ED O'LENIC, NATIONAL WEATHER SERVICE
February 2, 1996
 On Groundhog's Day, Ed O'Lenic, a long-term seasonal forecaster for the National Weather Service,answers questions about the harsh weather most of the U.S. has been experiencing. Typically Mr. O'Lenic predicts weather patterns
at least three months in advance.
 Click here for Friday's segment on the Deep Freeze or for the transcript on the "Blizzard of '96"
Click to see a Forum Menu.
A question from Mike Mocklar of Asheville, Tennessee:
The Farmer's Almanac has been around for years, and has
made many accurate predictions. Do you utilize any of the same
techniques?
Ed O'Lenic responds:
The prediction techniques used by the venerable Farmer's
Almanac have never been published in a scientific journal, so its
hard to say whether or not we use any of their techniques. We are
open-minded about forecast techniques to the extent that we will
consider using any forecast technique which has a known, documented
record of useable forecast skill over at least 20 or more years of
actual or simulated forecasts. Each of the three techniques we use
was required to have such accuracy before they were put into use
in our operation.
By "useable", I mean that, if we are forecasting
only two classes of temperature or precipitation (either ABOVE or
BELOW) at a certain station, say Nashville, and we have a total of
100 forecasts for that station, the technique must predict the correct
sign of temperature or precipitation (either ABOVE or BELOW) at
least 57 times, on average.
The techniques we use include one which identifies trends in
temperature and precipitation over the last 10 and 15 years, respectively.
The second searches through nearly 5 decades of seasonally
averaged observations of U.S. surface temperature, global ocean surface
temperature and flow-patterns (alot like jet-stream patterns) looking
for relationships that have useable prediction skill.
The third method uses a mathematical model of the ocean and
atmosphere which can simulate to a useable degree of accuracy at
certain places and times, the real atmosphere-ocean system.
In the October 1981 issue of Weatherwise, pages 212-215,
John E. Walsh and David Allen performed a check on the accuracy of 60
monthly forecasts of temperature and precipitation from the Old
Farmer's Almanac at 32 stations in the U.S. They found that 50.7
percent of the monthly temperature forecasts and 51.9 percent of the
precipitation forecasts verified with the correct sign. These may
be compared with the 50 percent success rate expected by chance.
All of our forecasts, which we update each month near the 15th,
are available on our homepage at
http://nic.fb4.noaa.gov
The homepage also contains a wealth of other information.
A question from Barbara Silver of Sandusky, Ohio:
I have often heard meteorologists making exceptional long range
forecasts. How accurate can these really be?
Ed O'Lenic responds:
As I implied in my answer to question 1, one can compare
any forecast with a similar forecast made at random. On average,
over many forecasts, the random forecasts will be correct 50 percent
of the time and will be wrong 50 percent of the time. We feel
that, as a minimum, an accuracy level of 57 percent is required for
a forecast technique to be useable in some sense. Such a low level
of minimum accuracy means that users of such forecasts must be able
to take a long-term approach to planning and investment decisions
made using the forecasts. Clearly, this means that one should not
use seasonal forecasts to plan a wedding 6 months in the future,
for example.
In general, the accuracy of our forecast tools varies regionally,
seasonally and by whether we are forecasting temperature or precipitation.
Winter and spring forecasts tend to have highest accuracy. Autumn the
lowest and the remaining seasons lie in between.
A question from Laurie Singer, San Diego, California:
Why is it harder to predict weather conditions in coastal regions
than inland?
Ed O'Lenic responds:
Coastal boundaries are also boundaries of moisture and temperature.
This is part of the reason why there is a "sea-land" breeze blowing from
ocean to land during the daytime and a "land-sea" breeze in the opposite
sense at night. Such gradients also constitute a source of energy for
atmospheric disturbances. The coastal snow storms in which develop
near Cape Hatteras in the winter are an example.
On longer time scales, regions near coasts are actually easier
to forecast temperature for because they are strongly affected by the
temperature of the ocean. Since the ocean warms and cools much more
slowly than the air, it has a longer "memory" and can have a considerable
effect on proximate land areas. Thus, if the ocean off San Diego is
warmer than "normal" (e.g. the 30-year average ocean temperature at that
location), the land near the ocean will be similarly affected.
A question from Deborah Galton of Freeport, Maine:
What's ahead for the next three months for different regions of the US?
Our forecast for March through May, 1996 calls for temperatures
to average above normal from the Rockies westward, below normal over the
Great Lakes and over coastal sections of Mississippi and Alabama. Else-
where, there is no strong preference for either ABOVE or BELOW normal
temperatures. This is also the case for precipitation for the entire
nation during this season.
----------------------------------------------------------------------------------
ADDITIONAL THOUGHTS
Frederick Bartlett of Mercerville, New Jersey:
Given that predictions accurate to, say, 2 decimal points are
not possible for more than a few hours (minutes?) ahead,
why do TV weatherfolk insist on giving us 5-day forecasts?
Taking the longer view, since it seems that there is a cyclic
variation in climate patterns on a scale of centuries, the
mechanism behind which is far from fully known, how sure can
we be that it will be warmer in a decade or a century?
----------------------------------------------------------------------------------
Marjorie Parent Greenman of Ann Arbor, Michigan:
Since most weather records cover only the last 150--200 years, it seems to
me that trying to predict a serious change in the weather is like trying to
predict what the weather will be next week or on June 27, 1997. We
can assume that while it might be cold, it more likely will be pleasantly warm.
We can say "Welllllll, based on computer modeling it appears that we are
having hotter summers and colder winters, so we'd better prepare ourselves."
But since a computer, like we mere mortals, can only work with the data
available, and the amount of data we have is practically nil when we consider
the age of the planet, isn't it possible, and likely that we are simply
entering a new cycle? (Maybe our great greats really did walk to school
through three feet of snow.)
If our summers are getting hotter and our winters
colder do you envision a more aggressive approach to research in nuclear
power (fusion vs. fission), horticulture and animal husbandry to
meet the challenge? And, even if it can only be said that the past year or
two have been aberrations wouldn't there be value in stretching the point
a bit in order to instigate the research above?
-----------------------------------------------------------------------------------------
Robert Blinder of Short Hills, New Jersey:
The last two of the past three winters have been the harshest
within the last half century. Is there any correlation
between these winters and the highly reported theme vis-à-vis
global warming? I have read reports that the cold winters
prove that the earth is getting warmer. I have also seen
the Global Warming crowd say that the warm winter of 1995
proved their theory. Which is it? I would tend to believe
the vast majority of respected scientists who say the whole
theory re: warming is a bunch of melted icicle.
--------------------------------------------------------------
Michelle Wetstein of Columbia, Maryland:
The limits of weather computer models:
What methods are you using to overcome one of the big problems
of long-range forecasting for any chaotic system: extreme
sensitivity to initial conditions? I thought that forecasting
the weather beyond four or five days with any accuracy was
next to impossible. I realize that forecasting climate is not
quite the same as forecasting day-to-day conditions, but don't
the same problems occur here?
Past climate records:
Is it true that the earth's climate went through severe
fluctuations in temperature up until recently (i.e. up until
about 10,000 years ago)? Are we in a relatively calm period
of climate compared to this past record?
-----------------------------------------
Alice Dominguez of Sunnyvale, California:
A question of history
Do we know that trends of weather are measurable in decades and not in
centuries? How far back do we really have accurate records? Can we tell
by looking at our recent data how the trends of the past are changing?
The weather may be changing day to day or year to year, but what about
decade to decade? How do we know our information is not partial information?
-----------------------------------------
Joseph McDonald of Burlington, Vermont:
Can we control the weather?
Weather forecasting has improved dramatically over the last
15 years, especially in the 3 to 6 day forecasts. This year
in the Northeast we received warnings several days in
advance of major snow storms, which really helped us prepare
for them.
My question has to do with controlling the weather. Do you
think there will be much progress in the next 20 years
in controlling, or moderating, destructive weather events,
such as droughts, floods, hurricanes, and tornados. Are
any federal or private agencies doing work in this area?
-----------------------------------------
Dennis Bell of Seattle, Washington:
I, for one, would have more confidence in medium and
long range forecasts if, along with the familiar charts
of weather trends, we were shown similar charts displaying
exactly how the weather over the most recent like periods
actually happened. My gut feeling is that most of the long
range forecasts are skewed toward persistance, and that
the further out one looks, the closer one approaches the
inevitable coin-toss. Boiled down to a single line:
Just exactly how good are these long-range forecasts?
-----------------------------------------
Rick Sigrist of LaRue, Ohio:
This past year has seen violent weather across the country.
Is this trend going to continue in 1996? Also, there does not
seem to be much improvement in long range forecasting
beyond a week. Is there any hope for accurate 3 month trends
predicting for the future? | 
