The California Urban Heat Island Effect

Last week Anthony Watts had a post at WUWT in which he talked about a new effort to find out just how much the urban environment was affecting the temperatures at Californian weather stations. The study is being carried out in conjunction with the EPA, and the announcement came by e-mail rather than a more conventional press release.

I was interested since, as part of series that I carried out looking at the US Historic Climate Network (USHCN) data, I plotted temperatures for the stations in each state as a function of latitude, longitude, elevation and local population. The first three values were identified with the information at each station. The local population for a town can be found on the web in several different places, and very largely I relied on the city-data web sites for information (see, for e.g. this for Sacremento).

The question arose as to which particular temperature should be used for that of the station, since the USHCN provides annual average temperatures, as raw data, Time of Observation (TOBS) corrected and “adjusted.” When the original post for California was written, only the last of these was available, and thus it formed the basis of the analysis. Shortly thereafter, in 2010, the USHCN site also provided the raw data, and the TOBS temperatures for each station, each year. The data was therefore re-analyzed using the TOBS values. But the plot that was originally generated was plotting the current population against the average temperature since 1895.

As the study grew to include more states, that plot seemed to be an error, since populations can change very rapidly, and go up as well as down. So, towards the end of the series the average temperature was taken only for the past five years, since this was likely to reflect the impact of current populations. At the same time, since there is little difference between the two sets of values in this period, the “adjusted” values were used to derive the plot. It looks like this:


Figure 1. The comparison of average California station temperature plotted relative to adjacent population, with a log-normal plot.

Now the “discovery” of a log-normal relationship is not new. Oke has been studying the topic for decades, and has proposed such a relationship. But it does have a side effect. Consider what happens when the trend line is shown on a normal plot:

Figure 2. The comparison of average California station temperature plotted relative to adjacent population, with a normal scale on both axes.

There is a “kick-over” in the rate of temperature rise at around a population of 10,000. (In fact this is a curve and the sharp transition is an artifact of the software, but it illustrates the trend). Temperature gains for smaller gains in population are higher below that level, while those above that population require a larger population growth to get the same increase. (Failure to recognize this is one of the underlying faults of the Berkeley Earth Project work on the topic.) Since the GISS data on temperatures also does not recognize any difference in population size below 10,000 it is also a fault of that data set.

I am curious to see how the California study pans out, I did drop a note with this finding to William Dean, as the e-mail suggested, and he was courteous enough to reply noting that this was “an interesting approach.”

As I pointed out to him, the strength of that relationship is, perhaps, borne out not only by the R^2 value, but by the consistency of the coefficient over the plots for a number of states. The tabulation is as follows:



I have had to cut the list in two to allow screen capture.


Figure 3. Correlation Coefficients for the relationship of temperature to local conditions with temperatures in degrees C.



And similarly for the table where I have converted the temperatures to def F.


Figure 4. Correlation Coefficients for the relationship of temperature to local conditions with temperatures in degrees F.

New Jersey combined temperatures

Crossing from Delaware into New Jersey as I come toward the end of the data acquisition part of the project I started last year looking at state temperatures over the past one hundred and fifteen years, I find that New Jersey has a dozen USHCN stations.

Location of the USHCN stations in New Jersey (CDIAC ).

According to the list, the only GISS station in the state is in Atlantic City. There have been 3 stations there, one that ran from 1895 to 2008, down at the Marina. This clearly shows the drop in temperature in the 1948 – 1965 period that I have been mentioning in the last few posts on the subject.

Longer term temperature profile reported for the GISS station in Atlantic City (GISS ).

However, as has become evident in many states that I have reviewed, the one that is being used by GISS has a much more recent history, only having been in operation since 1951.

That record also clearly shows the temperature drop, though with the start in 1951, it is not as clear that this is an anomaly from the overall rising trend.

Reported temperatures for the GISS station currently being used in Atlantic City (GISS ).

Given the steady rise in temperature of the station at the Marina, I was curious to see how far from the sea the new station is. It turns out to be at the airport, which is 9 miles from the sea, and 23 m above sea level.

Location for the current GISS station in Atlantic City, New Jersey.(Google Earth)

And then as I start to import the data for the USHCN stations, I find that the first one is still at the Atlantic City Marina:

Location of the USHCN station in Atlantic City, at the Marina (Google Earth)

New Jersey is 150 miles long and 70 miles wide, running from 73.9 deg W to 75.58 deg W, and 38.9 deg N to 41.3 deg W. The mean latitude is 40.1 deg , that if the USHCN stations is 40.3 deg N, and the GISS station is at 39.45 deg N. The elevation of the state runs from sea level to 549 m, with a mean elevation of 76.2 m. The mean USHCN station is at 53.9 m, while the GISS station is at 23 m.

Because of the short interval for which information from the current GISS station has been presented, the difference between it and the USHCN average is relatively short.

Difference between the data presented for the GISS station in New Jersey and the average of the USHCN stations

For the state itself, turning to the Time of Observation corrected (TOBS) raw data, and seeing how the temperature in the state has changed over the years:

Change in the TOBS temperatures, on average, for the USHCN stations in New Jersey.

It can be seen that there has been, with the exception of the time from about 1950 to 1965, a steady increase in temperature. As I had noted in an earlier post on Rhode Island the sea surface temperatures (SST) have risen by about 1.8 deg F per century. This is relatively close to the value shown in the above graph. (Note that the homogenized data plot shows a temperature rise of 2.45 deg F per century.)

Turning to the geographical factors, starting with latitude:

Effect of station latitude on temperature in New Jersey

Remember from previous observation that longitude is really a proxy in many cases for changes in elevation, and New Jersey is, in the main, relatively flat:

Effect of station longitude on temperature in New Jersey

There is really no significant effect of longitude, whereas when one looks at elevation:

Effect of station elevation on temperature in New Jersey

It is clear that the broadly consistent finding from other states on the role of elevation is valid also here, even with relatively smaller elevation changes.

When looking for populations, Charlotteburg has only one farm by it at the moment, but there are two sets of sub-divisions being developed in the neighborhood, which may have a significant impact on recorded temperatures in the future, though the reservoir may have a stabilizing effect.

Location of the USHCN station at Charlotteburg, NJ (Google Earth)

Indian Mills also did not come up with a citi-data site, so a check with Google Earth showed that it was close to Medford Lakes and that the station was surrounded by houses (with large lots). So I used the Medford Lakes population. Moorestown is on the edge of Philadelphia, but has a separate population,

Looking therefore at the effect of population, considering the average of the last 5 years temperatures against the local population:

Effect of local population on TOBS temperature for the USHCN stations in New Jersey.

Interestingly the homogenization of the data for the USHCN reported temperatures also creates a higher R^2 for this state.

Effect of local population on homogenized temperature for the USHCN stations in New Jersey.

Which suggests there might be some difference between the two sets of data, as there would appear to be. The recent drop is a little less than common to many earlier states.

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.

Maryland combined temperatures

Moving up the East Coast from Virginia, the next state is Maryland, where I will look to answer the continuing question on whether it too is showing the significant drop in temperature that seems to hold along the Atlantic Coast from around 1948 to the late 1960’s. It was not evident in the first temperature plot that I made for Missouri when I began this series, and only drew my attention as I reached states along the Atlantic Coast.

USHCN stations in Maryland (CDIAC)

Maryland has 16 USHCN stations from Beltsville to another Woodstock. It is interesting that there are two stations that are far to the west in the state, while all the rest lie relatively close to the coast. So I’ll come back to that point a little later in the post. Oh, and there is still that problem at the CDIAC site:



Mayland has one GISS station on the list, in Baltimore. And there are two Baltimore entries on the GISS site, one runs from 1895 to 2006 (Baltimore WSO City) and one from 1950 to date (Baltimore Blt/Washington International). It is, of course, the latter that has the correct co-ordinates for the current single GISS location that they use. I say of course since this (truncated station data) has been a common discovery as I have moved through the individual states. I had a look at the longer time record station and it does not show that temperature drop that I mentioned at the beginning of the post.

Historic temperature data for Baltimore MD, as recorded at the GISS station no longer used by them.

In contrast the site that has been retained does clearly show the drop in average annual temperature, from 13.67 deg C in 1953 down to 11.85 deg C in 1963, (a drop of 1.82 deg C, 3.3 deg F) consistent in size with the temperature drops seen further south.


Annual temperatures as reported by GISS for the station in Baltimore MD.

Interestingly one of the USHCN stations is located at the Maryland Science Center in Baltimore, and when one looks at the homogenized data for that site, the temperature drop is much less easily discerned:

USHCN homogenized temperatures for the MD Science Center in Baltimore (CDIAC)

And when the Time of Observation corrected raw data is examined there is no discernable drop, suggesting perhaps that the phenomenon is closely related to changes in the Sea Surface Temperature (SST) over that time interval, since the Science Center is surrounded by Baltimore.

Time of Observation corrected (TOBS) temperatures for the MD Science Center in Baltimore over the period where other states show a fall in temperature.

If I turn to the state as a whole, there is a difference between the sole GISS station temperature and that of the average of the USHCN stations, even with the homogenized data and the shorter time interval.

Difference between the GISS temperature for Maryland and the average of the homogenized USHCN station temperatures, in the interval where the GISS station has been functioning.

For the state as a whole, the data shows the three dgree drop in temperatures, both for the homogenized data, and for the TOBS values. The difference in temperature rise for both is relatively similar 1.59 deg F per century for the TOBS temperatures and 1.69 deg F for the homogenized. Given the strong influence of the adjacent ocean this is not perhaps surprising.

Average temperature for the state of Maryland over the years, from the TOBS data averaged for the stations in the state.

The two stations over on the far west of the state (Cumberland and Oakland) are also at a much greater elevation, however when I averaged the two over the same interval I get that same drop in temperature that I have been discussing.

Fall in temperature in the period from 1940’s through the 1960’s for the two western stations in Maryland, averaged TOBS data.

Maryland is 250 miles long and 90 miles wide, lying between 75.07 deg W and 79.55 deg W, and 37.88 deg N and 39.72 deg N. The mean latitude is sensibly 39.5 deg N. The USHCN average latitude is 39 deg N, and for the GISS station 39.28 deg N. The elevation rises from sea-level to 1,024 m, with the mean being 107 m. The USHCN average is 97.5 m, while the GISS station is only 6.1 m above the sea.

Looking at the way in which geography changes the temperatures around the state.

Variation in station temperature with latitude in Maryland (TOBS data)

Variation in station temperature with longitude in Maryland (TOBS data)

Variation in station temperature with elevation in Maryland (TOBS data)

Seeking information on populations around the various stations, Royal Oak didn’t appear in the citi-data list, and so I used the zip-codes site to get 483. Woodstock is a suburb of Baltimore, and is folded into Baltimore I suspect, so I again used the zip-codes site to get a population of 7,192.

Variation in station temperature with adjacent population in Maryland (TOBS data)

It dawns on me that the difference between the homogenized and TOBS data is generally quite small the closer we get to the present, and this should mean that ther should be a correlation with homogenized data, now that I am only using the past five years of temperature data to correlate with relatively recent population. (I wrote this before plotting the final graph).

Variation in station temperature with adjacent population in Maryland (using USHCN homogenized data)

And finally there is



Maybe I was being a little optimistic.

Virginia combined temperatures

This is a continuation of the series where I look at the temperatures recorded for an individual state, over the past 115 years, and see what that data tells us. Moving on from West Virginia which I covered last, to Virginia returns me to the Atlantic coast, and a certain curiosity as to whether the significant drop in temperatures between 1948 and 1965, found in other states further South along the coast, also occurred here. (For the impatient, it does). Virginia has 19 USHCN stations from Blacksburg to Woodstock, and has two GISS stations on the list, at Richmond and Roanoke.

Location of the stations in Virginia (CDIAC )

Unfortunately the CDIAC site still has the problems that I noted last week, when writing about West Virginia and it doesn’t appear possible at the moment to get the data directly from the site, which is an irritant. There are five Richmond sites as supplying data to GISS, though only one, Richmond Byrd, with the correct location, and it has data from 1911. Richmond is relatively close to the coast, and shows the temperature drop I referred to above, reaching a high in 1949, then dropping to a low in 1966.

UPDATE: Browsing Climate Audit, I came to Martin A's suggestion that we should consider the first difference trend for temperatures. I hadn't thought to do that, but since the data is easy to hand, I ran the plot and have added it to the end of the post. As Hu McCulloch noted, it doesn't seem to add much to the information on the state temperatures, having just about averaged out over the century. So, with respect, I don't think I'll add it to the repertoire.

Annual temperatures as reported for the GISS station at Richmond, VA.

Roanoke is one of the westernmost stations in the set, and it only has data from 1948, so that although there is a fall in temperatures from the beginning, which bottoms out in 1982, information on the temperatures in the 30’s is missing.

Annual temperatures as reported for the GISS station at Roanoke, VA.

When I combine the temperatures for the state from the USHCN network, and compare this with the average for the two GISS stations, then I get a graph that shows the change in range of the two stations, but that, recognizing that impact, shows that the GISS stations have always shown a higher temperature (by about 2.8 deg F).

Difference between GISS station average temperature and that of the USHCN average temperature per year.

In terms of the overall change in temperatures of the state over the century of data acquisition:

Change in average station temperature with time, for Virginia

There is still that drop in temperature from around 1950 to about 1968, with a consequent pick-up in temperature. Obviously we are no longer in that group of states that lost temperature over the century.

Virginia is 430 miles long and 200 miles wide. It runs from 75.22 deg W to 83.62 deg W, and from 36.52 deg N to 39.62 deg N. The central latitude is 37.49 deg N, that of the USHCN average is 37.7 deg N, and that of the GISS stations is 37.41 deg N.

The elevation in the state runs from sea-level to 1,742 m, with a mean elevation of 290 m. The average elevation of the USHCN stations is 281 m, and for the GISS stations 172 m.

It was a little more difficult to get the information on population for Virginia, since the site names did not easily fit with the source sites that I use. Bremo Bluff did not have a citi-data site, so I used the Zip-code site to find that it was 795. Burkes Garden was a little more of a challenge, with a population of 260 coming after a greater search, though it is 7 miles from Tazewell which has a population of 4,282. Dale Enterprise turns out to be on the outskirts of Harrisonburg (I had to use Google Earth to find that one) Hot Springs is (via Google Earth) actually now in Clifton Forge. Lincoln is near Purcellville, using the same approach; Piemont is in Orange, VA (via Google Earth) And so, ultimately it was possible to get some information on population sizes for all the stations around the state.

Looking therefore at the effects of geography and people on station data for Virginia.

Average station temperature for Virginia as it compares with station latitude.

The correlation is not as good as it normally is, and that may be because of the large variations in elevation within the state. That elevation also influences the apparent correlation with longitude, but remember that this regression line went up on the other side of the mountains in West Virginia.

Average station temperature for Virginia as it compares with station longitude.

The regression coefficient is also much greater with elevation.

Average station temperature for Virginia as it compares with station elevation

Correlating temperature over the past five years with recent population, gives:

Average station temperature for Virginia as it compares with population near the station.

If one were to take out the temperatures pre 1915, the homogenized and TOBS data would have been relatively equivalent until just after 1980 when there is an increase in the homogenized average.



UPDATE: Here is the finite difference plot, i.e. I have plotted the temperature change each year, by subtracting fromt hat annual average the temperature of the previous year, as a function of time. It is for the average of the USHCN TOBS data for the year, and it appears to suggest that, overall, the changes average out - which would suggest there hasn't been that much change in overall temperature, though the plots at the start of the post would suggest a rise of some 1.6 deg F per century.

Change in average Virginia temperature from the previous year, as a function of time.