Understanding Alaska's Climate Variation

John Papineau, Ph.D

NWS Anchorage, Alaska



Introduction

The climate of Alaska (i.e. long-term weather patterns) is a product of a number of different external and internal 'forcings'. Climate varies over time as these forcings change (often small changes). The primary external forcings are: solar output, atmospheric gases (volcanic eruptions, CO2), the temperature of the water in the Pacific Ocean, as well as the strengthening or weakening of ocean currents. Changes in one or several of these forcings typically causes changes within the atmosphere as well (internal forcings), like the repositioning of the polar jet stream and the Aleutian low pressure system or the frequency of La Nina's and El Nino's. We will look at some of these forcings in some detail later in this study.

Because of the unsteady nature and interactions in the aforementioned climate forcings, significant changes in temperature and precipitation (rain and snow) can occur in Alaska from one month to the next, from season-to-season, from year-to-year and even from one decade to the next. Figure 1 shows winter temperatures for the past 45 years at Yakutat, Anchorage, and Fairbanks (three different climate zones). Note the variation in temperature from one winter to the next. Overall there is no easily recognizable pattern that emerges from these plots, in fact, there are times when the trends at these three stations do not even match (i.e. they are out of phase). The main feature that is apparent from these graphs (and other stations as well), is the abrupt warming that occurred in the late 1970s (horizontal green line represents the average temperature).

It is these variations that make climate forecasting a challenge to say the least. There are times however when some type of grouping or clustering of temperatures (or precipitation) occurs. If we can associate, or what in statistics is called 'correlate', these groupings with a one or more climate forcing(s), then we might have a hope in at least forecasting some type of climate trend. Before proceeding in this study we need to consider how weather and climate data is collected and analyzed.



Data Considerations

Observations of temperature (air) and precipitation form the foundation of climate data across the globe. Additional climate data consists of sea surface temperatures (SST's), soil temperature, amount of snow cover or ice cover at a certain time, wind speed and direction, etc. Currently temperature and precipitation data is collected at several hundred locations across Alaska. This data is collected daily or in some cases hourly. It is from these observations that climate statistics like; mean (average) monthly temperature or mean monthly precipitation are calculated. Monthly data then of course can be aggregated to form seasonal or yearly values as well.

One important consideration in climate research is the number of stations that collect data and the length of record of those station. In Alaska there are a few sites that have sporadic data dating back to the late 19th century (St. Paul, Nome). However, most stations in our state brought into service between 1930 and 1955. This means that we do not have a very long record to work with. In addition, in order to analyze climate we need a continuous record for the longest possible period of time from as many stations as possible. The most reliable and data rich period for Alaska runs from the early 1950's through the present. There are as noted above a handful of stations that have longer periods of record, which can be used for specific types of analysis. In summary, for it's size Alaska is a relative data poor region, however there is enough data to at least make some estimates concerning short-term climate variation.

An additional consideration is the movement and re-location of a particular climate station within a city or town. Let's use Anchorage as an example. The official weather station for the city was established in 1915 and was located near the lower end of Ship Creek. Between 1923-1943 the station was re-located several times in what is now the downtown area. From 1943-1953 the station was located at Merrill Field, but from 1953 to the present it has been sited at Ted Stevens International Airport. These station re-locations are important because as everyone who lives in Anchorage knows, winter temperatures (and precipitation) vary considerably from west-to-east and north-to-south across the city. These spatial variations in the data have to be taken into consideration during climate analysis, otherwise they might indicate a change in climate which in reality is due to the re-location of the equipment.

There is no set methodology to analyze climate data, researchers may use different techniques. In this study I have chosen to work with mean monthly values as well as seasonal values (i.e.- November through March and April through October). The first question that has to be answered is: what constitutes a significant deviation (what we will call an 'anomaly') from 'normal' conditions? Temperatures fluctuate at virtually all timescales, day-to-day, week-to-week, etc. Typically we calculate the mean (synonymous with 'average') for a period of time that is of interest to us. For example, the mean monthly January temperature at Fairbanks for the years 1930-2001 is: -10.1o F. Note that the monthly mean is calculated by using the daily means, the daily means are typically calculated using the daily maximum and minimum temperatures at a particular climate station (at automated weather stations the average of the 24 hourly observations is used). Monthly precipitation on the other hand is the total precipitation (rainfall and melted snowfall) that occurred during the month. The mean monthly precipitation is the average value that occurs over many years.

Let's return to the Fairbanks temperature example. If the mean temperature for a given January is -6.3o F, what does this tell us about temperatures during that month and is this value significant? First, -6.3o F minus -10.1o F = +3.8o F, which indicates that it is was a little warmer than 'normal'. Where normal is defined as the long-term mean (in this example -10.1o F). Now we have to ask the question: is this temperature difference significant? The answer is no. Fairbanks residents might expect a slight decrease in their heating bills for that particular month, but not much. The reason that this difference is not significant is based on an additional statistic called the 'standard deviation', which simply measures the range of the data set. For example, the January standard deviation at Fairbanks is 10.2o F. If we add and subtract this number to the monthly mean of -10.1o F, then we know from statistical theory that two-thirds of all the monthly January temperatures will fall within this range (-20.3o to +0.1oF). The remaining one-third of the values represent significant temperature anomalies- these are the Januaries that residents will remember as being very cold or very warm, when compared to normal conditions.

In this study I have used the standard deviation as a measure of what constitutes a significant temperature or precipitation anomaly. Note that other studies may use different criteria. Seasonal data is handled in a similar manor. During the winter of 2000-2001in Anchorage for example, the mean temperature was 7.7o F above the long-term mean. The seasonal standard deviation is 3.8o F, hence a 7.7o F difference represents a major warm anomaly (twice the standard deviation). The anomaly did not last long however, the 2001-2002 winter mean was only 0.4o F above normal. [note that throughout this study, and especially in the graphs, since a winter season spans two calender years, the temperature or precipitation value is labeled as the later of the two years].



Weather Basics

The weather patterns that we observe across the North Pacific, including the Bering Sea and Gulf of Alaska vary considerably from week-to-week and month-to-month. During the cooler months of the year when a region of strong upper-level winds, called the polar jet stream, lies to the south of the Alaska Peninsula and the Gulf of Alaska (roughly south of 50o N latitude), temperatures in Alaska are cold, often below normal (Figure 2). The position and strength of the polar jet is important because it is along this feature that most large winter storm systems form and move. In general, air to the north of the polar jet is 10o-20o F cooler than air to the south of the jet.

At times the polar jet and associated storm track will move large north-to-south distances over the eastern half of the Pacific Ocean, in just a couple of days. This produces variable weather in areas underneath or near the polar jet. There are other occasions when the jet remains nearly stationary for several weeks at a time (or longer as in the case of a blocking ridge pattern). Over the course of a normal winter in Alaska, the polar jet moves north and south quite regularly. In some winters however, the jet may remain further south than normal and at other times further north. Closer inspection of the weather data during those winters indicates that usually two or three months will experience large polar jet variations, while the remaining months are near normal. When the seasonal statistics are calculated, however the anomaly dominates.

In short, weather patterns are pretty much in a constant state of flux, and what we as human's perceive as 'normal' is nothing more than a mean or average state of the atmosphere. What we are primarily interested in are the major deviations that these weather patterns make from those normals.



Climate Forcings

In the introduction the concept of climate forcings was mentioned, its now time to look at these in some detail. Keep in mind that our understanding of these physical processes is far from complete, know doubt there are forcings that we have not even identified yet. In this section I will introduce each topic then at the end of this section list the impacts on Alaska temperature and precipitation.

Pacific Decadal Oscillation

In the mid and late 1990's researchers noticed that fish catches in the Gulf of Alaska seemed to fluctuate on a time scale of roughly 20 to 30 years. Closer investigation showed that sea surface temperatures (SST's) across the region fluctuated roughly over the same time period (Manuta et al, 1997). The fluctuation from above normal water temperatures to below and vis versa has been termed the Pacific Decadal Oscillation or PDO. When SST's in the central North Pacific are above normal, SST's along the coast of Alaska and British Colombia tend to be warmer than normal, this is referred to as the positive phase of the PDO (Figure 3). When SST's in the central North Pacific are below normal, water temperatures along the coasts are usually above normal (negative phase of the PDO). One phase of the PDO last approximately 20 to 30 years, there are occasions however where the PDO undergoes a 2-5 year mini-reversal, before reverting to its original state.

Researchers measure the strength of the PDO by monitoring SST's across the North Pacific (Figure 3). This is accomplished through the use of buoys, ship observations and satellite monitoring. For each month of the year a mean monthly SST is calculated for a number of locations across the North Pacific. These values in turn are used to construct a PDO index, which ranges from about -3 to +3 (no units). A PDO index of 0 for example indicates that SST's across the North Pacific are very near the long-term normal water temperature for that month. An index value of +1.5 indicates that the North Pacific water is abnormally cool, while water along the coasts is abnormally warm.

Not all climate researchers and oceanographers agree in the existence of or influence that the PDO exerts on the climate of the Pacific Basin. Nevertheless there is mounting evidence that the PDO alters wind patterns and the storm track over the eastern half of the North Pacific. For example, Bond and Harrison (2000) found that during the positive phase of the PDO, low pressure storms (also called upper-level troughs) were 3 times more likely along the 170o W meridian than ridges (high pressure). Table 1 shows the phase changes that have been identified to date. During the negative phase, ridges outnumbered low pressure storms by two and one-half times along the 170o W meridian. The preference for either ridges or lows during a specific phase of the PDO has important consequences on both the short-term and long-term weather of Alaska.

El Nino and La Nina

Table 1- Phases of the PDO
Years Phase
1925-1946 positive
1947-1976 negative
1977-1996 positive
1997- ???? negative
The much talked about (and hyped) El Nino and the much less discussed La Nina, fall under the heading of:
El Nino and Southern Oscillation (ENSO). These events primarily occur in the equatorial and sub-tropical zones of the Pacific and Indian Oceans. (No equivalent in Atlantic). The leading indicators are significant changes in the SST's along the equator as well as changes in the surface pressure pattern (and associated winds) over the sub-tropical regions of the Pacific Ocean. El Nino's and La Nina's represent opposite phases of the same process. As you will see in the following list, El Nino's (and La Nina's), seem to start as fluctuations in equatorial SST's, which then in time modify the atmosphere, in that sense ENSO events are both an external and internal forcings.

An El Nino is characterized by the warming of water located in the central and eastern equatorial Pacific Ocean, from the surface down to a depth of several hundred feet. Most of these events begin sometime between March and May, last from 12-18 months but on occasions much longer. During an El Nino the following can generally be observed:

1. Easterly tradewinds diminish, often giving way to westerly winds.

2. The subtropical jet stream intensifies over the eastern Pacific and southwestern

USA.

3. Well above normal rain along the west coast of South America, southwestern USA and the islands the central Pacific. Drier than normal in the southwest Pacific.

4. Weakening of the Kurisho Current (warm water which moves up the east coast of Asia).

5. Extended periods of high pressure (ridging) positioned over the west coast of the USA- often in conjunction with a split polar jet stream over the eastern Pacific.



La Nina's on the other hand represent a cooling of central and eastern Pacific equatorial waters. The most pronounced atmospheric response to a La Nina is the increased amount of time during the winter that the polar jet stream spends at lower latitudes (30o-45o N), compared to non-La Nina years. The southerly trajectory of the polar jet tends to produce cooler than normal temperatures in places like Alaska and western Canada.

There are two ways in which ENSO events are measured, the most obvious is to monitor SST's along the equator. In order to do this a number of buoys have been deployed across the equatorial seas. These buoys measure the temperature of water from the surface down to several hundred feet. A second measure of ENSO is based on the surface pressure difference between Tahiti and Darwin, Australia. When the pressure is considerably higher in Tahiti than in Darwin, a La Nina typically occurs. When the pressure is considerably lower in Tahiti than in Darwin, an El Nino is in progress. This pressure difference is calculated monthly and has led to the creation of a index that gives a rough indication of the strength of an event. This index is referred to as the Southern Oscillation Index (SOI, see Figure 4), and ranges from about -2.5 to +2.5 (no units). An El Nino is consider to exist when the index is less than -0.50, and a La Nina when it is +0.50 or larger. When the index ranges from -0.50 to +0.50 the ENSO phase is referred to as being neutral (non-event). Since the monthly pressure data is in good agreement (highly correlated) with the SST data, and since it is much easier to monitor, the SOI index tends to be used more often then SST data.

Arctic Oscillation

Centered over the north pole (arctic) at a height of about 20 to 30 miles above the earth's surface (stratosphere), is a region of very strong winds that blow in a counterclockwise direction when viewed from space (hence it is an internal forcing). It has been recognized that significant changes in the strength and position of these winds alters the storm track and winds in the lower atmosphere. The strongest response to these changes appears to occur in the North Atlantic, however there is evidence that there is a modest amount of correlation between the Arctic Oscillation and position and strength of the Aleutian low pressure system. Like most identified climate forcings, the Arctic Oscillation has its own index, which in this case is based on monthly wintertime sea-level pressure patterns of the Atlantic and Arctic basins. The index values range from about -3 to +3 (no units). When the index is positive, winds in the stratosphere (polar vortex) tend to be stronger then normal, which in turn produces lower than normal sea-level pressure over most of the arctic.

Additional Indices

In their search for an understanding of how the earth's climate works, researchers have devised several additional climate indices that in and of themselves do not represent an identifiable forcing. These additional indices are used to help monitor the state of the climate. The most important of these indices for the North Pacific region is called the Pacific North American pattern (PNA). This index is calculate monthly, and represents the difference in the height of the 500 mb pressure level above the earth's surface, at four locations. Two of the locations are over the Pacific Ocean, one is over the western USA and the fourth is over the southeastern USA. Essentially this index shows the amount of north-south movement of the polar jet stream over the eastern Pacific and western USA. When the PNA is positive, there is a deeper than normal trough (low pressure) over the central and eastern Pacific (see Figure 2). This frequently produces warmer than normal temperatures in southeast, southcentral and at times the eastern half of the interior of Alaska. When the PNA is negative, the polar jet stretches almost straight across the North Pacific between 40o N and 50o N. This produces cooler than normal temperatures across just about all of the state, with the exception of the southern half of southeast.

Combined Forcings

So far we have been considering forcings as individual events, in reality of course they can occur simultaneously-it is this interaction of climate forcings which makes climate research challenging. For example, El Nino's that occur during the positive phase of the PDO do not have the same impact on temperatures and precipitation across Alaska that the equivalent El Nino would have if the PDO was in a negative phase. In some cases these climate forcings may positively interfere one with another, leading to a very large anomaly, at other times they may have a negative interference, effectively weakening each forcing. Some of these interactions will be discussed in the next section.

Table 2- PDO winter temperature anomalies (oF) throughout Alaska.
Station - PDO + PDO
Barrow -1.1o +1.1o
Nome -1.5 +2.0
Bethel -1.7 +2.3
St. Paul -0.8 +1.1
Fairbanks -3.1 +1.0
Anchorage -1.5 +1.4
Kodiak -0.9 +0.6
Cordova -1.2 +1.1
Yakutat -1.5 +0.8
Another consideration is the fact that these climate forcings may vary considerably in time as well. There is mounting evidence that long-term weather patterns over the North Pacific and Alaska (i.e.-position and strength of polar jet) may be for example, more strongly influenced by the Arctic Oscillation than it was in previous decades. After a number of years or decades this coupling may change so that AO has a smaller influence than it did previously. These kinds of topics are active areas of climate research.

Impacts on Alaska Weather & Climate:

As we begin to look at the impacts these climate forcings have on precipitation and temperatures in Alaska, keep in mind that there are many ways of displaying and analyzing data, this study only covers a few. More attention is given to temperatures than precipitation for the simple fact that snowfall is difficult to measure accurately, especially in windy regions that only receive small amounts (i.e.- Interior of Alaska and arctic coast). In short, measurement errors in these regions is large, so large in fact that the snowfall data should not be used in this type of study.

We will start by looking at the influence that the PDO cycle has on temperatures. Table 2 lists ten climate stations and shows the winter temperature anomalies for the last two complete phases of the PDO (1947-1976 -PDO and 1977-1996 +PDO). Note that all of these stations are cooler than normal during -PDO and warmer than normal during +PDO. These anomalies are not for the most part large, but considering they are for periods of 30 and 20 years respectively, they are significant. Anomalies of the opposite sign do occur during any given phase of the PDO, for example, during a -PDO phase, warm anomalies (we are talking about monthly anomalies here) can occur. However, for everyone that does develop there are from 2 to 3 cool anomalies. A similar process occurs during the +PDO, but in this case warm anomalies out number cool anomalies by 2 or 3 times. Summer temperatures behave in a similar fashion as winter, however the magnitude of the anomalies tend to be smaller.

Winter precipitation data shows an increase in rain and snow during the positive phase of the PDO when compared to the negative phase, for the Gulf of Alaska region north of about 57o N, and the Alaska Peninsula. The increase varies from one locale to the next but typically ranges from 10% to 40%. Annette on the other hand typically experiences a modest (10%) decrease in precipitation. Summer precipitation patterns are similar to winter although the increase in the Gulf of Alaska is not as strong as it is in the winter.

Now that we have established that the PDO has a significant influence on temperatures it is time to consider ENSO events. Look back at Figure 1, notice the color of each dot. Red dots indicate El Nino years, blue La Nina's, and green for neutral years. Notice that most of the red dots are above the horizontal line (the long-term mean). We can conclude that the average El Nino produces a modest increase in winter temperatures across the state. It is important to note that there are El Nino winters in which the average temperature is well below normal. It just depends on how the weather patterns over the North Pacific happen to set-up for that year. Although not shown, warm season temperatures (April-October) in Western Alaska and the interior, tend to be slightly warmer than normal during an El Nino. In the Gulf of Alaska, April-October temperatures are poorly coordinated with El Nino events. It is important to keep in mind that when an El Nino occurs, the impact that it has on temperatures in Alaska is highly variable- in other words, one El Nino may generate above normal temperatures, as for example during the winter of 1976-1977, when just about the whole state had well above normal temperatures. Other El Nino's may only produce a temperature anomaly in a limited region of the state, as in 1977-1978. The blue dots in Figure 1 represent La Nina events, it is evident from these graphs (and others which are not presented in this paper) that there is a moderate to high correlation between a La Nina and cooler temperatures. The correlation
Table 3- Winter temperature anomalies (oF). *Asterisk denotes high variability from one event to the next.
Station El Nino
- PDO + PDO
La Nina
- PDO
Barrow -1.2 +1.1 -3.0
Nome -1.6 * +1.4 * -4.3
Bethel -2.4 * +1.5 * -4.0
St. Paul -0.7 * -0.2 * -2.8
Fairbanks -1.3 * +2.8 -4.3
Anchorage -0.9 * +2.1 -3.7
Kodiak -0.4 * +0.4 * -2.8
Cordova -0.9 * +2.2 * -4.3
Yakutat -1.1 * +1.6 * -3.4
Annette -0.8 * +2.0 -2.9
between cool temperatures and La Nina's is much higher than it is for warm temperatures and El Nino's. In other words, there is more long-range predictability for La Nina's than for El Nino's. This results from the fact that during most La Nina's, the aforementioned polar jet stream tends to spend most of the winter well south of 55
o N, allowing cool (or cold) arctic air to move over the state from Siberia and the Chukchi Sea. Figure 5 is a graphical representation of Fairbanks and Anchorage temperatures, broken-down by ENSO events.

In Anchorage for example, the number of heating degree days during the typical La Nina increases by about 10%, (9% for Juneau and 5% in Fairbanks), hence residents can expect an increase in their heating billing accordingly. Note that temperatures during the summer months are effected as well, but the magnitude of the anomalies are not as large as they are during the winter.

The next step is for us to consider combinations of the PDO and ENSO. This of course can get quite complicated so the following discussion is just an overview. First we will look at

El Nino and La Nina events that occur during the negative phase of the PDO followed by the positive phase. Table 3 shows winter temperature anomalies (November-March) for select stations. El Nino's that occur when the PDO is negative tend to be slightly cooler when compared to the long-term mean, but the variability is large. During the positive phase of the PDO, El Nino's tend to be warmer than normal, however there is still considerable variability (in statical terms the standard deviation is large). Notice that La Nina's which occur during -PDO tend to produce abnormally cooler temperatures, with incidently very little variability. La Nina's which occur during +PDO are not listed because only one occurred during the 1977-1996 period (which by the way may give some clue about the development of La Nina's).

The La Nina of 2000-2001 was weak, nevertheless we would have expected from past history that it would have produced cooler than normal temperatures across the state- the data shows however that temperatures during that winter were well above normal. No one knows why this occurred, however, the during that period the PDO shifted temporally from being in a negative phase to a positive phase. Cooler water in the central North Pacific caused the polar jet to be re-directed over southcentral and western Alaska for much of winter.

Analyzed winter precipitation data shows as much variability as temperatures do. Let's use Seward as an example. If we add up all the precipitation that has been observed during El Nino's over the last 50 years (15 events for which there is data at Seward), the mean of those events is 30.9" which is almost exactly the same as the long-term mean of 30.1". Inspection of the individual events shows that the El Nino of 1976-1977 produced over twice the normal winter precipitation. There were however three El Nino's during which precipitation was abnormally low. For the remaining 11 events, precipitation amounts were clustered around the long-term mean. This high variability from event-to-event is common for stations in Alaska, as illustrated in Figure 6. As noted above, on occasions an El Nino will produce extremely wet conditions, but overall, above normal precipitation is not guaranteed. For the Gulf of Alaska region about one-in-three El Nino's produces a very wet winter and/or summer. The exception is the southern half of southeast Alaska, where large anomalies are a little less frequent. A wet (dry) summer in the Gulf of Alaska does not guarantee that the following winter is going to be wet (dry) as well, although it does frequently occur. On rare occasions a abnormally wet (dry) summer will be followed by very dry (wet) winter.

Figure 7 shows two examples of April-October precipitation anomalies. In the graph depicting Fairbanks precipitation, notice the low correlation between anomalies and La Nina's and

El Nino's. This is true for most of the state except the Gulf of Alaska region. In the Yakutat graph, notice since the 1970's how the majority of El Nino's have produced a number of positive anomalies (some of them very large). One can also start to see signs of a PDO influence on precipitation in general, with a majority of negative anomalies prior to 1977, during which the PDO was in a negative phase as well. Although not show in graphical form, winter precipitation during La Nina's varies considerably from one event to the next, the variability however is not as high as it is for El Ninos. In the Gulf of Alaska region somewhere between one-in-two to one-in-three La Nina's will produce a dry anomaly at any one given station (almost every La Nina will produce a dry anomaly somewhere in the Gulf of Alaska). About one-in-four La Nina's will produce a wet anomaly at a particular station. There was no attempt to correlate winter precipitation for the southcentral, the interior and the Arctic slope due to the aforementioned problem with accurate snow measurements. In addition, outside of the Gulf of Alaska region there is no obvious correlation between a particular phase of the PDO and the amount of precipitation that is recorded at a climate station.

Let's briefly look at the connection between temperatures and the Arctic Oscillation (AO). On the Arctic Slope, Barrow and Barter Island (1948-1988) have a weak correlation between the AO Index and winter temperatures. Prudhoe Bay (1968-present) has a higher correlation than the either Barrow or Barter Island, nevertheless it's still not very strong. South of the Brooks Range there is almost no correlation between the AO index and winter temperatures. In essence, the AO index is a fairly weak measure of the climate forcing in the western Arctic. This does not mean that climate forcings in the arctic do not influence northern Alaska, it merely suggests however that the AO index is not a very good measure of that influence.

Summary of the impacts on Alaska:

* El Ninos produce a wide range of temperature and precipitation responses; from well above normal, near normal, to well below normal. In other words, weather during these events is highly variable with little predictability. The specific impacts made by an El Nino depends to some degree on the phase of the PDO.

* Even though any given El Nino may or may not produce a warm anomaly at one particular station in Alaska, almost every El Nino will produce a warm anomaly somewhere in the state.

* La Nina's produce cooler and drier than normal conditions in both winter and summer. Very few La Nina's form during the positive phase of the PDO. Overall there is considerable predictability with La Nina events, the exception being the 2000-2001 event where temperatures across most of the state were well above normal. In the Gulf of Alaska region, there is a tendency for more wet years to occur during the positive phase of the PDO than during the negative phase.

* Although no one has done a detailed study of the frequency of storms that occur in or near Alaska, it can be said that, in general, regions that experience more southerly flow (that is from southwest through southeast) over the course of the season or year, will experience more storms (wind, precipitation).

* It is important to keep in mind that not only is there considerable variability from one winter to the next, but there is also a lot of variability in temperature and precipitation during any one given winter as well. Temperature anomalies (warm or cold) typically have a duration of 2 to 3 weeks, which is often followed by a period of near normal temperatures. On occasions, however, a temperature anomaly of one sign will be followed by a significant anomaly of the opposite sign. These 2 to 3 week periods of cool or warm temperatures are highly correlated with the location of the polar jet as it moves north and south across the North Pacific.

A cold (warm) winter occurs when the number or magnitude of the cold (warm) anomalies is larger than warm anomalies. Keep in mind that during cold (warm) winters, short duration warm (cold) anomalies can still occur. In other words, everyday during a cold (warm) winter is not necessarily colder (warmer) than normal. Here is an example; temperatures in Fairbanks during October and November of 1999 were well below normal. December temperatures remained on the cool side but were not cold enough to be considered an anomaly. Both February and March of 2000 however turned out to be warm anomalies. The net result was a winter mean temperature that was near normal.



Weather/Climate Predictability:

Those of us in the weather and climate business would, of course, like to be able to predict temperature and precipitation anomalies well in advance. Due to the state of Alaska's proximity to the storm track (Aleutian Low and Polar Jet), moderate changes in the position of the polar jet make for some significant changes in our weather. The seesaw nature of temperature and precipitation from month-to-month and season-to-season is often referred to as 'natural variability'. In other words, the weather, and hence climate, is always in some state of flux (i.e. changes in ocean currents, solar output, and volcanic eruptions). Therefore, trying to determine what is 'normal' is not as easy as it may appear at first. An example of non-natural climate variability (or climate change if it last for any length of time) would be anthropogenic releases of greenhouse gases and other pollutants.

A number of other researchers have noted over the past decade that it appears that the climate forcings themselves are not constant in time. For example, Overland et al 1999 have suggested that the mean winter position of the Aleutian Low and its central pressure vary on several time scales. There are times when the Aleutian low and polar jet are highly correlated with North Pacific anomalies (PDO and ENSO), while at other times it's correlation with climate forcings in the arctic increases at the expense of the PDO and ENSO. The fact that the frequency of La Nina's decreases during the positive phase of the PDO, may be related to the strength of the arctic forcings. All of this makes for some difficult analysis!

One of the primary challenges of the climate researcher is to find repeating cycles of long-term weather patterns (periodicity). Look back at any of the temperature plots in Figure 1. Can you find any patterns that are repeated? To be honest, its difficult to look at these graphs and make any sense out of them. However, one of the ways we can attempt to make some sense of data, is accomplished by approximating those graphs with a mathematical function. It turns out that trigometric functions such as sine and cosine (Fourier analysis) work pretty well for this application (other more complex functions like wavelet analysis work as well). I preformed this type of analysis on November through March temperature data for 11 sites across Alaska. The results clearly show that there is roughly a 20 to 30 year cycle in the temperature data, which corresponds to the PDO cycle. There is some evidence that longer cycles may exists, but due to the short period of record (50 to 80 years), we will have to wait until more data becomes available before they can be explored. There are a number of shorter cycles as well, primarily centered around 8 and 3 years. April through October temperatures were also analyzed, overall the results are quite similar to those found for the winter period.

It is interesting to note that the 20-30 year cycle tends to be fairly steady in time. That is they are quite consistent over the period of record, while cycles that are shorter than 20 years tend to be highly variable in time. This is evidence that the impacts of El Nino's and La Nina's, at any given location, differ greatly from one event to the next. It also confirms that the frequency of both ENSO events, varied considerably over the 20th century.

Another attempt at long-range forecasting is to write some type of complicated mathematical equation that is able to predict temperature anomalies, based on what has occurred in the past. For example, we might try to base an equation on climate indices like the PDO and ENSO. A generic equation of this nature might look something like:



Tanomal = AxF(PDO index) + BxG(SOI index) + CxH(PNA index)



where; A, B, and C, are weights that have to be determined by regression or some other method. F, G, and H are functions that have to be judiciously selected. Generally these functions will be trigometric, logmetric, or polynomial in nature. In order to have any forecast value, the index values would span a number of months (i.e.-3 to 6 months) prior to the forecast period. For example, if we wanted to solve Tanomal for the December through February period, the average PDO index for the months of June through November might be used (not limited to this period-whatever works!). In a similar fashion, a number of prior months worth of the SOI and PNA would be used as well. So why aren't these types of equations used more frequently? The general answer is that this method only captures a portion of the signal variation, in large part because the indices themselves are estimates of the climate forcings. Furthermore there are no doubt additional climate forcings we have not recognized at the present time, likewise the interactions between the known forcings could be so complicated that our 'simple' equations are not adequate.



Global Warming and Alaska:

There has been a lot made in the press over the past decade about global warming, especially at the higher latitudes. Essentially the increase in greenhouse gases like carbon dioxide and methane trap large amounts of longwave radiation, which in turn causes an increase in air temperatures. There is no question that the amount of greenhouse gases has increased over the past 50 years. What is in doubt is how the atmosphere and oceans are responding to those increases. For example, it is known that sea water can hold considerable amounts of carbon dioxide. Could increases in atmospheric carbon dioxide be mitigated by its up-take in the worlds' oceans? Additional complications arise because our planet is still emerging (warming) from the Little Ice Age which occurred between the 15th and 17th centuries.

One of the main findings of the computer model simulations of global warming is that the arctic warms much faster than the rest of the planet. The reason that high latitude regions of the world are suppose to warm faster than the rest of the planet is due to the fact that a small change in temperature would reduce the amount of snow and ice (sea ice, glacier ice) in these locations. The reduction in snow and ice coverage then allows more sunlight to heat the ground or ocean surface, which in turn produces slightly warmer temperatures. Once this cycle starts, it is thought by most researchers that it would be difficult for the process to stop (positive feedback). There are a number of unanswered questions however: how will precipitation patterns change due to the warming? How will ocean currents change due to influx of fresh water as sea ice and glaciers melt? How will SST's change? No one really knows the answers to these questions!

Is there evidence of temperature increases in Alaska? Figures 8 & 9 show temperature trends at four stations across the state. I have selected these particular stations because they are located in small towns or villages, where hopefully the influence of urbanization has been small. The dotted blue line in each of the plots represents a 5 year running mean. Each dot on the blue line is actually an average (mean) for that year as well as the previous four years. This has the affect of smoothing the data, nevertheless it makes it easier to follow the overall trend. The most conspicuous element of these four plots (as well as data from other stations which is not plotted), is the large temperature increase which occurred around 1977. We noted earlier in this paper that during the 1976-1978 period there were two consecutive El Nino's, the latter one being very strong. This period also corresponds to a shift in the PDO. Just prior to this warming trend, temperatures at most stations were well below normal, which made the warm-up all that more impressive. Notice, however, by the mid-to-late 1980's, winter temperatures had returned to near normal values.

Overall the temperature trends correspond pretty well with the phases of the PDO. There is some evidence in the temperature record of a number of stations located in the interior of Alaska, that a slight warming has occurred (Stafford et al). However most of the temperature records in this area only span one complete cycle of the PDO, hence it is difficult to ascertain if the warming is 'true' or is an artifact of the warm cycle of the PDO which spanned the 1977-1997 period.

Let's look at Barrow and the arctic slope a little closer. First, there have been both significant warm (early 1940's, early 1980's, 1990's) and cold (1920's, and 1950's) anomalies over the history of these stations. The current warm anomaly, which began in 1993 has been punctuated by several cooler winters (1995, 2000). The 'jump' in temperature from 1992 to 1993 has occurred in previous periods as well. The climate station at Prudhoe Bay, which has been collecting data since 1968, has a temperature trend similar to Barrow. Incidently the warming during the 1990's is also evident during the warmer months of the year (April-October) as well. In fact, even though the magnitude of the warming is not as large for the months of April to October as it is for the winter months (because of the very large amounts of longwave radiation lost during the winter), the percentage of months that experienced a warm anomaly is much higher for the April-October period. The 1990's warm anomaly also shows-up in the temperature records of stations south of the Brooks Range, however this anomaly is much weaker than past warm anomalies.

A number of people have attributed this current warm anomaly to global warming. It is the opinion of this researcher that the current warm anomaly occurring on the arctic slope represents a phase shift, which may or may not be linked to the PDO, but most likely to significant changes in the circulation of the atmosphere and ocean in the Arctic. Multi-year pack ice that covers a significant part of the Arctic Ocean was during the summer of 2002, reduced to its smallest areal extend since records have been kept. In order to conduct a detailed study of global warming and its influence on the arctic, data from northern Canada, Scandinavia, and the Russian arctic would need to be analyzed. In addition, proxy data such as glacial recession, melting of permafrost, soil temperatures, changes in the range of biological species, etc. would need to be examined as well.

Other forcings that were not considered in this study were the influence of volcanic eruptions, which can be very significant (Krakatau 1898, Pinatubo 1991). Typically however the impacts only last several years after the eruption. Over the years there has been interest in the impact that small changes in the output of the sun may have on earth's short-term climate (decades to several centuries). Nothing conclusive has been established as of now. It is most likely that climate variability is a product of simultaneous changes by more than one of the forcings.

What we do know is that the winter of 2000-2001 was one of the warmest on record in Alaska, only the southern half of the panhandle did not experience abnormally warm temperatures. This is interesting because we are in the negative phase of the PDO and the winter of 2000-2001 experienced a weak La Nina (SOI=+0.57). Closer examination of the PDO index indicates that during the late summer and autumn of 2000 the index ranged from -1.0 to -1.3, by December it was +0.5 after which it stayed positive until April. This rapid change in SST's in the central North Pacific may have been sufficient in and of itself to produce 'abnormal' southwesterly flow over Alaska (like Figure 2- lower panel). By the late spring temperatures returned to near normal values. The winter of 2001-2002 was uneventful with regard to temperature anomalies. We are then left with the question: was the warm winter of 2000-2001, the warm anomaly on the arctic slope over the past 10 years and the reduction of multi-year sea ice in the Arctic Ocean an oddity, or does it represent the start of some type of phase shift in the climate forcings? Stay tunned as the story unfolds.



Recommended WebSites:

+ Western Region Climate Center (good place for temp and precip data)

www.wrcc.dri.edu

+ Global Warming

www.oism.org/pproject/s33p36.htm

http://lwf.ncdc.noaa.gov/img/climate/globalwarming

www.ngdc.noaa.gov/paelo/globalwarming/what.html



References and Recommended Reading:

Bond, N.A. 2000: The Pacific Decadal Oscillation, air-sea interaction and central North Pacific winter

atmospheric regimes. Geophysical Research Letters, vol.27, no.5, pg. 731-734

Douglas, A.V., Cayan.D.R., Naimais, J. 1982: Large-scale changes in North Pacific and North American weather patterns in recent decades. Monthly Weather Review. vol.10, pg.1851-1862

Niebauer, H.J. 1998: Variability in Bering Sea ice cover as affected by a regime shift in the North

pacific in the period 1947-1996. Journal of Geophysical Research. vol.103, no.c12,

pg. 27,717-27,737

Mantua N.J., Hare S.R., Zhang Y., Wallace J.M., Francis R.C. 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society.

vol.78, no.6, 1069-1079

Overland, J.E., Adams J.M., Bond N.A. 1999: Decadal variability of the Aleutian Low ands its relation

to high-latitude circulation. Journal of Climate. vol.12, pg.1542-1548

Papineau, J.M. 2001: Wintertime temperature anomalies in Alaska correlated with ENSO and PDO.

International Journal of Climatology. vol.21, pg.1577-1592

Stafford J.M., Wendler G., Curtis J. 2000: Temperature and precipitation of Alaska: 50 year trend

analysis. Theoretical and Applied Climatology. vol.67, pg.33-44