AMO, NAO and predicting Atlantic Coast temperatures

Several months ago, after finishing the individual state temperature evaluations, I started putting the state data together in regional groups, starting with the Atlantic States. What motivated that start was the discovery that, along the East Coast, the individual states all displayed a drop in temperature, over the period from about 1950 to 1965 on the order of 4 degrees F. This was a surprisingly large number, in terms of the consistency over both space and time, with which the fall in temperature occurred. But there was some physical evidence of the impact of these changes since, following a piece on bird populations, it was possible to tie the fall in temperture to the migration of the Black Capped Chickadee, as an example, which moved south as the temperature fell, appearing for the first time in North Carolina at the height of the cold spell. When the average plot for all the states (excluding Florida) was derived the average fall was reduced to about 3 degrees F, with the drop in temperature reducing for states further South.

Averaged Time of Observation corrected (TOBS) temperatures for the Atlantic states, with state averages weighted by area in the overall average of the data.

Today’s post shows how this can now be explained, and how it might, from existing data, be possible to predict the future shape of this plot.

Looking at the range of states over which this fall occurred it was clear that it while the temperature drop made it across the Adirondacks into the Midwest it got no further and the Midwest had some differences in the pattern of temperature recovery. The weather normally travels from West to East, and so the logical cause seemed to be to look first at what happened in the Pacific, to see if US temperature variations could be explained.

Over the last couple of weeks this has taken the course of looking at the impact of the El Nino Southern Oscillation (ENSO) with its roughly 5-year cycle, first on regional temperatures and then on that of the individual states. While changes in the ENSO impacts moisture and drought patterns in the states, it did not, comparing temperatures, seem to have as great an impact on the temperatures along the West Coast, where the current flows impact. And further it was restricted to coastal states, in what correlation was immediately visible.

Coming back to the Atlantic Coast there is a similar oscillation in the sea surface temperatures (SST) called the Atlantic Multidecadal Oscillation (AMO). Looking at a plot of this average over time, it is immediately clear that there is a concurrent drop in these temperatures, in around the same time period, that might answer the question at to what had caused the temperature drop along the East Coast.

Sea Surface Temperatures in the Atlantic showing the AMO (NOAA )

One can then, in the same way as for the Pacific Coast States, overlay this curve on the average of the Atlantic States variation (with the Midwest added to show the limit of range) to validate the above statement. However in the first step the overlay was kept at the same vertical scale, so that the size of the anomaly in SST’s that forms the oscillation, could be compared with the size of the temperature changes on land.

Comparison of the fluctuation in AMO anomaly with the temperatures on the Atlantic and Midwest regions. Note that the AMO plot has been scaled and randomly placed vertically to allow distinction from the two other curves.

The immediate first observation is that the land temperatures have fluctuated over twice the scale (roughly) of the SST values, though it should be remembered that the latter are averaged over a much greater area.

Looking over the entire history of the temperature record, which for now is taken as post 1895, the plot for AMO values is doubled, to make it easier to see how the peaks and troughs may coincide, and then 52 deg F as added to the anomaly values to set them below the Atlantic States, and the values are averaged for each year.

Comparison of the form of the AMO (which has been doubled in size) with the changes in the average temperatures of the Atlantic States over the past 112 years.

So now the question comes as to the cause of the changes in the AMO, and one finds that this is currently a work in progress. The AMO itself was only described in 1994, and work on what is the cause, and what it correlates to is ongoing. But there are are several different other cycles in the Atlantic, and it is interesting to see what folk have already found by way of correlation.

One of these other cycles is the drift of the InterTropical Convergence Zone, or ITCZ. Back when I was reading “Stories for Boys” about sailing ships in the Atlantic the ITCZ was called the “Doldrums.”

Location and characteristics of the ITCZ (Da Silva 1994 ).

The Doldrums drift North in the summer. However the movement and nature of the ITCZ behavior has been suggested to be dependent on the AMO, rather than the cause (see for example Knight, Folland and Scaife)

What is actually more interesting, particularly from the “climate prophet” point of view is a suggested correlation with another Atlantic Cycle, that of the North Atlantic Oscillation.

North Atlantic Oscillation – NAO is defined as the fluctuation in the difference of atmospheric pressure at the sea level between two specific locations: Ponta Delgada, Azores and Stykkisholmur/Reykjavik.

Although the NAO is now undergoing a name change to the Northern Annular Mode, the relevant changes in atmospheric pressure (which are shown in color and globally at the Climate Prediction Center can be illustrated, to allow easier layman comprehension with this picture from JISAO).

Pressure patterns in the North Atlantic showing the pressure difference that builds up to value the NOA. (JISAO via Gosselin)

Comes now this new paper from Vukcevic which suggests that there is a correlation between the AMO and the NAO, but that the correlation becomes much better if the values for the NAO are delayed by eleven years. Further that the correlation becomes better when only the fluctuations in the pressures at Reykjavik are considered.

Correlation between an eleven year lagged Icelandic pressure and the 3 year moving average SSTs in the Atlantic – the AMO. (Vukcevic )

What is riveting about this is the current correlation not only the AMO and NAO, but that it is time lagged. And as Vukcevic notes, there is an eleven year time lag between the behavior of parts of the NAO and the consequent change in the AMO. Which means, if this is correct, then we know what the future behavior of the AMO is going to be, since the end of the red line above indicates the SST trends over the next eleven years. That in turn, as was shown earlier, means that the temperature trends for the Atlantic Coast can also be surmised for the next eleven years, and as has been noted elsewhere this holds true not only for the United States but also Europe in that the NAO also has been shown to influence the Central English Temperature (CET).

This is obviously quite intriguing and will no doubt be the object of some discussion over the next few months, particularly since a second paper is promised. It is unlikely that it is the drop in pressure itself that is the cause, but it is conceivable that this is an indicator (dare one say proxy) for some other event and as such it is indicating how this other forcing event is behaving, and through that route becomes a marker itself for the temperature changes that are coming. But then that raises the question as to what it is a proxy for?

The Temperature fall and the Chickadee

Yesterday (Monday) I was reading the Boston Globe and came across an article by David Abel about a recent report by the Audubon Society on their Breeding Bird Survey, and how it was showing that the changes in the climate was influencing bird ranges and populations. The surveys, which looked at the state of Massachusetts, only went back to 1965, and so, being curious, I wrote to Mr Abel pointing out the drop in temperature (about 4 deg F over 1950 - 1965) for the state that I wrote about a couple of weeks ago. I was, sort of hoping that he might look at the survey data before that date, and see if the bird populations had fallen with that temperature drop. Also I wanted to know if the numbers had returned the the 1950 level. However he merely sent me a note thanking me for my note.

But I was now curious to know the answer, and so I had a look for the data on which the Audubon article and the resulting newspaper story were based. And it turns out that the reason that the data only went back to 1965 is that when the BBS started to be implemented). So where could I go to get earlier data?

The answer turns out to be the Audubon Christmas Bird Count (CBC) But the count is very species specific, and since I have only a limited amount of time I decided that I would just pick one bird (at least initially) and see what the results showed. And if the story was about Massachusetts, then it seemed sort of logical to use the state bird, which turns out to be the Black Capped Chickadee .(And as it happens, though not told until tonight by the Bishop, this is also the Maine State Bird - about which more anon .)

So the first thing to do was to get the basic information on the population with time, looking at the annual counts for the different states along the Atlantic shore from the Audubon site. And that showed the answer to one of my questions. Let me begin by repeating the temperature plot for Massachusetts over the last hundred years from my look at the state. (I am using the USHCN numbers for this particular study) so that you can follow how I got there.

Average temperature as a function of time for the state of Massachussets

You can see the significant drop in average state temperature that occurred in the 1950 - 1965 time frame. So how did the bird population respond?


It has only been in the last few years that the state has returned to those 1950 temperature, so did the bird population change, and has it made it back? The Audubon page allows you to plot a graph:

Black Capped Chickadee population in Massachusetts using the Audubon CBC

The count doesn’t look that valid before 1940, and when I checked other states I had the same problem, so I am going to restrict the rest of this to the years after 1940.

You can see that there was a drop in the count at around the period that the temperature fell, and this was confirmed when I looked at the numbers for the other states sitting close to Massachusetts. But, as you can see, the question that I had asked David Abel as to whether the numbers had returned to the values back before the temperature fell, was answered with a Yes!

However we can now go on from there. Having numbers and a temperature plot, the question can be answered as to whether the population was affected by the temperature, as the two plots above seemed to indicate. So I re-plotted the data as count versus temperature. And this is what I got for the state:

Chickadee count as a function of average annual Massachussets temperature

The plot showed, when I fitted a polynomial, that there was an optimal temperature for the population. I checked this with similar plots for the states along the Atlantic that I covered in the earlier post. Some states didn’t get that warm, and some didn’t get that cold. But for New Jersey, Connecticut, Pennsylvania, Rhode Island and perhaps New York, the data plots showed about the same peak in population at around 51 deg F.

Here is the plot, for example, for Connecticut:

Chickadee count as a function of average annual Connecticut temperature

And similarly for New Jersey

Chickadee count as a function of average annual New Jersey temperature

So with Massachusetts temperature hovering around that level at the moment (as back in 1950) we’ll have to hope for no more change in state temperature, since too much fluctuation either way may cause the population to drop.

So the next question would be whether the population drop is because the birds died off, or moved, if the temperature fell off. I don’t have enough data to answer that part of the question, but I have an indication. If one looks at North Carolina (below which the more Southern states don’t see this species apparently) then when that cold spell hit, further North, the state suddenly saw birds appear. (And having discovered the state they came back when it was colder up North).

Black Capped Chickadee counts for North Carolina (Audubon Society CBC)

So it appears that, not only did I get an answer to my question, I also discovered another temperature index through the birds, and so maybe I’ll do some more correlations from time to time. (I was thinking about doing wild turkeys since they are common around much of the country, but there is the problem that they are harvested also so that the numbers are more artificial).

Oddly the plot of optimal temperatures for the Chickadee did not work for Maine, where the temperatures are cooler. Wonder if the Maine version is turning into a slightly different creature – given that it is the Maine state bird too.

Black Capped Chickadee counts for Maine (Audubon Society CBC)