Global Warming USA: the Long and the Short of it

Since 1895, the conterminous USA (lower 48 states) has warmed significantly:


The red line is a modified lowess smooth, on a time scale which mimics 20-year moving averages.

I decided to look for patterns in the geographical distribution of temperature change, using principal component analysis (PCA). The warming pattern of the 1st principal component (1st PC) matches the nationwide warming pattern excellently:

The warming pattern is on the right; on the left is a map showing how each of the 344 climate divisions match the 1st PC. Red dots means a match, with bigger dots meaning better match. The interesting thing is that the match is outstanding for the eastern half of the country, but not so great for the western half.

Does that mean that the western USA hasn’t matched the long-term trend shown by the national average? In fact, no.

PCA identifies patterns that are common to many of the climate divisions. The fact that the western half of the country doesn’t match so well isn’t because the western trend differs from the 1st PC, it’s because the fluctuations diverge. The eastern half of the country tends to show the same month-to-month fluctuations (these are all monthly data from NOAA, although I’ve plotted yeary averages for greater clarity), but the western half, in spite of showing a very similar trend, tends to show different fluctuations.

I’m primarily interested in the longer-term stuff, not the short-term month-to-month variations, so I decided to take a different tack. Instead of doing PCA on the base data (monthly averages), I first smoothed the monthly data for each climate division with a modified lowess smooth, using a time scale which mimics 10-year moving averages. Then I computed PCA on those values, to seek patterns which tended to be common among the “roughly 10-year moving averages” rather than the monthly data. This, I believe, will tell us more about how the trend patterns tend to group together, rather than putting so much emphasis on how the fluctuation patterns group together.

With that strategy, I got this for the 1st PC:

The 1st PC still matches the nationwide trend well. But now the geographical distribution of how it matches (the map on the left) is very different. Instead of an east-vs-west difference, it’s much more uniform across the country, demonstrating that the USA has participated in the nationwide warming more uniformly.

I do note that the match is considerably weaker in a region of the southeast, centered on Alabama and Mississippi. This is the location of the “warmhole” which I blogged about here. It appears that the strategy, to focus more on trends than fluctuations by using PCA on long-term rather than short-term data, has worked. It appears to have identified the most prominent pattern of deviation from the nationwide average behavior, i.e. the “warmhole.”

If we look at the 2nd PC, find something quite interesting:

There’s the warmhole again; the red dots show where the local pattern matches PC#2, while the blue dots show where it “anti-matches.” Looking at the time series pattern, we see that the warmhole so identified shows a sudden drop in the 1950s. This is in accord with what I found in my earlier post about the warmhole, in which (looking at it closely) the data suggested a sudden drop in 1958. Bear in mind that in the PCA, the time series pattern represents the difference between the “red-zone” pattern and the “blue-zone” pattern, so the warmhole cooled significantly in the late 1950s relative to most of the rest of the country (especially the north and the west).

Also of interest are patterns for different seasons. I’ll show just one now, the 1st PC for summer (June-July-August) temperature:

This illustrates that summer in the USA showed quite an outburst of hot temperatures during the 1930s, reaching a peak comparable to present temperatures. Other seasons show no such 1930s outburst; Autumn, for instance, shows strong warming over the last 20 years but nothing of the kind during the 1930s:

There are many other interesting things, including other seasons (winter and spring) and other principal components with interesting geographical patterns. I hope to post about those soon.


This blog is made possible by readers like you; join others by donating at My Wee Dragon.


Advertisements

18 responses to “Global Warming USA: the Long and the Short of it

  1. David B. Benson

    The 1930s were dust bowl times.

    • The lower 48’s Dust Bowl years are much beloved and oft’ cherrypicked by deniers. Even as there is evidence that parts of the whole problem were caused by human influences.

      Kinda’ makes one shake their head.

  2. Martin Smith

    This blog post illustrates a problem I often have when trying to understand your analysis or someone else’s. After reading the whole post, I have the sense that I understand argument, but at the same time I have an uneasy feeling because I don’t really understand some of the points made or terms used, so I have had to “fill in” my understanding in places by just assuming a point is correct or that whatever an unfamiliar term means, it too is correct.

    So I looked up PCA in wikipedia and tried to follow that explanation, but it is mathematical, of course, and I don’t understand how to apply it to a temperature dataset. What do components #1 and #2 represent in the temperature history for the geographic area?

    When you say “PCA identifies patterns that are common to many of the climate divisions,” does “climate division” refer to geographic regions?

    When you say ” because the fluctuations diverge,” are the fluctuations the differences from one year to the next and are they diverging from the trend? Or are the fluctuations different from each other in different areas of the western US?

    How would you then describe in everyday climate language what the difference has been between the eastern and western US? Is this close: In the eastern US, warming has increased steadily over the entire area, but more slowly in the southeast. In the western US, warming has followed a similar long term trend, but the year to year fluctuations in separate regions of the west have differed a lot from each other.

    [Response: Forgive me, I never know exactly where to draw the line between technical and not. My regular readers tend to be a technical lot (some are genuine experts); sometimes I forget that I hope to reach for a much wider audience.

    Yes, “climate division” refers to a geographic area of the U.S. NOAA (National Oceanic and Atmospheric Administration) defines 344 of them which cover the conterminous U.S.; each state is covered by about half a dozen or more.

    Yes, the fluctuations are the differences from one *month* to the next (since the base data are monthly) which aren’t part of the long-term behavior, just the ubiquitous up-and-down wiggling around. The long-term behavior, which could be called “trend,” follows a pattern which has persistence, but the fluctuations appear to be random, like the roll of the dice.

    Your description of “the difference has been between the eastern and western US” is pretty darn close. I would mention that it’s not the case that the western U.S. shows fluctuations that are all different — rather they tend not to follow those of the eastern U.S. which defines the 1st principal component. Their pattern of fluctuations will show a different set of commonalities which shows up in other principal components. The reason the easter US defines the 1st PC is that there are so many climate divisions there which show such strong commonalities.

    As for PCA itself, I did a basic intro here. Take some consolation in the fact that even if you’re a mathematician, understanding the math doesn’t guarantee true comprehension of what’s essentially going on. That takes a bit of experience and familiarity.]

  3. My crude-level internal rationalization of PCA involves a coordinate transformation. If we have n-dimensional space (i.e, n variables), is there a way to recombine these variables so that we can explain most of the variation with fewer than n variables – but new variables that are some combination of the old? PCA looks for that recombination and tells you how many of the recombined variables provide how much explanation.

    Take an example of an unfamiliar [equatorial] city where I am mapping vehicle movements, tracking longitude and latitude as a function of time. It’s a mess: every movement involves changing both latitude and longitude for each vehicle. I then do a mind-meld approximation of PCA on the data (i.e., I think about it), and realize that all movements share (roughly) a characteristic that each +latitude movement is correlated with an equal +longitude or an equal -longitude movement, and each -latitude movement is also associated with either an equal +longitude or -longitude movement.

    I then take a look at a map of the city and realize that most of the roads are oriented NE-SW or NW-SE – i.e., on a 45 degree angle from N-S/E-W. Suddenly the movement data makes sense – it is aligned with the roads, not with the four primary compass directions I was using to begin with. If I decide to reprocess my data, and track movement using a new coordinate system that has one axis running NW-SE (let’s call this axis “milk”), and a second axis running SW-NE (let’s call it “toast”), then nearly all of my vehicle movement is much more easily described in terms of my milk-toast coordinates. Most of the movements involve only changes in either milk or toast, and not both.

    Note that milk and toast are mathematical combinations of the previous compass coordinates, so I don’t really have new information in the data. It’s just that the new combination (in this case, a coordinate transformation) makes it easier to see the patterns and identify the dominant features of the data.

  4. Since you brought up the US SE “warming hole” again (comments on the earlier post were closed for a while), I wanted to bring up one thought I’d had at the time about the apparent shift in the late 1950s. Doing a little reading, it seems that the late 1950s was a time where the Mississippi River underwent some significant engineering work – namely the construction of the current Atchafalaya diversion project.

    I didn’t get far enough into finding data sources, but I wondered to what extent the use of this control structure has changed where and when the river flow is sent into the Gulf of Mexico. Could diverting large amounts of water into a different outlet to the Gulf have an effect on Gulf of Mexico salinity and surface temperatures – enough to affect SE US land air temperatures?

    • An ingenious speculation, but I’m guessing not, mostly because (as I understand it at least) the whole point of the diversion was not to create a new flow to the Gulf through the Atchafalaya, but to *prevent* an increase in the flow through it–something which was already tending to happen more and more often over time, and which would have had negative consequences for existing port facilities in places like Baton Rouge and, of course, New Orleans.

      https://en.wikipedia.org/wiki/Atchafalaya_River

      FWIW, the discharge volumes for the Mississippi and Atchafalaya are respectively 593,000 cfs (largest in the US, of course) and 225,000 (6th-largest, just ahead of the Niagara and just behind the Yukon.) So I’d think that, if the Atchafalaya ever does succeed in appropriating a Mississippi-sized share of the total, it would be a big enough swing to make some noticeable hydrological changes in the Gulf, all right–though I think knock-on climatological consequences on warmhole-sized scale wouldn’t be likely, based on nothing but intuition.

      Like you, I’ve wondered about potential anthropogenic factors, including changes in cotton production and changes in forestry (a big industry in the Southeast). But I don’t think the numbers support either speculation very well, though the data I’ve seen is pretty crude for this purpose.

      • Martin Smith

        What if the engineering work resulted in much improved drainage of the entire warmhole region so that the ground is more dry in general and there is less evaporation over the whole area?

      • Martin, if by ‘the engineering work’ you mean the Atchafalaya control project, then no, I can’t see any way that soil moisture over the entire warmhole region would be materially affected; the project was basically meant to keep Mississippi water flowing down the Mississippi, and not cutting an ever-deeper channel in the steeper and more direct Atchafalaya. I doubt it would have much effect on the drainage of the wider region at all, and to the extent that it could have, you’d expect that it would be of opposite sign–ie., that it would keep any still-extant riverine wetlands along the lower Mississippi from drying due to lower mean river levels. (I suppose there are some such wetlands, though in human-dominated areas (like New Orleans) the banks are highly channelized by levees (which, surprisingly, are not entirely human-made, though they have been ‘enhanced’ by humans.))

        On the other hand, maybe you meant some other work, either ‘instead’ or ‘as well’? Could there have been, for instance, a burgeoning of fields ’tiled’ for drainage? That’s a technology used in agriculture in other flattish places (I’m aware of it, for instance, in southwestern Ontario, where some of my family practiced it.) The whole point of it is to make sure fields can drain faster, so it would do just what you propose. But even after a bit of searching, I have little idea of the history of agricultural drainage in the region; further North, most of the work happened much before the 1950s.

        https://en.wikipedia.org/wiki/Tile_drainage
        https://en.wikipedia.org/wiki/James_B._Hill
        https://web.extension.illinois.edu/bioreactors/history.cfm

      • This paper gives a good historical overview of the various control structures at the Atchafalaya area. (The web site does not seem to be responding right now.)

        Joann Mossa (2013) Historical changes of a major juncture: Lower Old River, Louisiana, Physical Geography, 34:4-5, 315-334
        https://doi.org/10.1080/02723646.2013.847314

        In addition to preventing the entire Mississippi from cutting over to the shorter Atchafalaya channel, the control structures also maintain some water diversion to the Atchafalaya, and are used for flood control on the Mississippi.

        Last winter (when Tamino’s first “warming hole” post came out), I managed to find some flow data at the Atchfalaya gap. Prior to 1960, the minimum daily flow for a year was often between 100,000 and 150,000 cfs, and dropped below 75,000 cfs once. Since the early 1960s, it has rarely been less than 150,000 cfs. Maximum daily flow over the year doesn’t seem to have changed much. Annual average maybe has tended to increase(using the eyecrometer). Daily flow can vary from 75,000 cfs to about 1,600,000 cfs. That’s quite a range (and quite a lot of water!)

        I also recently found this paper:

        Spatial and Temporal Patterns of In Situ Sea Surface Temperatures within the Gulf of Mexico from 1901- 2010
        Jason Allard , John V. Clarke III , Barry D. Keim
        American Journal of Climate Change, 2016, 5 , 314-343
        http://dx.doi.org/10.4236/ajcc.2016.53025

        It has some interesting discussions of the variability of Gulf of Mexico sea surface temperature distributions, including discussion of water circulation variations. There are some interesting seasonal differences.

        In the earlier post, I had made this comment:

        https://tamino.wordpress.com/2018/02/20/us-warmhole/#comment-100959

        …which has another link to a paper.

        And for pure entertainment purposes, you can’t talk about the Mississippi and the Atchafalaya without reading this classic from New Yorker magazine:

        https://www.newyorker.com/magazine/1987/02/23/atchafalaya

      • Thanks for those links, Bob. You’re right, that New Yorker piece is a classic. And it suggests the possibility of increased danger to the extant structure of southern Louisiana via failure of the Old River Control and rechanneling of the mainstream into the Atchafalaya: what effect does increasing extreme precipitation frequency in the US have on the probability of the ‘Design Flood’ magnitude being exceeded?

  5. It would be interesting to compare 1930s warming patterns with the 1930s pattern for “John Deere’s” tractor and plow sales. Rain follows the plow. Farmers believed that. If seedlings died because of drought, they plowed again. Plains states were covered with drifts of dark dust.

  6. What do you make of the graph in the comment response here: http://euanmearns.com/energy-and-man-part-3/ ?
    I don’t understand why/how the originator of the “net energy cliff” graph and main author on the old Oil Drum site can be in such denial.

    • Perhaps I missed the ‘good old’ Euan Mearns, but what I have read of his work has consistently struck me as extremely biased. Consider a few sentences such as these from the link you provide:

      The new international sport is bidding for the highest CO2 reduction targets and levels of renewable energy penetration. Most of those involved do not know what energy is let alone how important it is to have affordable supplies delivered to individuals and companies when they most need it. Those leading the pack with 100% renewables targets in the near future – Scotland, Denmark and Germany – have not studied the implications of their strategies on the needs of people and society. Or if they have done, these studies have been conducted by the renewables advocates. A simple case of lunatics in charge of the asylum. Alternative strategies have not been considered.

      I think it’s obvious on the face of it that that is pure fantasy.

      From there it seems to be just a few steps to pure climate change denialism–reached long before you get to the comments (which–full disclosure–I didn’t). For instance:

      The Arctic sea ice canary is refusing to die. Whilst it seems inevitable that 7 billion souls must impact Earth’s climate and more critically a myriad of terrestrial and marine ecosystems, the impact on climate to date looks increasingly benign. As time passes the failure of OECD government policies to control global emissions and the failure of increased emissions to raise global temperatures will appear increasingly asinine.

      That’s not worthy of respect–and again, from my limited experience of his writing to date at least, it’s not all that new or different for Mr. Mearns. For past examples, he wrote ‘hit pieces’ about the Syrian drought, which he contended *based purely on (limited) precipitation data* didn’t happen, and about renewable energy in Europe. That latter was extensively mooted here between me and the late Ed Greisch, who eventually earned a tongue-lashing from our host for promulgating ‘bullshit’. To the best of my recollection, the claim was that wind power produced zero power across all of Europe for an entire day, whereas it transpired that the supporting data showed that for about four hours half-a-dozen or less European nations (not even including all the biggest wind producers) had had very low output (say 20 MW or something like that). Very, very overstated conclusions, at best; willful deception, at worst.

    • Energy Matters is an odd site. There is some very good analysis there, and a willingness to go back to data basics which is often enlightening, but it’s combined with embarassing denial, or at least strong lukewarmism. And quite a few of his commenters are in complete WUWT-grade denial. There is a lot of useful analysis of the genuine difficulties of decarbonisation, but mixed with the strange idea that it’s not actually needed anyway.

      The denial devalues the good stuff.

      • It’s rather illuminating, IMO, to compare Mr. Mearns’ piece on the Syrian drought–or as he would claim, ‘non-drought’–and the 2012 paper he used as source for PDSI values.

        http://euanmearns.com/drought-climate-war-terrorism-and-syria/

        http://ageconsearch.umn.edu/bitstream/155471/2/2_Al-Riffai.pdf

        A couple of illustrative points:

        1) Mearns rationalizes an exclusion of the arid Zone 5 as not possessing significant cropland, despite the fact that Al-Riffai (2012) shows that that zone produced fully half the value of wheat and vegetables as the moist Zone 1. (Of course, irrigation is a necessity in Zone 5, but that does not mean that drought has no impact–and by a strange coincidence, the PDSI value for Zone 5 was at an all time low for that zone. So Mearn’s exclusion also deliberately suppressed the worst case data.)

        2) Al-Riffai also speaks about the 2008 drought Mearns wishes away in several places. Perhaps most damning is this sentence:

        “The International Disaster Database of the Center for Research on Epidemiology of Disasters (CRED, 2009) ranked the droughts in 1999 and 2008 among the top 10 natural disasters in Syria since 1990.”

        Mr. Mearns chose to ignore that completely. And that’s too bad; had he followed up on that clue, as I did, he would have found some much more granular information here:

        https://reliefweb.int/sites/reliefweb.int/files/resources/23905_droughtsyriasmall.pdf

        The provinces primarily affected by poor rainfall included the top four wheat producers which account for 75 percent of total wheat production in Syria (Al-Hasakah, Ar-Raqqah, Aleppo or Halab, and Dier ez-Zor). Rainfed wheat area in these provinces normally amounts to more than 800,000 hectares, and is extremely reliant on timely rainfall during the growing season. Favorable rainfall in April and May are typically critical to
        successful growing seasons, and this year non-irrigated crops were already failing in late March. April rainfall was extremely low throughout northern and northeastern wheat regions this year, causing even greater moisture stress and decimating crop yield potential, (USDA, 2010).

        (p. 16)

        Al-Riffai, and hence Mearns, only presented annual PDSI averages, which obviously fail to capture shorter-term variability. The discussion following, and especially the tabulated data from the Vegetation Health Index–a remote-sensing data product–paint a very, very different picture than that which Mr. Mearns limns. (It must be noted, the editing in this part of the report is very bad, with English that at various times hardly deserves the term.)

        Perhaps most illuminating is this tidbit:

        Rainfall in eastern Syria fell to 30 percent of the annual average in 2008 – the worst drought for 40 years –and al-Khabour, a main tributary of the River Euphrates, dried up… The country’s agriculture sector, which until recently employed 40 percent of Syria’s workforce and accounted for 25 percent of gross domestic product, has been hit badly, but farmers themselves are worst affected… Poor and erratic rainfall since October 2007 has caused the worst drought to strike Syria in four decades. Approximately one million people are severely
        affected and food insecure, particularly in rainfed areas of the northeast – home to Syria’s most vulnerable, agriculture-dependant families.
        Since the 2007/2008 agriculture season, nearly 75 percent of these households suffered total crop failure… Depleted vegetation in pastures and the exhaustion of feed reserves have forced many herders to sell their livestock at between 60 and 70 percent below cost. Syria’s drought break point was the season 07/08 which extended for two more seasons, affecting farming regions in the Middle north, Southwestern and Northeastern [parts] of the country, especially the
        northeastern governorate of Al Hassakeh.

        (p. 26)

        What that shows, I think, is that Al-Riffai inadvertently suppressed drought effects in the 2008 event by in effect averaging across affected and non-affected areas of the same zones. That is, the drought was catasptrophic primarily in the east, whereas the climate zones stretch mostly east-west. For example, the eastern stations Mearns considers comprise Deir Ezzor and Palmyra, for which Mearns dutifully reports large negative rainfall anomalies of -31% and -22%, as well as Kamishli in the far northeast. For that last station Mearns reports a positive anomaly of 3%, but conveniently omits to mention that, according to the graph he presents, that’s mostly due to a very wet season in 2011, and that in 2008 the station recorded no significant rainfall at all.

        How could he get it so wrong, ignoring such widespread reporting and testimony, and failing to uncover easily available evidence to the contrary? I can only believe that he was eager to get to the conclusion he wanted:

        We have established that there was no drought of any unusual significance in Syria between 2006 and 2011, that climate change did not cause the crop failures which resulted in millions of farmers fleeing to the cities or that they triggered the Syrian uprising when they got there. The claim that refugees from Syria are in any way, shape or form “climate refugees” is therefore entirely without foundation, as is the claim that man-made climate change had anything to do with the Syrian civil war or the rise of ISIS…

        Climate alarmists are becoming progressively more strident, unscientific and indecorous in their attempts to get their message across to a largely disinterested public, but this surely marks a new low.

        Personally, I find that absolutely despicable.

  7. Sorry for the blown endquote tag there, but presumably it’s clear that the last sentence is mine, not Mearns’.

    [Response: Fixed.]