Thursday, July 26, 2007

Does GISSTEMP overcount rural stations?

In a previous post, Eli quoted from Hansen, J.E., R. Ruedy, Mki. Sato, M. Imhoff, W. Lawrence, D. Easterling, T. Peterson, and T. Karl, 2001: A closer look at United States and global surface temperature change. J. Geophys. Res., 106, 23947-23963, doi:10.1029/2001JD000354 as to how the GISSTEMP team adjusts urban data

The urban adjustment in the current GISS analysis is a similar two-legged adjustment, but the date of the hinge point is no longer fixed at 1950, the maximum distance used for rural neighbors is 500 km provided that sufficient stations are available, and “small-town” (population 10,000 to 50,000) stations are also adjusted. The hinge date is now also chosen to minimize the difference between the adjusted urban record and the mean of its neighbors. In the United States (and nearby Canada and Mexico regions) the rural stations are now those that are “unlit” in satellite data, but in the rest of the world, rural stations are still defined to be places with a population less than 10,000. The added flexibility in the hinge point allows more realistic local adjustments, as the initiation of significant urban growth occurred at different times in different parts of the world.
In the US, they used satellite observations of night lights to define what are rural, suburban and urban areas:
The percent of brightness refers to the fraction of the area-time at which light was detected, i.e., the percent of cloud-screened observations that triggered the sensor. These data are then summarized into three categories (0-8, 8-88, and 88-100%). From empirical studies in several regions of the United States, Imhoff et al. associate the brightest regions (which we designate as “bright” or “urban”) with population densities of about 10 persons/ha or greater and the darkest (“unlit” or “rural”) regions with population densities of about 0.1 persons/ha or less. As is apparent from Plate 1b, the intermediate brightness category (“dim” or “periurban”) may be a small town or the fringe of an urban area.
after this classification the number of rural stations in the US is reduced to 214 USHCN stations and 256 of the GHCN stations (obviously a lot of duplication here)
As the contiguous United States covers only about 2% of the Earth’s area, the 250 stations are sufficient for an accurate estimate of national long-term temperature change, but the process inherently introduces a smoothing of the geographical pattern of temperature change.
Outside of the US, they continue to use population data to define rural stations.

The bottom line is that the ONLY stations which contribute to the overall trend are the RURAL stations. Moreover, rural stations near heavily settled areas will be more strongly overcounted because the trends in the few rural stations in such an area will dominate all of the stations in the area and the nearby points on the grid to which the temperature data is fit.


Horatio Algeranon said...

Fractional Impervious Surface Area. The U.S.-constructed impervious surface area (ISA) in 2000 was nearly the size of Ohio."
(Roughly 1% of the total US land area)

For the past few decades the pace of building (roads, buildings, etc) has been increasing.

When you pave over a grassy field to put in a parking lot, it has an obvious impact on the local albedo and local heating -- and possibly on overall heating, as well.

It stands to reason that at least part of the urban heat island effect is actually "real" -- ie, that it is not just a matter of inflated temperatures that are not representative of the US as a whole.

I'm not sure what fraction but at least part of the increase may reflect actual heating of the US overall.

In other words, the mean temperature in the US may be higher than it was a century ago simply because of all the roads and buildings.

I'm not sure if this is part of the consideration when adjusting the temperature downward to "fix" the increased temperature measured in urban areas. If not, it probably should be. Otherwise, there may be an "overcompensation" occurring.

Horatio Algeranon said...

Correction to above:

..."simply because of all the roads and buildings."

should read "partly because of all the roads and buildings."

Anonymous said...

(caerbannog the anonybunny)

I originally posted this material in the "Wanna see some pictures lil' girl?" thread below, but since that discussion's getting a bit stale (and threatening to scroll off the main page), I figure that I should re-post it here.

(first post)
I was poking around a bit on the USHCN web-site. The folks' two favorite stations are Orland, CA and Marysville, CA. The "default" temperature plots for both sites are prominently featured at

But I decided to "drill down" a bit. I set the plot time-scale to 1970-2005 and looked at things like minimum temp vs time. Over this 35-year period, the trends for the Orland and Marysville stations are pretty similar. You have to go back prior to 1960/1970 to see much of a discrepancy between the two stations' trends, i.e. back to a time that pre-dates most urban expansion!

The post-urban expansion data for the past 30-40 years looks much more consistent than the earlier data! So whatever it is that was driving the differences between the two stations' data, it doesn't appear to have been urbanization!

Some "outlier" temperature readings back in the late 1800's impact the autoscaling of the Orland station plot for the default plot settings. Set the time-scale to look at the last 35 years or so, and the differences are much less dramatic, *especially* if you look at plots of the *minimum* temperatures.

You can hunt up the USHCN data at

A look at data from neighboring stations for this time-period will show pretty consistent trends, with overnight lows appearing to increase the most consistently (according to my crude eyeball estimates).

The differing urban vs. rural settings for the Orland and Marysville sites don't have nearly the impact that the "default" plot-settings imply.

(next post)
I just took a closer look at folks' renditions of the Orland vs. Marysville plots.

They show the Orland data going back to the 1800's, but they truncate the Marysville data to 1900-1910 or so (poor plot resolution makes it hard to tell exactly.) Data for both stations go back to the late 1800's, and both stations show anomalously high temp readings prior to 1900. The folks saw fit to include the high 1800's temp readings for the Orland plot, but *exclude* the high 1800's temp readings for their Marysville plot.

The Orland plot was autoscaled to keep the high 1800's temps on the plot, but the same was not done for the Marysville plot! So that's the way of doing business -- selectively truncate data and let autoscaling do its magic!

It's the same stunt the deniers pulled with their "hockey-stick" antics. Monkey with the plot scales to obscure the order-of-magnitude difference between the dynamic ranges of their "noise-only" hockey-sticks and Mann's data-driven hockey-stick.

Anonymous said...

(caerbannog the anonybunny again)

The Orland/Marysville data looks more interesting, the more I look at it. For the Orland station, the only data that go back prior to 1900 are "mean temperature" data. The min and max temperature data go back only to about 1905 or so, and don't show the high late 1800's "spike".

Ditto for the Marysville data. The only data that go back well before 1900 are the "mean temperature" data. The min/max data don't go back far enough to capture the late 1800's "spike".

The surfacestation folks clearly chose the "mean temp" data for their Orland plot -- nothing else goes back prior to 1900. I can't tell exactly which plot options they chose to generate the Marysville station temperature plot, but they sure did not include (nor scale the plot for) the high pre-1900 mean-temperature readings!

Anonymous said...

It's pretty clear what their underlying theme is: the data from surface stations are all bunk.

In fact, they pick and choose which stations to display the data for.

Would not want to show any data from a good station that shows a warming trend, now would they?

Anonymous said...

(caerbannog the anonybunny)

Would not want to show any data from a good station that shows a warming trend, now would they?

Heh, heh... you mean data from, say, the Orland station that they praise as "well maintained and well sited"?

Plot the Orland temp data for the last 35-45 years and you see a clear warming trend.

Anonymous said...

You're right, caerbannog

I should have said a "warming trend over the entire 20th century".

but actually, when it comes right down to it, if you look at the greenhouse forcing over the 20th century it really started ramping up in the 70's and shortly thereafter, the temperature also started to really ramp up pretty steeply.

That's not to say greenhouse forcing was not increasing before then, but just more slowly (and it actually leveled off from about ten years from 1940-1955 or so), which means that other factors (eg, cold fronts from Canada) might have counteracted the greenhouse warming.

EliRabett said...

There is something funny about Northern California data at the turn of the century. Although they did not comment on Maryville and Orland, the 2001 GISSTEMP paper did say

"The strong cooling that exists in the unlit station data in the northern California region is not found in either the periurban or urban stations either with or without any of the adjustments. Ocean temperature data for the same period, illustrated below, has strong warming along the entire West Coast of the United States. This suggests the possibility of a flaw in the unlit station data for that small region. After examination of all of the stations in this region, five of the USHCN station records were altered in the GISS analysis because of inhomogeneities with neighboring stations (data prior to 1927 for Lake Spaulding, data prior to 1929 for Orleans, data prior to 1911 for Electra Ph, data prior of 1906 for Willows 6W, and all data for Crater Lake NPS HQ were omitted), so these apparent data flaws would not be transmitted to adjusted periurban and urban stations. If these adjustments were not made, the 100-year temperature change in the United States would be reduced by 0.01°C."

Anonymous said...

Interesting. Photos from a satellite many thousands of miles up are valid scientifically to determine if a site is rural or urban. Photos taken 20 feet away from a site showing microsite contamination and non-adherence to WMO standards have no metadata value whatsover. As I said, interesting.

Anonymous said...

Don't forget irrigation cooling effect (ICE) of farming or just watering of lawns and golf courses near cities. The evaporation or transpiration of one millimeter deep of water at the surface requires about the same energy as heating a 2 kilometer high column of dry air above the area watered, one degree C.

Anonymous said...

There are lots of things that can affect the local temperature record -- eg, agriculture, both through vegetation changes and evaporative cooling from irrigation (not just blacktop near the station!) and there is a very good reason why no one station is used as an indication of a trend.

Anonymous said...

I guess we'll see once everything's gathered and categorized what's done with it. As long as all the data is made available, whatever they come up with can be checked.

Anonymous said...

"So that's the way of doing business -- selectively truncate data and let autoscaling do its magic!"

Pox on YOU! The charts are DIRECTLY FROM NASA, Watts did NOTHING to alter them at all! You people are just so full of yourselves; AGW is a proven HOAX!