Chemical transports and transformations
(top)
MM5
to drive a tracer model for insect migration
Jimy Dudhia hosted Akira Otuka (National Agricultural
Research Organization, Japan). From June 2002 to March
2003 Otuka worked on using MM5 to drive a tracer model
for insect migration and developed a real-time insect
migration forecast system. Destructive hopper-type
insects can apparently be well forecast as they are
carried to Japan by the low-level winds from Chinese
source regions in the spring. Past episodes of high
trapping rates were successfully simulated with a nested
version of MM5 using an inner ten-km grid.
Wildfire research (top)
Explosive crown fire dynamics explored with infrared
imagery
Crown fires are extreme fires in which the fire spreads
rapidly from tree to tree through the canopy. They
are important because their intensity precludes direct
suppression until conditions subside. Understanding
crown fire dynamics requires observations of the 3D
winds in and around the fire, which few remote sensors
can capture because of the rapidly changing (tenths
of a second), and small scales (under one m). Janice
Coen, Shankar Mahalingam (University
of California, Riverside), and John Daily (University
of Colorado) analyzed infrared imagery of a crown fire
gathered during the FROSTFIRE experiment using image
flow analysis techniques to derive wind fields, fire
spread rates, and vertical sensible heat fluxes of
the fire.
|
|
| Movie
6: Observed a sequence of surges of many convective
plumes representing a scale larger than individual
trees shows fingers of flame initially
burst forward tens of meters parallel to the ground
at low levels at speeds up to 28-48 m/s before turning
upward in updrafts up to 32-60 m/s.
These bursts appear to play an active role in propagating
the crown fire and point towards a powerful, dynamic
mechanism of fire spread. |
Mouse over image to begin
movie. Alternately, you may download
the animation.
|
They observed a sequence of surges of many convective
plumes representing a scale larger than individual
trees. They found that these fingers of flame initially
burst forward tens of meters parallel to the ground
at low levels at speeds up to 28-48 m/s before turning
upward in updrafts up to 32-60 m/s.
These bursts appear to play an active role in propagating
the crown fire and point towards a powerful, dynamic
mechanism of fire spread at the heart of anecdotes
where firefighters report being overtaken by fireballs.
This work contradicts prevailing notions that strong
environmental winds are the primary cause for the rapid
spread of crown fires by showing that these along-slope
winds exceeded the weak ambient winds (commonly supposed
to be driving the fire).
Coupled atmosphere-fire modeling
Coen applied NCAR's
coupled atmosphere-fire model to the Big Elk Fire,
which ignited in the afternoon
of 17 July 2002 northwest of Lyons, CO. Fire behavior
was extreme, reflecting the extreme conditions in the
area and throughout Colorado (including the lowest
fuel moistures ever recorded in the area). Initial
spread was rapid, moving up a south slope of ponderosa
pine with crowning and torching. This SIMULATION (figure
49, below) shows four hours in the afternoon of 17
July. Six nested
domains telescope from 10-km grid
spacing down to approximately 50 m. The larger-scale
atmospheric environment was given on a locally run
MM5 simulation (http://rain.mmm.ucar.edu/mm5/),
which produced the mesoscale light westerlies that
were observed. However, the winds in the fire environment
were dominated by afternoon upslope conditions, driven
by the solar heating of terrain and by the winds created
by the fire. The animation of the finest scale domain
covers the steep mountain valley (slopes with rise/run
of 0.7) where the fire ignited and (in the center)
Kenney Mountain, 700 m elevation above the valley.
The animation begins one hour before fire ignition
and shows the mountain/valley solar-induced circulation.
Once ignited, the fire grows rapidly up the south face,
igniting a crown fire that climbs toward the mountain
crest. This perimeter compares well with observations
taken on site that evening. Although the model is still
being improved, Coen is
also using simulations like this to stimulate discussion
among land managers to
determine user requirements for how models like this
might be used in a practical sense. Sensitivity tests
showed that although these fine resolution simulations
capture fine-scale features in the fire line, simulations
at 300-1000 m horizontal resolution still capture the
overall fire spread and could be run much faster than
real time on a single processor of a PC, thus showing
potential for an operational application of this type
of modeling.
|
|
| Movie 7: Red is a buoyancy isosurface of 10 degrees warmer
than
the environment, and the misty white field shows
smoke. The arrows are the near-ground-surface wind
vectors, which are affected both by ambient meteorology
and by the fire itself. |
Mouse over image to begin
movie. Alternately, you may download
the animation.
|
Implementation of a wildland fire component in WRF
In a joint project between MMM and the Wildland Fire
Collaboratory, Edward Patton (visitor,
Pennsylvania State University) and Coen continued
to port the fire-component of the Clark-Hall model
into the WRF framework. Porting
the fire code into WRF will benefit the community by
allowing new users to take full advantage of the many
services that come with a community model, i.e., WRF
is fully supported, uses state-of-the-art technology,
runs on many computing platforms in both serial and
parallel environments, is easy to switch and/or add
physics or numerics, and comes with pre-built analysis
tools. This conversion provides the ability to immediately
link with the emissions and chemistry components of
WRF as well as readily available real-time data initialization
and assimilation. A working version is expected to
be available in the first six months of FY04.
|
