Mike Stavish and Dan Weygand, WFO Medford, OR
Event Sequence
On Thursday, February
7, 2002 a rapidly developing and fast moving eastern Pacific cyclone brought
a 2 to 3
hour long episode of high winds to the Southern Oregon coast and subsequently
to portions of the Umpqua and
Willamette valleys. These winds were not adequately forecast. Wind gusts at
Cape Blanco rose from 38 mph at
11 am to 89 mph at 1 pm, a peak gust of 85 mph was recorded at Gold Beach
around 2 pm, and a gust in excess
of 100 mph was recorded that afternoon at Winchester Bay. A tractor-trailer
tipped over at Port Orford around
130 pm. By 3 pm, south winds at the coast had shifted to the west and began
to decrease. By 4 pm sustained
winds at Cape Blanco and Point St. George had returned to near 25 knots. Inland,
Signal Tree RAWS recorded a
peak gust of 77 mph between 330 pm and 430 pm, a peak gust of 39 mph was measured
in Roseburg at 409 PM, and
a peak gust of 70 mph was noted in Eugene at 440 pm. The high winds resulted
in downed utility lines from a
large number of fallen trees.
Most forecasters would
agree that the AVN tends to provide the best model solution of conditions
in the
eastern Pacific. However on this day, it was noted that over the previous
24 hours the ETA had done a better
job with the surface winds over the Southern Oregon coastal waters, while
the AVN had over-forecast the winds
by 10 to 15 knots. The preferred solution on the past 2 shifts had been that
the coastal winds on Thursday
would be strong southwest in the morning and back to east or southeast as
an approaching weak surface low
made landfall near Brookings. It was unclear whether the low would arrive
as a closed low or a weak trough,
which posed difficulty for the wind forecast. The winds would likely have
an easterly component, but either
southeast or northeast was unclear depending on the location and strength
of the low.
On the early morning
(midnight) shift of February 7th, the 00Z run of the AVN, ETA and NGM models
had
differences regarding the strength, timing and position of a disturbance expected
to reach the south Oregon
coast. The AVN 12-hr forecast surface chart indicated that a developing surface
low near 38N 140W would move
east to near 135W by 12Z (fig.1). The AVN forecast
southerly winds of 40 to 45 knots in advance of the low
over the far northern California coastal waters. It had the low move across
the waters in the afternoon
before heading east to northeast across Siskiyou County. The ETA model forecast
was significantly weaker with
the surface low pressure and forecast only an open wave trough with wind speeds
of 15 to 25 knots in the
southern Oregon coastal waters.
It was determined
that the AVN model winds were more realistic than the ETA. This was based
upon the Goes-10
IR satellite derived winds below 700 mb which showed 50 knots of wind along
35N between 130W and 140W. As a
result, the forecast was tailored toward the AVN, but this was not deemed
to be sufficient data to wholly
follow the AVN solution. Less than 12 hours earlier, the AVN solution had
noticeably forecast coastal wind
speeds that were too strong, and the ETA model was deemed the better model
at the time. Nonetheless, windy
conditions were added to the forecast. This included mention of gusts to 40
mph along the Curry County coast
and gusts of 35 knots over the coastal waters south of Cape Blanco. Also,
wind gusts to 35 mph were included
for the Rogue Basin and 40 mph for the Siskiyou mountains and east side. It
was thought that the surface low
would track along the Oregon/California border with the strongest southerly
winds only affecting the South
Coast, Rogue Basin, Cascades and South Central Oregon. Coordination calls
were made to Eureka, Pendelton and
Boise specifically to highlight that the AVN may be handling the offshore
surface features better according
to the observed satellite-derived winds. Winds of 20 to 30 mph with gusts
to 40 mph in adjoining areas were
suggested.
Based upon the morning
briefing, coordination with the marine forecaster, and analysis of the differing
12Z
model solutions it was determined on the morning of the day shift that a compromise
between the AVN and ETA
was the most prudent course. Despite differences in strength, both models
were consistent and similar in
tracking a weak low east from near 40N 135W into western Oregon between 12Z
and 00Z (figs. 2a and 2b).
Thus,
the strongest winds were expected along the far northern California coast
and a forecast of breezy conditions
was issued along the Oregon coast with windy conditions at coastal headlands.
The 00Z and 12Z runs of the MAV
guidance both indicated sustained south winds of 30 knots for 00Z Thursday
afternoon at Brookings. A wind
advisory was issued for the Shasta Valley of Siskiyou county, California,
which often receives strong south
winds as strong surface lows move inland over the Pacific Northwest.
In the morning, south
winds along the southern Oregon coast had been below 20 knots. Around 11 am,
as the
rapidly developing surface low approached the coast, coastal wind speeds increased
to near what they had to
been forecast to be. Satellite features between 1030Z and 1930Z showed classic
cyclogenesis features
including a baraoclinic leaf, dry slot intrusion, and cusp formation, which
should have helped forecasters
identify the surface low position and rapid development (see
Satellite loop). The indication of a more
northerly storm track, growth of the dry slot behind the storm, and the fact
that the storm was not yet at
the coast were all indications that winds in the afternoon would be stronger
than forecast. However, wind
speeds were still not expected to be significantly stronger than forecast
and a decision was made not to
update the forecast given the rapid movement of the storm and the fact that
the forecast had been updated
only 15 minutes earlier. At around noon, wind gusts at Cape Blanco jumped
to 60 mph from 38 mph the previous
hour. At 1 pm, a peak gust of 84 mph was recorded at Cape Blanco. A Nowcast
was issued for the Curry County
coast at 107 pm followed by a High Wind Warning for the Coos and Curry County
coast at 116 pm. South winds
veered to west and began to decrease at the coast around 300 pm.
Conclusion:
This type of event
is rare but such an event is likely to occur again in Oregon. The most important
lesson is
that forecasters must be able to recognize the precursors of explosive marine
cyclogenesis. An explosive
cyclone is defined as one in which the surface pressure is falling at a rate
exceeding one Bergeron per day.
At 40N this would equate to a pressure fall of 18 mb or greater in 24 hours.
In their paper, "Explosive
Marine Cyclogenesis: Surface and Upper Air Indices", Lyons and Scoggins
note 5
features normally present prior to or during such an event:
1. A pre-existing
baroclinic zone with strong meridional temperature gradients and strong west
or southwest
winds from 500 mb upward to 200 mb.
These conditions were
present in this case. Note the low level thermal gradient over the eastern
pacific in
the 12Z AVN 850 mb chart ( fig.3 ).
2. A strong jet streak
west or northwest of a pre-existing surface cyclone accompanied by a strong
cyclonic
vorticity center and associated PVA into the area above the surface cyclone.
There was a strong
100 to 135 knot jet streak at 300 mb ( fig.4 ) and
a strong vorticity center at 500 mb (
fig.5 ) overtaking the low level thermal gradient.
3. A synoptic scale north to south oriented upper level trough becoming negatively tilted with time.
The SPENES message
of 1445Z from the satellite analysis branch of NESDIS noted the trough beginning
to take
on a negative tilt with a baroclinic leaf signature ahead of the vorticity
center (See Satellite
Loop ). The
upper trough became negatively tilted between 12z and 00z (figs. 6a
and 6b ).
4. Strong upper level
divergence to the east of the upper trough axis, and strong low level convergence
beneath the upper level divergence region.
The storm encountered
a dual jet stream structure along the Oregon coast which enhanced upper level
divergence in the vicinity of the developing surface low. The right rear quadrant
of a westerly jet streak
over Washington overlapped with the left front quadrant of the approaching
southwesterly jet streak offshore
( fig. 7 ).
5. Development of
strong convection in the vicinity of the surface cyclone center as the cyclone
begins to
develop.
Convection was not observed.
Lessons Learned:
1. Due primarily to
much improved computer model forecasts, forecasters have to ward off the reluctance
of
issuing a forecast that diverges strongly from the model data. Be vigilant
in determining when observed
conditions vary from the model solution.
2. Forecasters need
to remember that the model that verified well yesterday may not have a good
solution
today. Also, a consistent model forecast is not necessarily an accurate model
forecast.
3. As our relationship
with neighboring offices continues transition from coordination of written
forecasts
to the era of collaboration with GFE, communication between offices will take
on increased importance.
Occasionally, much stronger winds may be confined to one office's area. Smoothing
of adjacent grids is
possible without neighboring offices having an overall forecast of similar
conditions.
4. The forecast for
this event was hampered by a dearth of observed surface conditions in the
vicinity of the
developing surface low. There were no ships reports in the area, the wind
sensor at Buoy 06 (40.8N 137.5W,
just north of the low track) had been inoperative since October 2001, and
Buoy 59 (38N 130W, just south of
the low track) was out of service. Also, the storm made landfall nearly at
the center of the largest stretch
of the United States west coast without a near shore buoy. The satellite derived
winds (labeled as "below 700
mb") were perhaps the most important precursor to the rapid cyclone development.
The 40 kt southwest winds
derived in the vicinity ahead of the the developing surface low were much
stronger than model data indicated.
Also, skillful analysis of infrared and water vapor satellite imagery and
an accurate determination of how
conditions aloft will translate to the surface are both imperative.
5. Forecasters from
Pacific Northwest marine offices should maintain an annual review of the precursors
to
explosive cyclogenesis as outlined by Lyons and Scoggins.
6. In the operational
environment, identification of the problems of the day, and an awareness that
these may
change over the course of the day are vital in maintaining a constant weather
watch.