> members.
>
> The day 4 tubing picks
up two tubes of significant outliers
> headed by p7 and p21 respectively:
> - p7 tube picks
up p12, p24;
> - p21 tube picks
up p2, p4, p13.
>
So... if on closer inspection
the p21 tube looks a lot like the JMA, EC
and UK progs, can we extend
that back to the initial
analysis/starting-condition
to determine whether the difference between
their models and GASP lies
mainly in the analysis in a particular region?
(Probably a naive question;
but maybe we can all learn something from
this)
>
John McB
Pete Steinle
Am already in the process
of doing exactly that (but with p4 - which
also had the trogh over
SA closer to the OS models). Its a bit of a
worry & confusing:
A worry because GenSI has
just gone operational in the last few weeks,
so I have been waiting for
a bomb.
Confusing because Chris T.
has been testing GenSI and LAPS and getting
the opposite problem - in
one case it did NOT develop a Tasman low
enough.
..and a pain the posterior,
as there were no real signs of either of
these problems in the many
months testing beforehand!
Pete
Jin Lee
I plotted spread in MSLP
among the global models and also spread in
Z(500hPa)(attachment). You
can see at analysis time an area of spread
northwest of Kerguellen
Island - about 2 hPa in MSLP and 30 gpm in Z(500
hPa). Admittedly these are
very small numbers but if you step through to
subsequent forecast hours
the spread area propagates downstream and grows
until the spread reaches
west of W.A. at +5 forecast days.
Spread: MSLP 00
24
48
72
96
120
Spread Z500
00
24
48
72
96
120
Satellite
05/27
12Z 18Z
05/28 00Z
A lot of times I've seen
areas of inital spread coinciding with areas of
large baroclinic instability.
In this case the initial spread at 500 hPa
matches up with an upper
trough. To see whether there was a significant
weather system in the vicinity
of the initial spread I looked at some
satellite pictures. It appears
there was a hint of enhanced
convection(associated with
CVA maximum?) at 12Z northwest of the Island
where the initial spread
seems to have originated(attachment) and this is
more evident at later times.
So it may well be that the
forecast uncertainties in the global models
resulted from downstream
propagation of uncertainties in the initial
condition.
Cheers,
Jin
