IBM demonstrates regional weather
forecasting system at the annual IBM stockholder's meeting
The capabilities developed and utilized for the Olympics, AMS, CMA and
SC97 have since been employed in other operational forecasting settings.
For example, at the annual IBM stockholder's meeting on April 28, 1998
in Chicago, IL this capability was replicated as part of the company's
technology demonstrations at that meeting. The system was adapted to the
Chicago area as shown below. For this and any of the subsequent images,
you can view a higher-resolution version by simply clicking on it.
You can also interact with this map via a scene in PanoramIX
or simplified VRML. A flythrough animation
of the terrain is also available for you to watch.
New 24-hour mesoscale forecasts were produced at 8 km resolution in
a region roughly 680x680 km in extent during the meeting on a three-hour
update cycle. The computation took place on six 120 MHz P2SC thin nodes
in an SP that was brought to the meeting. One 135 MHz P2SC wide node
was used for I/O. Two workstations (an IBM RS/6000 43P-240 and an
IBM Intellistation M-Pro) were available to interact with the model and
analyze results in a fashion similar to what was used at the Olympics and
SC97. In this case, raw observations were received regularly over
the internet at FSL in Boulder, CO, where the LAPS pre-processor step was
executed. The LAPS results were then networked to the SP via the internet
from FSL for ingest into RAMS. A 24-hour forecast over the aforementioned
domain took about 2 hours. This improved performance was due to the
incorporation of a more efficient microphysics package developed by FSL
and additional optimizations. The previous implementation at SC97
achieved similar performance with eight 160 MHz P2SC thin nodes.
Compared to the system used to support forecasting at the Olympics, this
was faster performance with only one-fourth the number of thin nodes.
Output from RAMS every 10 minutes of forecast time were provided for
browsing visualization. Animations were produced routinely as the primary
mechanism to evaluate the model output through the procedures proven at
the Olympics and other installations. Image-based rendering of three-dimensional
scenes via PanoramIX
and geometric descriptions of three-dimensional scenes via VRML
incorporating simplified
geometry were available. The system also permitted simple tracking
of the simulation, so that the interactive tools could be utilized while
the model was running. If there were problems in the model run, then the
execution could be terminated and the model restarted with new input observations.
This operational numerical forecasting coupled with routine, interactive
three-dimensional visualization was unprecedented in such a setting.
In addition, interactive visualization applications for the analysis
of post-processed model results and as well as the output
from LAPS were available operationally. All of these
capabilities are illustrated below.
The following image is from one of the animations
produced during the meeting. (The animation
can also be viewed at higher resolution, but the file is 2.5 times bigger.)
The 85-frame animation shows a prediction of high thin clouds to the northwest
of Chicago. It further illustrates the lake-effect winds across Lake
Michigan, a prediction of cool, breezy and dry weather in Chicago on April
28. Predictions from the system for April 27-28 were quite accurate.
A comparison of several model runs with the weather observations for the
same period showed that the rms deviation in temperature was about 2o
C. and the rms deviation in wind speed was about 2 m/sec averaged across
the entire domain.
The image shows a terrain map, pseudo-colored by temperature overlaid
with coastline, county and river maps for 4 PM on April 28. Predicted
winds are illustrated by arrows, colored by speed. The clouds are
visualized as a white, translucent isosurface of cloud water density.
Some major cities in the area are also marked as well as predicted temperatures
at those locations. This particular time step can also be examined
via a flyover animation, simplified VRML
geometry and a PanoramIX scene. Other
results from the same model run are shown in the following images.
The first illustrates surface winds with streamlines, pseudo-colored by
speed with directional arrows that highlight the lake breezes. The
terrain surface is pseudo-colored by predicted dew point temperature.
This next image shows surface winds with flags, pseudo-colored by speed
pointing in the direction of the wind. The terrain surface is pseudo-colored
by predicted humidity.
This image again illustrates surface winds with arrows. The terrain
surface is pseudo-colored by predicted surface pressure. Instead
of a continuous pseudo-color presentation, the domain is segmented into
10 pseudo-colored filled contour bands.
Here is a sample image, which shows the area being
forecasted and some of its major cities at 7 PM on Sunday, April 26, from
a model run the day before. The local terrain is visible, overlaid
with state, county, river and coastline maps. The topography is colored
by predicted precipitation, where blue regions illustrate up to one inch
of total rainfall. The predicted rainfall at specific cities is indicated.
Local high and low pressure regions are marked by an H and L, respectively.
Predicted clouds are illustrated by white, translucent surfaces.
A cyan colored surface is a forecast of a rain shaft corresponding to a
squall line moving across the region. The model accurately predicted
the location and time of this precipitation when compared to the actual
observations. This can be seen in a 145-frame animation.
This particular time step can also be examined via a flyover
animation, simplified VRML geometry and
a PanoramIX scene.
The next example image is for 6 AM on Wednesday April 29, from a model
run the day before -- the last completed during this experiment.
The local terrain is visible, overlaid with state, county, river and coastline
maps. The topography is colored by predicted precipitation, where
blue regions illustrate up to 0.5 inch of total rainfall. The surface
is also overlaid with arrows which point in the direction of winds and
are colored by speed (white is 30-35 mph). Predicted clouds are illustrated
by white, translucent surfaces. A cyan colored surface is a forecast
of a rain shaft corresponding to a band of rainfall moving through north-central
Illinois. The model accurately predicted the location and time of
this precipitation when compared to the actual observations the next day.
This can be seen in a 126-frame animation. (The
animation can also be viewed at higher resolution,
but the file is three times bigger.)
This particular time step can also be examined via a flyover
animation, simplified VRML geometry and
a PanoramIX scene.
After each RAMS execution, all of the results
are collected and reorganized into a form that can be used by standard
meteorological analysis tools as provided by FSL, for example. This
post-processed data were made available for interactive three-dimensional
visualization and analysis via a viewer application. This includes all
of computed variables from the model, but at hourly resolution unlike the
browser application that worked with a subset of variables but at six times
the temporal resolution. Here is a sample image and animation
created with this application for a forecast initiated on Saturday, April
25 at 2300 UTC (6 PM CDT).
A surface variable (mean sea level pressure) has been selected for
display as pseudo-color, which is overlaid on a topographic map.
Rivers (blue) and coastlines (black) are draped on the surface. An
upper air variable (reflectivity) has been selected for display via surface
extraction. The surface at 20 dBz is requested in translucent brown.
Another field (specific humidity) has been selected to show as a vertical
slice, which is pseudo-color contoured. Any of the three-dimensional
fields available from the model can be visualized with either of these
methods. The upper air wind data are shown via streamribbons.
The seed points are along vertical profiles, which are specified interactively.
The upper air wind data can be seen along three vertical profiles, which
are specified interactively. The direction of the model wind field
along these "virtual soundings" are shown via vector arrows pseudo-colored
by horizontal wind speed. The length of the arrows also corresponds
to the horizontal speed. Each profile is realized as a pseudo-colored
tube, which is contoured by the variable selected for isosurface realization
(i.e., reflectivity). No tubes are seen on two of the soundings because
the model does not predict any reflectivity at these locations. They
can be seen in the animation, which shows
a prediction of squall line moving through the area.
Here is another sample image and animation
created with the RAMS viewer for a forecast initiated on Sunday, April
26 at 1500 UTC (10 AM CDT).
A surface variable (total precipitation) has been selected for display
as pseudo-color, which is overlaid on a topographic map. Rivers (blue)
and coastlines (black) are draped on the surface. An upper air variable
(cloud ice) has been selected for display via surface extraction.
The surface at 0.1 gm/m^3 is requested in translucent brown. Another
field (reflectivity) has been selected to show as a vertical slice, which
is pseudo-color contoured. The slice is only partially visible for
where the data actually exist. Any of the three-dimensional fields
available from the model can be visualized with any of these methods.
The upper air wind data are shown via streamribbons. The seed points
are along vertical profiles, which are specified interactively. The
direction of the model wind field along this "virtual sounding" is shown
via vector arrows. Both the arrows and ribbons are pseudo-colored
by horizontal wind speed. The length of the arrows also corresponds
to the horizontal speed. Each profile is realized as a pseudo-colored
tube, which is contoured by the variable selected for isosurface realization
(i.e., cloud ice). No tubes are seen on the soundings because the
model does not predict any cloud ice at these locations. The results can
also be viewed with the visualizations produced with the browser.
As part of the effort in developing forecasts, the results
from the LAPS pre-processing, assimilation step are analyzed as well as
RAMS output. Tools similar to those for RAMS were also developed for the
study of LAPS results. A discussion of this capability and example
results are available for you to examine.
lloydt@watson.ibm.com
