The generality of precipitation/temperature patterns: North Pacific versus North Atlantic

In a series of earlier blog posts (here and here) it was demonstrated that the relationships between England and Wales Precipitation (EWP) and Central England Temperature (CET) show consistency across months, but that the hotter and cooler parts of the year show different relationships:

1. a positive CET/EWP relationship exists from November to February and in a minor way for the fiscal year from July to June
2. a negative CET/EWP relationship exists from April to September
3. no significant relationship in March and October

Despite the rapid change the globe’s climate due to emissions of greenhouse gases by Australia, South Africa, the Gulf States and to a lesser extent other mineral- and fuel-producing nations, these relationships have not substantially changed since 1974.

In this post I will see if an analogous condition to that of the UK also holds in the only other analogous climate region of the globe – southeastern coastal Alaska. Whilst generally extremely similar to the UK in its ecology and environmental history, there are major geographic differences owing to he extreme height of the coastal mountains, which reach much further above the glacial equilibrium line than does Mount Everest.

At the beginning of this year, southeastern coastal Alaska was divided after an examination of long-term station records into four climate divisions:

• AK 9: East Gulf (red, on right)
• AK 10: North Panhandle (blue, on right)
• AK 11: Central Panhandle (purple)
• AK 12: South Panhandle (dark green, on right at bottom)

I have chosen to investigate only AK 9 (East Gulf) for this study, to simplify matters because it is the largest and most “central” of Alaska’s six “maritime” or “southern” climatic divisions – which also include AK 8 West Gulf (green, around Kodiak) and AK 13 Aleutians (purple, in far southwest).

Reliable temperature and precipitation data for Alaska go back only to 1925, a little more than a third the length of the EWP data. For this reason, I have decided not to separate years with and without the dominant influence of greenhouse gas emissions by mineral exporting countries like Australia, South Africa and the Gulf monarchies: they are too likely to dominate the sample and the UK experience is that the change in correlation prove insignificant even though the averages do change significantly.

So, here are the scatter plots by months of precipitation versus mean temperature for the East Gulf division of Alaska, extending from Valdez to Sitka:

July:

As we can see here, in this the hottest month of the year, a general negative precipitation versus temperature regime prevails, exactly as seen for EWP versus CET in the United Kingdom over a record three times as long. The major outlier is the very wet July 1958, which was the second wettest on record with an estimated district average of 473.46 millimetres but no cooler than average.

August:

August, still in the hotter part of the year, shows a similar trend to July, which is in strong agreement with our earlier results re the relationship between EWP and CET in the various months. The relationship is not tight, and I have not measured the correlation coefficient. Two Augusts:

• 1969, the coolest on record but sixteenth driest of 91
• 1981, the second wettest on record but 0.4˚C hotter than all-series average 

show very distinct departures from the pattern of hot/dry and cool/wet.

September:

Here, we see that the seasonal change from hotter weather being drier to warmer weather being wetter than normal appears to be occurring one month earlier than we saw for the EWP versus CET graph in our earlier blog post. The scatter plot for September in the East Gulf division is basically flat, and is flat even with the extremely cold outlier of September 1992, whose estimated district average precipitation total is near normal.

September 1992 500 millibar chart anomaly vis-à-vis 1880 to 1974 mean. Support for the Twentieth Century Reanalysis Project dataset is provided by the U.S. Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment (DOE INCITE) program, and Office of Biological and Environmental Research (BER), and by the National Oceanic and Atmospheric Administration Climate Program Office.

As one can see, the extremely cold polar air over Alaska’s northeastern Gulf coast does not really have a mean onshore flow component, so that the cold was not accompanied by excessive rainfall or snowfall.

October:

In contrast to the CET versus EWP graph, October in southern Alaska already shows very clearly the typical winter pattern whereby precipitation and temperature show a direct correlation. This pattern has long been known via the National Weather Digest (here) for the main city in the region – Juneau – and the figures for October clearly show this pattern extending generally in the region. Cold months of October have anomalous flow from the major cold-air source region of the Yukon.

November:

As the discussed 1986 article about the winter climate of Juneau – located since this year in the Central Panhandle climate division to the southeast of the region graphed – would imply, the direct precipitation/temperature relationship increases in intensity for November.

Two facts one will note with this graph is that there are very few outliers, and that the shape is more curved than linear. The curved concave-down line of best fit implies that the temperature distribution is skewed due to the greater frequency of relatively mild and hyper humid maritime weather vis-à-vis frigid, dry continental conditions. The lack of outliers is such that even November 1956, on the “lower” right, was still warmer than average, and November 2002 at the extreme top was still wetter than average:

November 2002 500 millibar chart anomaly vis-à-vis 1880 to 1974 mean. Support for the Twentieth Century Reanalysis Project dataset is provided by the U.S. Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment (DOE INCITE) program, and Office of Biological and Environmental Research (BER), and by the National Oceanic and Atmospheric Administration Climate Program Office.

The key point from this November 2002 chart, which those only familiar with sea level charts might not instantly grasp, is that the powerful anticyclonic anomaly over southeast Alaska is still wetter than average on its warm western side because the anomalous flow is onshore.

December:

December, as the month where the winter solstice occurs, reflects clearly the pattern of mild, hyper-wet weather opposed always to frigid, dry weather. In fact, nothing approaching a moderate outlier can easily be seen here – which suggests much more powerful correlations than for EWP versus CET. The curved, concave-down line of best fit is also more clearly visible than for November, as is the record cold and dry December 1933 with its extremely strong flow from the frigid Yukon:

December 1933 500 millibar chart anomaly vis-à-vis 1880 to 1974 mean. Support for the Twentieth Century Reanalysis Project dataset is provided by the U.S. Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment (DOE INCITE) program, and Office of Biological and Environmental Research (BER), and by the National Oceanic and Atmospheric Administration Climate Program Office.

It is noticeable how strong the anomalous flow in December 1933 was vis-à-vis any of the other months whose flow patterns have been diagrammed here.

January:

Vis-à-vis the almost perfect relationship seen in December, January does not show quite so consistent a positive correlation, nor so curved a line of best fit. The line of best fit is much closer to the “familiar” linear shape than for December or even for November. More significantly, the famous month of January 1949 is an extremely powerful outlier being very close to the wettest on record, receiving 703.33 millimetres, but being no warmer than average at -7.9˚C:

January 1949 500 millibar chart anomaly vis-à-vis 1880 to 1974 mean. Support for the Twentieth Century Reanalysis Project dataset is provided by the U.S. Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment (DOE INCITE) program, and Office of Biological and Environmental Research (BER), and by the National Oceanic and Atmospheric Administration Climate Program Office.

This anomaly occurred because in January 1949 – as can be seen above – the flow anomaly was westerly (moist) but came from the cold Bering Sea and no warm source was accessible due to the powerful North Pacific anticyclonic anomaly.

This month was the snowiest January on record over Alaska as a whole, and the coldest on record over a large portion of the western United States, where it has been rivalled only by the Januaries of 1916, 1930, 1937, 1950, 1957 and 1969. It was extremely warm, however, over the eastern United States and Eurasia, being almost the “year without a winter” in the UK.

February:

February retains the basic winter scatter-plot pattern we have seen since October – mild and hyper humid versus frigid and dry. If anything, the line of best fit is more akin to the curved December shape than was seen for January.

Although not to the same extent as with December, there are no strong outliers. Even the record wet February 1964 with a strong high-level low pressure anomaly over Alaska itself was warmer than the all-series mean (which more than CET is distorted by greenhouse emissions from Australia and other resource-exporting nations), and the record dry February 1989 (driest for any month throughout this super-humid region) still very cold.

February 1989 500 millibar chart anomaly vis-à-vis 1880 to 1974 mean. Support for the Twentieth Century Reanalysis Project dataset is provided by the U.S. Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment (DOE INCITE) program, and Office of Biological and Environmental Research (BER), and by the National Oceanic and Atmospheric Administration Climate Program Office.

Even more than December 1933, we see extremely large heigh anomalies vis-à-vis mild, wet winter months. It is the extreme height anomalies combined with anomalous continental flow that made this month the driest of any month on record over almost all of climate divisions AK 9, AK 10, AK 11 and AK 12, whilst Barrow on the dry, frigid North Slope had its mildest month between November and March on record.

March:

In contrast to the EWP versus CET plot, March does not show any change from the winter months in its precipitation/temperature correlations over southern Alaska. As with October, the graph represents almost a straight line of positive slope, suggesting reduced skew in the temperature distribution but no change regarding the basic contrasts between wet and dry air masses.

April:

With the days becoming longer than the nights, and continental temperatures becoming hotter relative to maritime ones, we should expect that April would show a reversal or weakening of the consistent contrast of warn, hyper-humid maritime months versus frigid, dry continental months that are shown consistently over the East Gulf district between October and March.

In fact, even for April the correlation between precipitation and temperature (coefficient not measured) can be seen from the graph above to be positive. Nonetheless, it is weaker than the correlations we saw between October and March. However, the shape appears to show one key trait found for Juneau by Bradley Colman in the winter but not in the summer: a skewed temperature distribution with the median and mode warmer than the mean.

May:

Here at last we se a more definite transition to the typical hot-season pattern whereby hot months are drier and cool months wetter than the long-term mean.

What is noteworthy is that May gives no appearance of a transitional month, and outliers are not pronounced. Even the record hot and dry May of this year fits a line of best fit dating back to 1925 extremely well – indeed when you see the bullet in the top left, May 2015 fits the line as if there had been no radical man-made climate change as is demonstrated by rainfall and runoff data in southwestern Australia and parts of Chile.

June:

June, like May and July, behaves as one would anticipate from our earlier study of EWP/CET correlations like a typical summer month. Hotter-than-average Junes tend to be dry and cooler-than-average Junes wet over the East Gulf division.

As with May, there are almost no marked outliers, although the line of best fit is a little curved. this curved line, although predicted by Bradley Colman in 1986 for all months in the winter half-year, is emphatically not expected for a summer month. The differences between sea and land temperatures in the summer are less than in the winter months, and it stands tougher to get the southeasterly flow that would be needed for the hottest temperatures in summer, than it is to get frigid winter northeasterlies.

Conclusion:

Even without calculating Spearman ρ and/or Pearson r for each month, in the case of the Alaska East Gulf climate division – and almost certainly all of the maritime North Pacific – it can be concluded that:

1. a very strong positive relationship between temperature and precipitation is observed in the winter half-yea between October and March
2. a similarly strong negative relationship between temperature and precipitation is observed in the summer months from May to August
3. September does not show a significant correlation between temperature and precipitation
4. April appears to show a slight positive correlation, but it would be interesting to speculate whether longer records would show it as more of a transitional month
5. Vis-à-vis the EWP and CET areas, climate division AK 9 is similarly located but further north.
6. This more northerly location may explain why the reversal in correlation coefficients occurs earlier in autumn and later in spring.
7. The patterns of monthly relationships between precipitation and temperature in southeastern Alaska (north Pacific) almost certainly are analogous to those over the United Kingdom (north Atlantic). The difference in (6) is almost certainly replicated over Scotland.
8. Analogies between these two coastal regions are likely to be useful if topographic differences are taken into account.