Winter 2003-2004 Outlook

Mother Nature is Reluctant to Tip Her Hand About the Winter of 2003-04

Written by: William R. Deedler, Weather Historian
National Weather Service Detroit/Pontiac, MI
November 15th, 2003

Quick jumps on this page:

Latest indications from the Climate Prediction Center /CPC/ show that for the Pacific Sea Surface Temperatures /SST/ will hover from normal (or in the neutral state) to slightly above normal (or a very weak El Niño). Various computer models have now converged on this predicted pattern. The CPC also included all the Great Lakes in equal chances /EC/ of below normal, normal or above normal temperatures and precipitation (including snowfall) for the upcoming winter. This variance of possibilities for the upcoming winter is well reflected in the 13 chosen winters in this year's winter outlook. The CPC stated the following in regard to the reasons for the uncertainty in the winter outlook for 2003-04:

"Without a strong El Niño or La Niña signal as a guide, there is more uncertainty in our forecast, but we do expect winter to bring its typical weather variability," said John Jones, Jr., Deputy Director of the NOAA National Weather Service. He added that forecasters, in cases where a dominant climate feature in the Pacific Ocean is missing, rely on historical trends of temperature and precipitation averages as well as dynamical and statistical models".

Local Study Findings


Our earlier autumn outlook called for considerable temperature swings (a.k.a. roller-coaster pattern) to materialize this fall. It is felt that the roller-coaster pattern will carry over into the winter, though not being quite so frequent. This is quite a change from last winter, where weeks of what seemed like never-ending cold prevailed. Therefore, the temperature pattern of the Winter of 2003-04 is expected to be more variable with ups and downs. Periods of cold will be offset by occasional periods of above (or even well above) normal temperatures. That being said, overall temperatures should still average near normal to below. The in-house study suggests the coldest weather of the winter will likely go along with climatology with January and February having the best chances (there were only a few cold Decembers in the data).


As evidenced by the 13 winters in the study, snowfall amounts were quite variable (more so than any preceding study), with snowfall ranges from as little as near eight inches (Saginaw's 7.8" during the Winter of 1941-42) to a whopping 87" /87.2/ at Saginaw in the Winter of 1966-67. This, in itself, displays no persistent discernable pattern for snowfall. However, there were other more subtle trends noted while looking back over the available years and recent patterns. While the average for snowfall in the research came out near normal, the number of "snowy" winters did outnumbered the "snowless" winters. It was noted that when the winters contained below normal snow, they also tended to be the driest overall. The study hints that (and this too, would be unlike last winter) the heaviest snow may fall from Detroit's northern suburbs, northward up into the Flint and possibly the Saginaw area, rather than from Detroit south to the Ohio border. Overall, most areas should see close to normal snowfall with locally above normal.

Local Research

Along with predominant trends of an influencing El Niño, La Niña or the relatively non-influence of a Neutral pattern, local historical trends of temperature and precipitation along with current meteorological models are also used locally at the NWS in White Lake. Also, past upper wind patterns of the specific winters in the study are examined (when available) along with the relatively recent patterns of the last year or so. In addition, the Northern Atlantic Oscillation /NAO/ (including the Arctic Oscillation /AO/) and the Eastern Pacific Oscillation /EPO/ of the past winters in the study (back through 1950) are studied and also compared to their recent trends for any pattern similarity and recognition. Finally, trends of the current sun spot cycle (including intensity, position and trend in the cycle) are compared to the years of the researched winters.

Previous Summer-Fall Periods

The last time a neutral state occurred - where any specific Pacific Ocean weather pattern (la Niña or El Niño) is pretty much absent - was just a few winters ago, during the Winter of 2000-01. As was noted back in the 2000-01 Winter Outlook (and again in the recent Autumn Outlook), the majority of neutral autumns tended to contain aggressive early cold snaps which created frosts and freezes. At the same time, however, it was also noted there was still the likelihood of some mild Indian Summer interspersed among these cold snaps. This again (like the Autumn of 2000), has appeared this fall but was not the case last fall (2002), when mid October pretty much became the "dividing line" between summer and winter. Another interesting item noted and seen during both falls has been the marked fluctuations in temperatures (more than usual, sometimes every few days). One just has to think about this past fall to get an idea of the pattern.

The dominant weather patterns of the past year or so, to a large extent, continues this Autumn. Overall, temperatures have averaged almost consistently around normal to below while our two main storm tracks have held remarkably persistent across the Northern Ohio Valley/Southern Great Lakes while the other is further north, across the Northern Great Lakes/Southern Canada. Recent changes have been noted lately as an active and strong clipper track has materialized over the northern US. At the same time, some relatively strong high pressure systems while bringing powerful chills, also have brought interludes of unseasonably mild weather with a return flow from the south. Finally, the Pacific Jet has recently grown quite active, occasionally bringing in mild Pacific air well inland. This also may be a reflection of the weak El Niño state developing over the Pacific.

As stated in the past several outlooks, some distinct pattern traits of the past year have resembled weather patterns of some seasons in the 1970s into the early 1980s (especially the winters). During the past year or so, some major patterns resembled that of 1973-74 and 1977-78. Trends of this past summer into the fall continue to show similarities to both years. After a generally cooler and drier than normal summer in 1978 (and really, for a good part of the region this past summer, except south of Detroit), this autumn's weather, with its frequent temperature gyrations (due to an amplified and progressive upper air pattern) brings both autumns ('74,'78) and some other past falls in the study into view. The impressive early cold outbreaks this fall have also resembled that of 1974. Ironically, both contained an impressive cold blast at October's opening which brought some flurries to portions of the region.

Autumn Gyrations - is the trend our friend ?

As mentioned above, this falls's weather, thus far, has contained what was anticipated in our fall outlook. The aggressive early cold blasts (producing frost/freezes), followed by strong warm ups creating delightful Indian summer weather, were all the result of the very amplified and progressive (or " roller-coaster") upper air pattern. For the most part, fall temperature patterns are generally erratic by nature with all the up and down temperatures as seasons change. Some falls in the study (including this fall) have displayed even quicker temperature swings, reflecting the overall progressive, somewhat chaotic upper air pattern. Therefore, it was theorized that checking out the other similar autumns in the study (those with fully amplified and progressive upper air patterns), might provide us with a few clues as to the upcoming winter patterns. So then, which falls in our study contained similar exaggerated "gyrating" patterns and did they persist into the winter? A perusal back through the last century and the 13 autumns and winters in our study, lead to the following outcome. The Falls of 1919, 1932, 1952, 1966, 1974, 1978 all contained similar "roller-coaster" patterns. When one looks at the subsequent winters, however, there were still various outcomes. While the amplification of the upper air pattern may not have changed much into the winter, its dominant placement during those winters may very well have. In other words, while the preceding autumns may well have been "topsy-turvy" (due to a progressive amplified pattern), the placement of the upper air trough during the winter may have settled down, occasionally, over a particular region. This in turn, dictated our subsequent winter weather, whether it be mainly mild, cold or stormy. Therefore, it stands to reason that both the amplified pattern and dominant placement of that pattern (troughs and ridges) has more of an effect on our expected winter weather, especially in a neutral or very weak Pacific patterns. In some winters, the roller-coaster upper air pattern persisted with little dominant placement of the upper air patterns. Therefore, it is easy to see why this variability can be problematic when trying to pick out a trend, but then again, variability too is a trend! Our earlier preliminary outlook (issued October 16th, 2003) and current fall trends continue to point that way. While on the subject of upper air trends, let's check out past trends and current trend of the NAO and EPO.

Mixed NAO and EPO Patterns this Autumn Resemble Few in the Study

Latest indications show that the NAO cycle has been decidedly more negative than positive since mid summer. This negative trend is different than last fall when the NAO was primarily neutral or positive. This increasingly negative pattern has helped set the stage for the strong cold blasts experienced so far this fall. When the negative pattern relaxes or shifts east, the dominant positive EPO has helped warm things up.


While the NAO has favored the negative side, the EPO up until November has favored positive territory - a change beginning? Basically the EPO pattern has been almost completely opposite of last fall when each negative dip of the EPO became stronger and stronger with time. This is one of the primary reasons why the weather this fall has been mixed as far as warm and cold blasts while last fall, the cold dominated under a decidedly negative EPO from October into winter.

Therefore, generally speaking, the two oscillations (NAO/EPO) have been running opposite to their last fall trend, the NAO has been more negative, while the EPO has been more positive. Unfortunately, the major oscillations are not predictable for more than a week or two. Still, checking back over the available oscillations (back to 1950) for the fall and early winters may bring some similarities to recent trends. Therefore, let's see if any fall oscillation periods (Sep-Dec) in the 10 available falls (after 1950) resemble our current trend.

I was hard pressed to find a semblance to our recent past this fall (Sep-Nov) for both the NAO and EPO in the same year. The closest for trend sake (meaning similar positive/negative phases but not necessarily the same timing of those phases) were the falls of 1966, 1977 and 1980 for both the NAO and EPO. During both falls, the NAO cycle was primarily negative while the EPO was primarily positive. Interesting enough, during all three winters, a very weak El Niño also prevailed and the EPO started more negative phases late in the fall (not unlike this fall so far). In the Winter of 1966-67, temperatures averaged near normal but it was a stormy winter (the major January Blizzard of 1967 comes to mind, for instance). The Winter of 1977-78 was a cold and stormy winter (this winter was in last year's list also since the Pacific water temperatures floated around normal during much of the late 1970s). In the. Winter of 1980-81, temperatures also averaged below normal (but not as much as 1977-78) while snowfall averaged somewhat below normal.

November Indications?

Though the weather patterns tended to be quite variable in the researched seasons, a somewhat more noteworthy trend did surface while studying the Novembers (Fig -1) in the research. While November's average temperatures were quite variable, nearly all the Novembers had measurable snow. While this in itself is not unusual since measurable snow in November is normal, what is unusual is that well over half (or 8 out 13 - 62%) contained above normal snowfall (and some of those were quite exceptional). This does give a bit more credence to the likelihood of an early measurable snow despite the occasional Indian Summer-like weather popping in from time to time. Take note of the November snowfalls preceding the winters in the study.

                           Fig - 1

MONTH/YR              SNOWFALL          NORMAL (1971-2000)
                                        Nov at Detroit 2.7"  
NOV 1900                 4.2" *                  
NOV 1919                 0.3"  
NOV 1932                10.0" *
NOV 1941                 0.5" 
NOV 1952                 0.9"          (0.1" IN OCT) 
NOV 1958                 4.4" * 
NOV 1966                11.8" *
NOV 1974                 7.7" *
NOV 1977                 7.4" *
NOV 1978                 6.1" *
NOV 1980                 3.4" *        (2.9" IN OCT) *
NOV 1992                 0.9"          (0.4" IN OCT)
NOV 2000                 1.3"     
* Above Normal Snowfall                (Measurable snow in OCT)


The following winters were researched containing basically Neutral to Weak El Niño conditions:

                            Fig - 2

Winter           Average Temperature                Snowfall Year
           Detroit    Flint      Saginaw        Detroit   Flint    Saginaw

30 Year
Normals-   27.1       23.9       24.0           44.0      48.3       44.5 
100 Year
Average-   26.5        -          -              -         -          -

1900-01    23.6       21.0*      22.5           41.7       -         54.3
1919-20    20.9 (9)   16.3*      16.7 (4)       43.5       -         41.2
1932-33    30.5-16-   30.2*      28.7 -8-       25.9      18.5*      24.5 -12-
1941-42    28.0       25.7       25.1           23.4-17-  26.9  -6    7.8  -1-   
1952-53    32.3 -8-   28.6 -4-   28.3-10-       16.6 -7-  23.4  -4-  20.0  -3-
1958-59    23.1(18)   17.9 (3)   18.3 (7)       37.2      61.5  (8)  44.3
1966-67    27.3       23.4       23.7           50.6      78.6  (2)  87.2  (1)
1974-75    28.1       28.1 -7-   27.8-12-       63.1 (7)  82.9  (1)  45.4
1977-78    20.4 (7)   19.1 (5)   17.9 (6)       61.7 (8)  50.6 (20)  55.6 (20)
1978-79    21.3(10)   18.4 (4)   17.4 (5)       35.6      52.0 (19)  55.9 (19)
1980-81    24.6       23.6       21.9           38.4      36.4 -20-  41.4 
1992-93    28.9       26.1       25.0           52.2      53.2 (16)  47.7  
2000-01    25.0       22.9(18)   22.5           39.0      53.4 (14)  67.8 ( 7) 

Averages:  25.7       23.2       22.8           40.7      48.9 **    45.6
(  ) Coldest or Snowiest Ranking 
-  - Warmest or Snowless Ranking
*Not an Official Record and not in ranking (Official Records Began in 1942)
**Sample Incomplete and Therefore Not a reliable Snow Average

Grasping Winter's Hand

A look at all the information mentioned above indicates out three main possibilities for the upcoming winter (and reflects the variability well). 1- An overall stormy winter (above normal snow) but with temperatures averaging around normal. 2- A cold winter (below normal temps) with normal to below snowfall, 3- A mild winter (above normal temperatures) and below normal snow. Now that this is spelled out, which way does the in-house study favor?

The above scenarios are mainly a result of the previously mentioned dominant placement of the upper wind trough (500 MB). Looking at the three winter possibilities, one may be able to visualize the dominant placing of this trough. Should the dominant trough settle out west over the western US, this would lead to more ridging (or zonal flow) in the east and thus, a milder winter with less snow. Should the trough set up shop mainly over the Midwest and Ohio Valley (and similar to last year), expect another stormy winter with snow, freezing rain, sleet and even thunderstorms as temperatures continue to fluctuate though still averaging near normal. One may be able to visualize the upper air trough placement during the stormy winters in the study. During the stormy winter's of '66-67 and '77-78, the main upper trough had a decided preference for the Midwest and/or the East Coast, shifting back and forth. Lastly, if the dominant trough parks over and east of the Great Lakes, a cold winter would prevail but most likely with less synoptic snow but more lake effect snow.

While on the subject of 1966, interestingly enough, a very strong trough developed during the first week of November over the Midwest/Ohio Valley (similar to the one seen late this October). A deep low pressure system developed and moved northward through the Ohio Valley bringing heavy rain and snow to the region. At the same time, strong Santa Ana winds developed out West under high pressure and buffeted southern California as temperatures surged up to 101'F in Los Angeles. Sound familiar? The storm in late October 2003, unlike in 1966, did not deepen enough nor contain enough cold air to bring snow to the region but strong Santa Ana winds did materialize under the strong high pressure ridge out West and helped bring the devastating fires to that region.

When looking at the overall temperature and snowfall averages in the study, temperatures average below normal, while snowfall is near normal. A majority of the winters contained below normal temperatures (7 below normal, with some of these winters placing in the top 20 coldest). If one considers the predominant trend of the NAO and EPO so far this fall and matched seasons (mentioned just above), this implies temperatures averaging normal to below and snowfall near normal to above. Latest Pacific SST's have shown a bit more warming to a border-line weak El Niño and this trend could hold through the winter. In addition, recently added as another interesting piece to this puzzle was the occurrence of a very strong solar flare display observed at the tail end of October. Some scientists theorized that strong solar flare storms may temporarily alter the upper air pattern over the globe. This may have contributed to the mild Winter of 2001-02 after a second strong solar flare peaked late December 2001.

Storm Tracks:

It is believed that split flow upper air patterns will play a more important roll this winter than last winter. The Pacific jet stream this fall remains strong and active, occasionally phasing with the northern Polar or Arctic Jet, resulting in strong low pressure systems and cold outbreaks. How strong (and split) the Pacific jet remains will also dictate the cold severity of our winter. A strong split flow would tend to temper cold outbreaks routinely. This also brings up the EPO mentioned earlier, and its dominant positive phase this fall, which in turn, does not favor the building of an upper level ridge on the West Coast.

There were primarily two storm tracks that affected the Great Lakes during the researched winters. They were the Alberta Clippers and the Colorado (or Texas Panhandle) Lows. The clippers ushered in Arctic and also modified Pacific air along with light to moderate amounts of snow and rain. Colorado (or Texas Panhandle) Lows generally deepen and intensify as they track through the Southern Plains, heading northeast through the Midwest and into the Great Lakes. Frequently, heavy amounts of snow, rain and mixed precipitation accompanied these storms. It should be mentioned that a third track, the Ohio Valley Low, which was very active last winter (in fact, much of the past year) also remained active in a few winters.


The temperature pattern of the Winter of 2003-04 is expected to be more variable than last winter with frequent ups and downs. That being said, overall temperatures should still average near normal to below.

While the average for snowfall in the research came out near normal, the number of "snowy" winters did outnumbered the "snowless ". The study hints that (and this too, would be unlike last winter) the heaviest snow may fall from Detroit's northern suburbs, northward up into the Flint and possibly the Saginaw area, rather than from Detroit south to the Ohio border. Overall, most areas should see close to normal snowfall with locally above normal. We shall see.

Look for a mid-winter update in late January is the U.S. government's official web portal to all federal, state and local government web resources and services.