Transcript Slide 1

An Investigation of Fog
Patterns in the Lower
Columbia Basin of Oregon
and Washington
- Darren Van Cleave -
Motivation
The lower Columbia Basin of Oregon and Washington is routinely stricken
with persistent inversion-triggered fog in the winter months. Although this fog
poses a major traffic hazard to various freeways and airports in the area, no
thorough investigation of these fog tendencies has been performed. Thus, the
purpose of this study is to examine the various conditions that lead to fog
formation in that area.
Prior Research
This project is a continuation of a study begun in the summer of 2005 by Jon
Mittelstadt (Science and Operations Officer with the National Weather Service in
Pendleton, Oregon) and Darren Van Cleave. Unfortunately, this initial effort was
hampered by a lack of time and resources. As mentioned, there have been no
other fog studies pertaining to the Lower Columbia Basin. However, this research
is similar to studies conducted by Mark Burger1 of fog patterns in the San Joaquin
valley of California. This study uses concepts of pressure heights developed in the
Burger study, as well as fog persistence concepts used in a study by Jonathan
Slemmer2 of fog patterns in the Salt Lake City area. Lastly, this study uses
probability figures similar to a study by Wayne Shaffer3 of fog patterns in the
Knoxville, Tennessee area.
1. http://www.wrh.noaa.gov/wrh/03TAs/0309/index.html
2. http://www.wrh.noaa.gov/wrh/04TAs/0401/SLCFog.htm
3. http://www.srh.noaa.gov/mrx/research/Fog/fogTYS.htm
Data
- ASOS
Surface observations were taken from ASOS (Automatic Surface Observing
Systems) records dating back to 1998 for three cities in the Lower Columbia
Basin: Pendleton (Oregon), Pasco (Washington), and Walla Walla (Washington).
All of these ASOS stations are situated at airports. Because large-scale fog in
the Lower Columbia Basin is experienced primarily from November through the
first week of February, the data set was limited to those days. Special
observations were eliminated, and the hourly observations were used
exclusively.
- NCEP
Readings of 700 millibar pressure heights were taken from NCEP (National
Centers for Environmental Prediction) reanalysis data. The heights were derived
from a geographical point near Pasco, which gave a solid representation for the
Lower Columbia Basin.
Methodology
This study utilized exploratory data analysis in each aspect of the research.
For simplicity, a fog event was defined as any time in which an ASOS station
reported fog without reference to any precipitation (for example, a report of
fog/rain would be ignored). Persistence figures were calculated for the three
stations. Probabilities of several meteorological aspects were calculated for time
periods prior to fog events, and the results were expressed in pie graphs. Lastly,
graphs of 700 millibar pressure heights versus time were rendered with fog
events noted at the respective times.
For some of the figures, composites were used in place of individual station
readings (for the pressure-heights figures, composites of the three sites were
developed by noting times in which fog occurred at all three stations
simultaneously). The use of composites was justified given the fact that all three
sites have meteorological similarity, and also because large-scale fog events will
most often generate fog at all three locations. Additionally, the goal was to
obtain a representation of the Lower Columbia Basin as a whole, and not just
the individual sites.
Methodology – Data Problems
For reasons unknown, the ASOS data archives had occasional periods of
missing data. On average, the sites had 65 missing observations, which is only
0.33 of one percent of the entire length of over 19,000 observations since 1998.
Unfortunately, the algorithms for several of the exploratory data analyses
demanded observations for every hour. To compensate, missing data points
were filled in with observations from the previous hour (a special thanks to
Kathy LeBlanc for assistance in this matter). If the gap was two or more hours,
data from before and after the gap was used. A more statistically sound means
of compensation such as a Kalman filter would have been preferable. However,
this was not practical since the observations included discrete events such as
precipitation and fog formation.
Example of Data Problems:
Missing 3 Hours
Fog Persistence in Pasco
Pasco has the least fog of the three cities due to its
low elevation. As an inversion in the Lower
Columbia Basin strengthens, the fog bank tends to
rise, leaving areas which are nearest to the
Columbia River under a stratus deck. This deck of
low clouds enshrouds the nearby hills and creates
problems for roads which traverse them. Thus, while
the Pasco airport may not register as much fog as
the other cities, it is still affected by prolonged
events.
Fog Persistence in Walla Walla
Walla Walla experiences more fog events as well as
longer event lengths than the two other locations.
This is largely a result of the surrounding
topography. The city lies in a relatively flat area
directly on the edge of the Blue Mountains, with the
flat Columbia Basin on the other side. The lack of
surrounding hills causes the fog deck to remain
nearer to the ground for longer periods of time. Only
under very strong inversions will the fog deck
ascend the Blue Mountains. Therefore, Walla Walla
has less occurrences of low-cloud fog events than
Pasco, and thereby experiences lengthier surface
fog events.
Fog Persistence in Pendleton
The characteristics of fog events in Pendleton lie
somewhere between those of Pasco and Walla
Walla. It is interesting to note that the rate of
decrease in the length of events in Pendleton slows
considerably after 10 hour events, and this holds
until the 16 hour events are reached. Thus, a
persistence forecast seems plausible after a fog
event has lasted 10 hours. However, the
effectiveness of that conclusion is questionable
given the small number of events that last over 10
hours.
Wind Probabilities
Fog in the Lower Columbia Basin requires
weak mixing (light winds) in the twelve
hours prior to its formation. These pie
charts depict the average wind speeds for
twelve hours prior to all fog events which
lasted at least six hours. When the three
sites are averaged, light winds (between
two and 6 miles-per-hour) are present
75.6% of the time.
Pendleton
Walla Walla
Pasco
Wind Direction Probabilities
This figure depicts fog events which
lasted eight or more hours. As
evidenced in the figure, fog
development is dominated by
northerly winds (95.1% of the time)
in the twelve hours prior to an
event. This is due to advection of
fog from the north. At the onset of
an event, fog forms closest to the
Columbia River (in the center of the
basin) and prevailing northwesterly
winds advect it towards the Blue
Mountains. If the fog bank can
reach the foothill canyons of the
Blue Mountains, the diurnal
upslope winds quickly carry it up
the canyons.
Precipitation Probabilities
Fog formation in the Lower Columbia
Basin often requires at least one
night of clear skies to allow for strong
radiational cooling of the surface.
This composite charts shows the
most recent precipitation for all fog
events. For the majority of the events
(69.5%), there was no precipitation
recorded in the previous twenty-four
hours. A much smaller portion
(18.3%) of the events had
precipitation in the past twelve hours,
and even less (12.3%) experienced
precipitation in the previous twentyfour hours.
Fog Formation (Visible Satellite vs. 700 Millibar Height)
These figures show the relationship between 700 millibar heights and fog formation
11/27/02
11/28/02
11/29/02
Fog Formation vs. 700 Millibar Height
This figure plots times in which all three stations
reported fog onto 700 millibar pressure heights for
the same period (the heights are detrended).
Note that the vast
majority of the
events occur at or
near peaks in the
pressure heights.
This clearly
demonstrates the
relationship
between 700
millibar pressure
heights and fog
formation.
Fog Formation vs. 700 Millibar Height
This figure shows fog events in Pendleton plotted against the 700 millibar
heights. In this figure, only the events which last greater than five hours and
have no precipitation in the previous twenty-four hours are shown.
Despite the consistent
association between
pressure heights and
fog formation, pressure
heights are difficult to
use as a predictive tool
due to the discrete
nature of fog formation.
One possibility not
covered in this study is
to use pressure heights
to predict the length or
intensity of fog events.
Conclusions
-Persistence
It is obvious that Walla Walla receives much more fog than Pendleton and
Pasco, both in terms of the number of events and the length of events. For
Walla Walla and Pendleton, there is a period between ten and sixteen hours of
fog event length in which the probability of fog lasting for another hour
significantly increases, and so a persistence forecast may be possible for those
times.
- Probabilities
For the Lower Columbia Basin as a whole, light winds between two and six
miles-per-hour (75.6% of the time) are needed to enable weak mixing within the
twenty-four hours prior to a fog event. Northerly breezes are necessary (95.1%
of the time), usually to advect the fog into an area from the north and northwest.
Dry conditions within the twenty-four hours prior to fog events prevail (69.5% of
the time).
- Pressure Heights
Pressure heights of 700 millibars are clearly linked to fog formation in the
Lower Columbia Basin. However, their usefulness as a predictive tool is doubtful
for three reasons: the discrete nature of fog, the occurrence of anomalous
events (not all major events occur during a pressure ridge), and the fact that the
existence of a stratus deck is not taken into consideration.