ONI and individual state temperatures

The influence that sea surface temperatures have on land temperatures and climate is an ongoing debate, and I mentioned this somewhat in my last post on this topic. However, in looking back over that post it was perhaps too general an approach to look at the impact of the El Niño events over the large scale of the West Coast, and thereon East. Given that effects were, as I showed, more regionalized and remembering that we are in a La Nina winter, I’ll just repeat the anticipated effect plot from Kumar, et al.

Impacts of a La Nina winter (after Kumar et al).

The disadvantage of using the regional average can be seen just in the US West Coast. At the upper end the season is cooler and wetter, while down in the south it is drier and warmer – the average might well be “no change.” So since we have the individual state temperatures for the period I looked at last time, I thought it might be interesting, before looking at other Oscillations, to just check how this event correlated on a more localized basis.

The ONI plot from 1950 (GGWeather )

It is logical to start on the West Coast, and given that the predicted impact this winter will be on the upper Northwest, the first comparison is with Washington State. For now I am going to use the homogenized data set, rather than the TOBS data, though I may come back later to look at how that changes things. (This is the fun of doing this without an agenda, we don’t need to have the data fit any pattern, so it is more informative to look at options).


The ONI plot overlain on a plot of Washington state temperatures.

It can be seen that while there was some correlation, in places, overall the agreement is not very good.

The other states that seem to be most impacted are the southern tier, that would include California, Arizona, and possibly New Mexico and Texas . From the regional comparison the Pacific SST effects seem to weaken somewhat once one gets over the Rockies, hence the caution as to how far we might expect the impact along the South Coast – so we shall see.

Turning first to California, recognize in the beginning that with the state being as long as it is, there are internal temperature variations along the state, Overlaying the ONI plot on the relevant part of the California temperature curve:

Relation of California temperatures to ONI temperature anomalies

My sense is that the correlation is a bit better, but still lacking. So let’s try the Arizona comparison:

Relation of Arizona temperatures to ONI temperature anomalies.

There does seem to be more of a correlation here, than with the earlier comparisons.

Moving on to New Mexico, and the same superimposition:

Relation of New Mexico temperatures to ONI

Well what correlation there was in the first states seems to be getting less here, lets try Texas.

Relation of Texas temperatures to ONI

Well, using that well known calibrated eyeball, it would appear that the correlation seems to get worse as one moves away from the Pacific.

Well this wasn’t totally what I was expecting, though I mentioned at the top that I suspected that the effect might not reach as far as New Mexico. so I think I will cogitate a little more on this before venturing an opinion. However it might be worth looking at relative precipitation levels, since this seems to be more the effect that is most obvious. (Though that also gets into cloud formation . . . . . . )

A new book and the temperatures in the Mountains

One of the many significant salient points that Donna Laframboise makes in her new book on the IPCC lies in the predilection of the IPCC authors to rely on computer modeling over factual data. I will probably have more to say on this in my planned review of the book (which so far is fascinating to read and damning in the details) once I finish it, but it does suggest that posts such as these on the actual variations in US temperatures over the past century will languish in outer darkness – oh, well!!

To recap where we are, after looking at the temperatures of the individual contiguous states of the American Union, I have started to look at regional changes and to compare them. A significant drop of around four degrees Fahrenheit that occurred between 1950 and 1965 along the Eastern Seaboard, and which inter alia led to a southern migration of the black-capped chickadee, has since been reversed in the warming of that region post-1965. In contrast in the middle of the country there has been sensibly no warming trend since 1895 (which might explain why the Governor of Texas finds it harder to believe in global warming, since his state hasn’t seen any).

Yet on the West Coast the trends show a relatively consistent increase in temperature since 1895. And so the next place to look is to see what happened in those states that lie along the Rockies and provide the mountainous barrier between the West and the Middle. In the following post I am going to derive this graph, and comment along the way as I get there.

Average variation in time for four regions of the country, with the results adjusted as shown to separate the curves, and show them in order (bottom to top) from West to East.

For this exercise I am going to include Montana, Idaho, Wyoming, Nevada, Utah, Colorado, Arizona and New Mexico . I have previously used both the homogenized data from the USHCN and the Time of Observation corrected (TOBS) data, with a preference for the latter since, as I noted with the individual state evaluations, it has shown a consistently better correlation across the states with latitude and elevation than the data after manipulation.

Combining first the average temperatures for the states, using the homogenized data, one gets a relatively flat curve, but one with a decided kick over the last thirty years, the sign that many consider the marker for Anthropogenic Global Warming.

Average of the state temperatures for the Mountain Region, averaging the values for the different states in the region.

However, before there is much comment on this, it should be noted (when this is compared with the figure above it) that that trend is not as clear in the four regions as I ultimately graphed them. But before getting to that let me just put up a set of comparison graphs, using the USHCN homogenized data, to show that the temperatures of the eight states that I looked at appear to vary relatively consistently. By setting them one above the other it should be easier to cross-compare.




These are relatively consistent, though they appear to smooth a bit with movement south, but to get back to that kick in the end of the plots. As you may know, and I have commented in the individual state plots, the USHCN data is “homogenized” from the raw data. When one takes the average of this “manipulation” and subtracts the Time of Observation corrected values from it, then one gets this curve as the average change made by the climate scientists in reporting the values for the stations in the evaluation (all 233 of them).

Difference between the average USHCN homogenized temperatures and the raw temperature averages, as corrected for time of observation (the TOBS values).

Given that, the rest of the analysis will go back to using the TOBS values. Firstly this changes the average a little:

Average temperatures for the Mountain states, using the Time of Observation corrected raw data.

The change in the slope of the curve drops the rise in temperature per century from 1.7 deg F to 1.08 deg F.

However this curve was obtained by just averaging the average state temperatures, there are varying numbers of stations and varying areas for the different states. When one weights the value by the relative number of stations in the state (equivalent to averaging all the stations in the region), then one gets:

Average temperatures for the Mountain states, using the Time of Observation corrected raw data and averaging the 233 stations in the region.

In each state, the area that each station “covers” (i.e. the area of the state divided by the number of stations) varies from 8,500 sq miles in Nevada, to 2,123 sq miles in Utah. To try and account for this I have weighted the state values by the areas of the states, and this gives this plot:

Average temperatures for the Mountain states, using the Time of Observation corrected raw data and weighting the state averages by the area of the state.

With this plot the steadily increasing temperature suggested in the USHCN plot does not appear as evident, and the temperatures since about 1990 seem relatively steady.

So how do the regional values now compare?

Comparison of average temperatures for the different regions of the United States (as I have defined them)

Obviously the mountain temperatures are lower than those of the other regions given that there is a fall in temperature with elevation. For the combined region this looks like this:

Variation in state temperature as a function of average station elevation in the state.

The r-squared value is much lower for this than it is for the individual stations in the region, and so I will have to look at the effects of latitude, elevation and population separately.

But to conclude I wanted to see how the temperature patterns changed across the country, and the overlay of the regional temperatures made that a little difficult to see, so I “shifted” the curves by adding and subtracting temperatures from each set, so that the comparison of shapes could be made, and that is where we came in:

Average variation in time for four regions of the country, with the results adjusted as shown to separate the curves, and show them in order (bottom to top) from West to East.