Temperature drop with elevation

This morning I visited Watts Up with That and read a post on the accuracy of the new set of temperature stations around the country. It points out that the new set of stations show, on average, a lower temperature than the record claimed by Jim Hansen.

In his rebuttal to this Nick Stokes points out that the average elevation of the two sets of stations is different and that when that is taken into account, allowing a temperature drop of 6 degC/km then the difference between the two sets of data is explained.

However the original network had an average elevation of 1,681 ft (512 m) and the new network has an average of 2,223 ft (667.6 m) giving a difference of 155.6 m or an anticipated temperature difference of 0.1556 x 6 = 0.93 deg C.

Well, in the state data that I calculated some time ago I plotted the temperature against elevation and so I quickly (too quickly as I found out) went through the various posts for the different states shown undeer the Follower pictures on the right hand side of the page, and listed them with the average slopes of the graphs I had plotted. Ran an average, and posted the result on WUMT.

Turns out I made a slight error since I plotted the temperature in deg F and the elevation in meters, so that the correlation I got was in bastard units. And I mis-entered one of the data points into the table. It translates, when done correctly into an average decrease of 0.016 deg F per meter, or 0.0089 deg C per m, 8.9 deg C per km, 50% higher than that quoted by Nick.

For those interested I have tabulated the average slope against average state elevation and then plotted it, just for grins.

Figure 1. Temperature trend with average elevation for each of the Contiguous United States

What is interesting (apart from the two negative values for North and South Dakota) is how the trend shifts as one gets closer to sea level. I did not plot this relative to their actual distance from the sea, but it reinforces my conclusion that the sea temperatures have more of an influence than we are giving them credit for.


There is no more

Delaware combined temperatures

The last post in this series looked at the temperatures for Maryland and so, moving up along the coast, the next stop is Delaware.

Delaware USHCN stations (CDIAC)

Given the small size of the state, I thought it might also be interesting to compare the results with those for the GISS station in Washington D.C., since the latter would otherwise be left out. They are at about the same latitude, and of somewhat similar elevation, differing only in the size of their populations.

Temperatures as reported for the GISS station in Washington D.C. (GISS )

Given the built-up nature of the area around Wilmington there could be some debate as to the relative population sizes about the various stations, but for the moment I will accept the values from the citi-data sites that I have used to date. (The question is raised particularly regarding the Newark University Farm, which GISS considers to lie within metropolitan Wilmington).

It turns out that Washington is, on average, about 2.76 deg F hotter than the average for Delaware, though the difference has changed, with a steady increase until around 1980, and a fall thereafter.

Difference between the temperature reported for the GISS station at Washington DC and the USHCN average homogenized temperature for Delaware.

Looking at the overall change in temperature over time for Delaware alone, there is still that drop in temperature that occurs between around 1948 and 1965:

Average temperature for the USHCN stations in Delaware after homogenization.


Before homogenization, however, looking at the Time of Observation adjusted raw data, the trend is not as significant:

Average temperature for the USHCN stations in Delaware raw data after correction for time of observation (TOBS).

The temperature drop from around 1950 to 1965 is still present, but the overall temperature increase has fallen from 1.9 deg F per century down to 0.5 deg F per century.

Delaware is the second smallest state (after Rhode Island) and is only 100 miles long, while 30 miles wide. It stretches roughly from 75 deg W to 75.75 deg W, and from 38.5 deg N, to 39.8 deg N. The mean latitude is sensibly 39 deg N, the average of the USHCN stations is 39.3 deg N (D.C. is at 38.85). The state elevation runs from sea-level to 137 m with the mean at 18.3 m. The average of the USHCN stations is at 28.6 m.

The small number of stations makes the correlation coefficients of little real value, but they are included for consistency. It will be interesting to see how these numbers fit in when I compile the overall statistics.

Change in average station temperature in Delaware as a function of latitude.

Change in average station temperature in Delaware as a function of longitude.

Because of the relatively small change in elevation for the different stations, the correlation with elevation is not as evident here.

Change in average station temperature in Delaware as a function of elevation.

When I looked at the TOBS data available there is insufficient recent information to provide a realistic plot of temperature against local population unfortunately, but I’ll include the plot for consistency.

Change in average station temperature in Delaware as a function of population around the station.

It is clear in Delaware, as elsewhere, that the homogenization of temperature data, has led to an increase in reported temperatures with time.

Increase in temperature from the TOBS data to the reported homogenized temperatures for the stations in Delaware.

West Virginia combined temperatures

This series, which has been looking at the temperatures of the different states around the country, is now beginning to approach an initial end, as I look at the data for West Virginia, which IIRC is the last state to complete that does not border on the Atlantic. (Foregoing for the moment looking at Alaska and Hawaii) West Virginia comes after looking at a few states along that littoral, ending last week with North Carolina. One of the features that was clear in those states was that there was a significant drop in temperature around 1950, while on the other side of the mountains, there were a number of states where the temperature has been falling since 1900. It will be interesting to see which of these categories West Virginia falls into.

There are thirteen USHCN stations in West Virginia, and three GISS stations on the list, Beckley which only carried data from 1951, Charleston which only has data from 1949, and Huntington which also only has data since 1948.

Location if the USHCN stations in West Virginia (CDIAC )

In passing it is interesting to note that this is the second week that those going to the USHCN site are faced with this comment:

PLEASE NOTE: CDIAC is currently experiencing hardware/software issues that are affecting the performance of the USHCN web interface.

We hope to have these issues resolved shortly. Accessing data via the FTP area is working normally. We appreciate your patience and understanding. (7/14/2011)

And when I try and access data for the different stations, I cannot. However, having had similar trouble earlier in the summer (starting when I was working on Arkansas in April. That hiatus lasted through Louisiana but was cleared up by the time I got to Wisconsin so that the site was down for about three weeks. This time I had already acquired the West Virginia data, and so the post can proceed.

Taking a quick look at the GISS stations, it is, of course, now no longer possible to see the temperatures in the 1930’s when, until recently, it was recognized that the USA had it’s highest temperatures. Nor is it as easy to monitor for this state the transition from the possible rising temperatures pre-1950 to the possibly falling ones of the 50’s and 60’s.

Temperature change with time for Beckley WV as recorded at the GISS station

Temperature change with time for Charleston WV as recorded at the GISS station

There is a drop in temperature for this station, which is more distinct than that for Beckley.

Temperature change with time for Huntington WV as recorded at the GISS station

This shows more clearly the peak and fall of temperature until about 1965, followed by a weak rise which we have seen in the states to the South along the Atlantic.

Looking at the average USHCN temperatures, using the homogenized data set, there is relatively little temperature change for the state over the century, though the drop and then recovery of temperatures shown in the GISS data also holds for this series.

Average temperature in the state of West Virginia since 1895 using the homogenized temperatures from the USHCN.

The temperature rise in that time has been 0.04 degrees per century, which is hardly significant. Looking at the Time of Observation corrected raw data for the state:

Average temperature in the state of West Virginia since 1895 using the raw temperatures modified to account for the time of observation, from the USHCN.

West Virginia is 240 miles long and 130 miles wide, running from 77.67 deg W to 82.67 deg W, and from 37.17 deg N to 40.67 deg N. The mean latitude is 38.6 deg W, that of the USHCN stations is 38.8 deg N, and that of the GISS stations is at 38.17 deg N. The highest point in West Virginia is at 1,482 m at Spruce Knob, while the lowest is at 73 m on the Potomac. The average elevation is at 457.2 m. The average elevation of the USHCN stations is 369.8 m, while that of the GISS stations is 364.8 m.

All the stations save Pickens WV were in the citi-data set, for Pickens which was too small, I had to get the population (125) from Zip-codes.com.

Looking at the effect of geography on temperature for the state:

The effect of station latitude on temperature in West Virginia

Looking at the effect of longitude, as I have mentioned several times earlier, this is an artifact of elevation changes, included now only for completeness.

The effect of station longitude on temperature in West Virginia

As one sees the temperature, which fell with longitude on the other side of the mountains, now rises with longitude as the mountains reduce in size to the west. The correlation is with elevation:

The effect of station elevation on temperature in West Virginia.

Looking at population, the GISS stations, as is the apparent custom in the states around the Union, are located in cities with the largest population of the stations recorded for the state. The average population is 38,718, while the population around the USHCN stations is 9,872. (That would make a difference of perhaps 0.6 deg F). As a reminder the citi-data information on population is relatively recent, and I have correlated it with the average of the temperature at the stations over the past 5-years.

The effect of adjacent population on the station temperature in West Virginia.

Finally a look at the effect that homogenization of the data has on the average values:

Changes in average station temperature as a result of homogenizing the data.

The impact is focused around 1955, before then the homogenization increasingly added temperatures the further from that date, and similarly from that time forward. Wonder why?

Kansas combined temperatures (revisited)

As I move around the states looking at the differences between the GISS station temperatures, the USHCN homogenized temperatures and the original raw temperature data (albeit modified by a correction for the Time of Observation (TOBS), we have returned to Kansas, which I had originally written about before the TOBS data became available, and then written again looking at that data.

Since then the style of the presentations has changed a little bit, so I am going to redo the Kansas post, and just put the combined temperature reference into the list down the right side of the site, so as to bring it (and later some of the other states) into the same format as I have now evolved. So to begin , Kansas has 31 USHCN stations, that are reasonably evenly distributed around the state:

USHCN station locations in Kansas (USHCN)

Back when I did the initial TOBS data analysis I was still sufficiently naïve that I did not realize how, by manipulation of the station picks, GISS could manipulate the temperatures that it reports. Chiefio has explained both how the current selection raises the overall U..S.A. station record by 0.6 deg C, over the USHCN average, by selective deletion of stations, for 2008, for example. There is also a new set of 59 GISS stations that have been added, but these do not have the historic data that I have been comparing to, so I am going to ignore those additions for the rest of this series. However there is one additional point that this study has brought out about how NASA shall we say “fudges” their data.

I first mentioned, when looking at Idaho’s temperatures, the GISS habit of marrying a short-term station with a long-term one. E.M. Smith (Chiefio) has done a detailed post explaining how, by using this type of combination it is possible to generate a trend that does not exist in the initial data. What has since been interesting, as I have continued with this series, is finding just how many state temperature situations that applies to. And the first one that I noticed, but not then realizing the reason, was Kansas, the second state that I had looked at There are 5 GISS stations on the list, including Witchita , Topeka, Concordia,, Dodge City and Goodland. It is Goodland that only has data since 1948.

Goodland KS GISS station data

Because of the small number of stations, the impact of the partial temperature record is significant on the relative difference between the GISS and USHCN average values (using the homogenized USHCN data for this)

Difference between the average GISS station temperature and that of the homogenized USHCN stations

Turning to the TOBS data, and looking at the average temperature change in the state, over time, the temperature rise is 0.85 degrees per century (the homogenized data would suggest a rise of 1.2 degrees.)

Average TOBS temperatures for Kansas (USHCN)

The geography of Kansas is that it is 400 miles long and 210 miles wide, running from roughly 94.5 deg W to 102 deg W, and 37 deg N to 40 deg N. The highest point is at 1,231 m, and the average elevation is at 609.6 m. (The average USHCN station is at 511 m and the average GISS station at 586 m).

Looking therefore at the variation in temperature with the geographical parameters of the state:

Effect of station latitude on recorded temperatures in Kansas

This is a state where the effect of latitude is clearly defined. Given that the state elevation is steadily rising to the West, as we move towards the Rockies it is not surprising that we see:


The stronger and real correlation is with the changing elevation



There was not, in the original search for data, much problem in finding population numbers

The effect of current population size on average station temperature for Kansas

Note that for the curve above, because it is the only source of information for the larger cities in the state, I have included the GISS station data.

And finally there is the difference that shows the effect of USHCN homogenization of the data relative to that recorded and adjusted only for time of observation.


I am replacing the original two posts of Kansas on the comprehensive list with this post so as to give a greater consistency to the formatting.

The Idaho combined temperatures

There was a blizzard, so I gather, outside our hotel in Portsmouth, NH last night. But we went to bed after driving in as the snow fell lightly, and by the time we left this morning the lot was plowed, as were the streets and highways. It was only when we had to guess where the driveway was up to the house (and got it wrong) that the depth of the snowfall became really evident. (It took 45 minutes to get us dug out). Which is a good reminder that it is time to get back to the changes in temperature over the years. And I was looking at the other side of the country, where they also know how to deal with large snow falls.

Idaho has 29 stations in the USHCN, and a quick glance at the distribution suggests that there are more stations in the southern half of the state, than vice versa.

Idaho weather station locations (USHCN)

Downloading the data, as previously, into a spreadsheet, and noting that indeed Idaho is a relatively elevated state, I turn next to the GISS data, and there are, according to Chiefio, two GISS stations now being used in Idaho, at Pocatello, and at Boise. Pocatello just has data since 1947. A quick check on the GISS site says that the next station is the one at Aberdeen, which is 19 km away, so this is the right one.

Pocatello Municipal Temperatures

.Boise, on the other hand, has a full set of data from 1880 on.

Boise ID GISS station temperatures.

Interestingly it also shows that 1934 was a lot hotter than now. Incidentally there has been some discussion at both Climate Audit, and WUWT relative to the adjustments that GISS make to their data. You have to dig deep into the comments at the WUWT story to get the explanation as to why GISS values for the years keep changing (Eric Smith comments at 9:42 and 9:55 pm and then at 10:09 am on Dec 26th) The latter points to a post wherein Eric Smith (Chiefio) shows how data manipulation can keep temperatures rising, after a record temperature has been set and not broken). As part of that discussion, there is the comment that GISS is tending to marry a short-lived station (post 1930’s) with a longer one in order to emphasize the recent rising temperatures which are more evident with the short term station than the longer term. We saw this pattern in Oregon, and here it is again.

Turning back to this little study, after downloading the temperature records and then going through the locations to find populations, I have to use Google Earth to find Arrowrock and note that it is on the side of a dam:

Arrowrock station by a dam.

So I give this a population of 100 – though that is not strongly defendable. Yet looking at earlier plots of population, the dam sites seem to lie consistently above the line, suggesting it is equivalent to a larger population than that picked.

Bern also needs me to go look on Google Earth since it is too small for citi-data (which seems to only related to places at a level of more than 100 folk). There are a couple of dozen houses that are close, so again I’ll use 100 as the population.

Dworshak Reservoir has 305 according to a reference though again too small for citi-data.

Fenn similarly, where a site gives a population of 40.

Lifton has about 6 houses, and what may be a boat dock, and some apartments (via Google Earth – not much else) so I put down a population of 50.

Porthill, which had a population of 126 in the 2000 census, now has no-one living there. But since someone is reading the numbers I will put down 1. (The numbers from citi-data –see, for example, Priest River are from 2009.

In regard to the geography of Idaho the state is 475 miles long (N-S) and 305 miles wide. The Longitude runs from 111 deg W to 117W, and from 42N to 49 N. The highest point is 12,662 ft, (3860 m), with a mean elevation of 5,000 ft (1,524 m) , though the lowest point in the state is at 710 ft (216 m). The average station is at 1,087 m (including the GISS stations). There are 6 USHCN stations above average elevation, the GISS stations are not.

So how do the temperatures of the state compare? Starting with the difference between the GISS stations and the homogenized USHCN, the average difference, over the century, is that the GISS stations are showing a temperature 4.42 degF hotter than the average for the USHCN.


Difference between the average temperatures recorded by the GISS stations and that of the homogenized average data from the USHCN stations in Idaho. (Note the change when the second GISS station record is added).

Looking at the overall temperature change in the state over the century, and switching to the TOBS data set, since this does away with the adjustments that occur with homogenization.


The temperature rise for the TOBS modified raw data over the period has been at the rate of about 0.8 degF per century, whereas the homogenized data suggests that the rise has been on the order of 1.5 degF over the same period.

Turning to the impact of geography on temperature, the state is more sparsely populated than many in the study vide the amount of extra searching I needed to do to find the populations of some of the hamlets), but let’s see how that plays out.

The effect of latitude:


Against the general trend here there is an increase in temperature as one goes North. Perhaps an artifact of the locations and their elevation?

There is the same contradiction to the overall trend in the country with the plot of Longitude:


The strong correlation with elevation is, however sustained, and may have had influence on the two plots above:


As I noted earlier the stations in Idaho seem to be located in more places with very few residents than I have found in other states.


Yet there remains a good correlation with population across the spread of the plot.

Oh, and yes,

Massachusetts combined temperatures and a summary

With New York completed last week I have actually got a collection of states that runs from one ocean to the other, but to make the Eastern connection more significant I am going to look at the temperatures in Massachusetts this week. Given the number of stations that there were in New York, it is a bit of a relief to find that there are only a dozen USHCN stations in the state. There is also, according to Chiefio’s list, only one GISS station in the state, in Boston. So that should make the initial data download relatively painless.

There was one snag in finding the Blue Hill Observatory, but it turns out (using Google Earth) that it is in Canton, MA, whose data I used. Having driven West from Boston in the past, I remembered that the road climbed almost all the way to the New York border. And, yet there seemed to be few stations with higher elevations, so I looked a bit more closely and there were only 2 stations in the western half of the state, and only two whose elevation was above the average. The rest of the stations had an elevation below 65 m, while the average elevation of the state is at 150 m. Given, as I will show that there is a strong negative correlation between temperature and elevation, the station locations are weighted to those closer to the sea.

Having said that I am going to use the TOBS plots to show how the temperatures have varied in the state over the last 115 years, and to look at the usual suspects (Latitude etc) to see how these affect the numbers. And because I have gone from one coast to the other, I thought I would tabulate some of the data that I have been plotting.

So firstly, how accurate is the GISS temperature in modeling average temperatures in the state.


As you can see, even with the average being taken from stations at lower altitudes, reading the GISS temperature at Logan Airport in Boston, which is sensibly on the sea, gives a temperature that is, on average 2.48 degrees warmer than that average, though it is going down a little.

For the state, as a whole, the temperature has been rising over the past 115 years, at a rate of around 2 deg per century.


However the temperature rise seems a little steeper pre-1950 than it has been since. The homogenized data that the USHCN generates suggests that the temperature rise has been somewhat higher, at 2.77 deg per century, with a higher R^2 of 0.43.

Looking at the effects of station location, the change in latitude continues to have an impact, though not nearly as significant as it has been in other states. This could, in part have been due to the two highest average station temperatures being for stations relatively close to sea-level. At the same time there are not that many stations in the state. (Interestingly the homogenization improves the R^2 value in this case, to 0.122, with a higher coefficient of -1.98, which is interesting if one looks at the table toward the end of the post).


The drivers of the eastern most stations being on the coast, and those further west being higher, means that there was a strong correlation with longitude:


Although the correlation is quite strong, I still feel that this is more due to the other factors that just longitude, and though the correlation with elevation is not quite as good (because of the higher temperatures of the two stations down at the coast) yet that also provides much of the explanation.


The homogenization of data reduces this correlation down to an R^2 of 0.33, one of the problems, I am beginning to think , of the process that they are using.

And the correlation with population continues to exist – note that the two highest temperatures are with the two largest cities.


Having now examined the states from Massachusetts to California, I thought to put the coefficients (as in temp = a x property + b, and tabulating the values of “a” and the R^2 values for the different states along the line. For population values, it should be remembered that the property has been converted to a log value first. (see the figure immediately above).


It can be seen that there is some consistency in the values for latitude, elevation and population, but not as consistent a set with longitude – bearing out my conjecture that there are other factors in play that influence the longitudinal correlation.

Oh, and one final thought – I did plot the difference between the homogenized data that the USHCN provides and the TOBS raw data, here it is.