Comparison of RAMS results to
those of other models
Results from RAMS on the previous page showed
predictions of weather for the northwestern United States on November 5
through 18. A group at the University
of Washington runs another mesoscale model, MM5,
at resolutions of 36, 12 km and 4 km on a semi-operational basis.
The forecasting domain that they used does not correspond exactly to what
was used for SC99. However, the 12 km MM5 nest overlaps with most
of the 16 km nest used for RAMS at SC99, so these will be compared.
In addition, the MM5 runs employ 33 vertical levels vs. 31 used for RAMS.
Details of the MM5 forecasting environment and model configuration are
available.
Like the RAMS runs, they are using Eta results from NCEP for both initialization
and boundary conditions for a 48 hour simulation, produced twice each day.
All of the results are posted as a set of static two-dimensional graphical
images on the web. The MM5 forecasts are computed on a 14-processor
SUN
Microsystems Ultra Enterprise 4000 Server. Currently, the model
is run for the entire 48 hour period over the 36 and 12 km domains, and
for 24 hours (12 to 36 hours into the forecast) for the inner 4-km domain.
A 48 hour forecast requires 5-6 hours to complete on this machine.
No information on the internal time step is available.
In an attempt to assess the differences between these two forecasting
efforts, a qualitative comparison of RAMS results at 16 km resolution with
MM5 results at 12 km resolution is discussed below for forecasts initialized
at 0Z on 16 November 1999. The comparison is limited at best for
a number of reasons, including:
-
Different domains are used in each forecast (resolution [grid spacing horizontally
and vertically], extent [three-dimensional mesh size], centroid [center
of the domain], and projection [cartographic coordinate system])
-
Weather variables in the plots from MM5 are expressed in different parameters
and/or units.
-
The only available information on MM5 results are static plots. There
is no access to actual data that could be compared. In addition,
only limited time steps common across the various forecasts were available.
(This was not a problem for RAMS since all of the data were available.)
-
Simultaneous animation is limited as a comparison technique because each
set of hourly plots from a particular MM5 forecasts uses different scales
for contours, color, etc.
One set of MM5 results are plots of 925 mb temperature, sea level pressure
at winds at 10m. The closest to that from RAMS is 950 mb temperature,
mean sea level pressure and surface winds. The following images are
the results at 0Z, which are basically derived from the initialization
from the Eta forecast.
The general distribution of temperature and pressure is similar between
RAMS and MM5. However, there is inconsistency in the wind direction
between the two results, although there is good correspondence in the speeds
at least as coarsely quantized via wind barbs.
Here are the same visualizations for 15 hours into the forecast.
There is similar structure in the pressure and temperature fields between
the two forecasts. However, the pressure from RAMS is about 5 mb
higher. The temperature data from RAMS covers a smaller dynamic range.
The inconsistency in the wind direction between the two results still persists,
although there is good correspondence in the speeds at least as coarsely
quantized via wind barbs.
Animations from both sequences are presented below.
lloydt@watson.ibm.com
Last updated January 4, 2000
