Gary Weymouth

I've been puzzled (for a long time) by apparently low 1000-500
LAPS thicknesses on DIFACS charts.
These are typically 2 to 4 gpdm lower than for the other
(GASP and international) model output thicknesses.

Today I plotted up what McIDAS thinks the thicknesses are
(dataset RTDB2/LAPS, parameter THKN, eg for 12Z tonight).
These McIDAS LAPS thicknesses are higher than those on DIFACS,
and seem to me to agree more closely with other model output.
It looks to me that on the face of it, the DIFACS display of
LAPS thicknesses is the odd one out.

Does anyone have an explanation for the differences?

Regards,
Gary

Charlie Sanders

> Does anyone have an explanation for the differences?

Not really, but I have checked a few things.

LAPS difacs displays are produced by a special display program. I do not
know anything about its internals.

You are correct in noting differences between the Difacs displays and
the
McIDAS TKKN (and the result of subtracting the 500 and 1000 Z in
MciDAS).

I compared the +36 prog valid 12Z 18 Jan with the fields in the database
as shown by McIDAS and noted the following

        - McIDAS field is 20-40m higher over most of the tropics and
          in S Indian Ocena around Kerguelen.
        - Generally little difference elsewhere, although I may have
          missed some differences.

I tried to run some of the other display programs from the LAPS output
files without success. Perhaps Tan Le may be able to produce a 1000-500
thickness display from the raw model output netcdf files to compare
with that generated by the LAPS display program. Perhaps there is a
difference in the vertical interpolation algorithm between the LAPS
display program (generating the difacs display) and the post-processor
(generating the data for the database, as displayed by McIDAS).

I also checked the display jobs, and LAPS uses its own display program.
GASP uses the Diagnostic Post Processor and computes thickness as 500
height - 1000 height

UK, US, EC all use the Diagnostic Post Processor compute thickness as
        500Z - (MSLP-1000)*8.1
which is an approximation. This approximation overestimates thickness
where the temperature is warm and the pressure high, or where the temperature
is cold and the pressure low, and underestimates where the temperature is
warm and the pressure low or where the temperature is cold and the pressure
high.

We could change the UK and US display programs to use 500 height - 1000
height as we do receive the 1000 height now (we did not when the
displays were set up). We cannot do this for the EC as we do not receive the 1000
height.

The differences between the accurate formula and the approximation seem
to be of the generally less than 10m. For the LAPS the biggest difference
between the approximation and the more accurate formula was about 11m
over SE Aus. For GASP it was about 200m over Antarctica, 150m over
Greenland, 200m over the Andes and about 300m over Tibet, all areas of
very high topography.

Over the Australian area the largest difference for GASP was about 11m
over SE Aus. The US model had differences of about 12m near Wilsons prom,
Cairns and Giles and  was very "noisy". Over the Aus region, the largest
differences between the two formulae for the UK model were over Vic,
East NSW and SE Qld, all about 12m. Globally the UK had large differences
over Antarctica (60m), Greenland (150m) Tibet (60m), Canadian Rockies (60m)
and Andes (100m). The largest differences between the formulae for the
US model were about 20m over the N Atlantic and Arctic Oceans.

I think the approximation is OK over Australia.

Charles