The forecasts this week here in Canberra have been well received by the
public and can be attributed to helpful guidance from the models. The
rainfall we got was keenly anticipated from a week prior to occurring
and the heavy winds of Wednesday were well captured by the models. On
behalf of this office, many thanks.

        It is however, a good time to enquire as to why there remains NO
guidance at all in this country that focuses on the 'precipitable
water'. Neither 'Kenny' nor NMOC guidance of any kind [GASP analysis
accepted]. Is there something out there that I'm unaware of? The US
excessive rainfall policy regularly talks about this variable in their
discussions, and quite frankly, it is an extremely important element in
forecasting good rainfall events, be it in a rain band like that which
we recently had, but also from convective events as well, incl. flash
flooding etc. I strongly recommend that something be available from LAPS
and GASP that allows us to see this variable. I would also be
particularly keen to see what sort of biases these models have in
relation to the amount of moisture they predict. It could be a helpful
factor in allowing forecasters to better use the rainfall guidance.

   Also, I gave a talk in HO in June highlighting the strong correlation
between 3 hourly temperature and pressure changes in the Meso LAPS
model. The data I showed was based on manually extracted values over
about 3 weeks. I've continued to do this and now have just over 100 days
worth of information based on the 12Z run. They continue to strongly
support the comments that were raised in that talk and I strongly
believe that would be of some interest to anybody interested in
improving the MSLP guidance from this product, which is significantly
failing to capture the local pressure changes across the region of
interest to us.

        The following table lists the time period in which the forecasts were
valid, the correlation co-efficient between the 3 hourly temperature and
pressure changes, and the z-score quantifying the strength of the
correlation.
 0 to  +3 hrs: 15Z-12Z  -0.4479   -4.797
 +3 to  +6 hrs: 18Z-15Z  -0.2854   -2.920
 +6 to  +9 hrs: 21Z-18Z  -0.3703   -3.869
 +9 to +12 hrs: 00Z-21Z  -0.7901   -8.809
+12 to +15 hrs: 03Z-00Z  -0.4633   -4.990
+15 to +18 hrs: 06Z-03Z  -0.3625   -3.778
+18 to +21 hrs: 09Z-06Z  -0.1996   -2.013
+21 to +24 hrs: 12Z-09Z  -0.5657   -6.379
+24 to +27 hrs: 15Z-12Z  -0.4837   -5.252
+27 to +30 hrs: 18Z-15Z  -0.3577   -3.724
+30 to +33 hrs: 21Z-18Z  -0.4415   -4.717
+33 to +36 hrs: 00Z-21Z  -0.8120   -11.272

        With the possible exception of the +18 to +21 hr period in which the
correlation is significantly different from zero at 4.4% the other
correlations are extremely significant. Whilst I concede that there is
some reason for negative correlations between temperatures and
pressures, eg in the afternoon when rising temps coincide with falling
pressures, to have EVERY single correlation a negative one to the extent
that we have here is clearly not correct. The two most excessive
correlations occur in the 3 hours to 00Z [i.e. +9 to +12 hrs and +33 to
+36 hrs], when the strongest temperature rises occur. During this time
the model drops the pressure greater on the days in which the
temperature rise in the most. However in this 3 hour period, strictly as
defined, the pressure actually rises slightly. It doesn't fall. I would
be more than happy to have some discussions about why the model does
this. Local pressure differentials between the ranges and coast are
crucial to a number of significant weather features across the region.
The failure of the Meso LAPS to mimic these changes in a significant way
makes some aspects of this guidance difficult, if not impossible, to
interpret.

        Having said that, the main thing I'd like to re-affirm is the
appreciation of the guidance we've received over the past week or so.
Thanks.

        Dave.

Phil Davill

Hi Dave,
 Yes there is precip water available.
 With the new 72hr laps avalaible i have been experimenting with ADDE over some data. By reducing the domain slightly and being selective in what I get (ie sigma levels, height, mixing ratio, u, v mslp rain, screen temp
and precip water) the file size is under 2 meg. Kenny on loading will calculate TD, humiditiy, divergence, vorticity etc. One advantage of doing it this way is we get ll levels at or below .85 as opposed to the 5 levels
that are sent to the regions.
 A minor modification to Kenny (version 10.3 I am working on) will display the Laps precip water.
Precip water is also available on the US (NCEP) model data and possiblly the GASP data that Neil Adams excellent web based interactive display (I am at home so can not check that for you).
 I assume you could view the LAPS precip water on goats or using mcidas.
Phil Davill