Metadata Dubuque and UHI
I’m in the process of remaking all the metadata from scratch and looking once again at the question of UHI. There are not any global conclusions we can draw from the data yet; I’m just in the process of checking out everything that is available that could be used to illuminate the problem. The problem, as I see it, is as follows. We know from numerous studies that UHI is a real phenomena. But, it’s far from simple. First lets get down to causes. The causes of UHI according to Oke are as follows. Note 20 + years of research has no expanded this list in any significant way
In the context of global temperature we are concerned with canopy layer UHI, that is UHI below the top of buildings where surface station thermometers are. Lets go through the causes one by one.
1. Increased absorption of short wave radiation. As Oke notes the construction of buildings adds surface area to the urban landscape. Building walls absorb SW radiation and the arrangement of building can lead to multiple reflection.
2. Increased LW radiation, primarily from air pollution over the city
3. decreased LW radiation from the surface. Skyview factor is critical here and skyview is directly related to the geometry of buildings. In a flat open plane free of building the surface has a clear view of the sky and can radiate accordingly.
4. Anthropogenic heat: this is excess heat from buildings and traffic. This is should be noted changes dramatically with the type of city and with latitude. The ranges of excess watts is rather large between city types.
5. decrease evapotranspiration: This results from changing the surface of the earth.
6. decreased turbulant heat transport. Again, building geometry plays an important roll here as does the local wind.
In addition to these variable that one can control when developing and urban area there are several that cannot be controled: the wind, clouds and nearby water.
With that in mind, let’s start to look at some metadata for Dubuque and its nearby airport. First a google map view
In the following charts the city pin will be marked witha blue cross and the airport with a red cross. In terms of incoming solar we should be aware that changes to evapotranspiration can cause changes in cloud cover. Looking at Modis cloud cover, we see the count for days of cloud cover. Less clouds is more sun.
Next we want to look at transformations to the surface. In the map that follows we have marked the urban built areas as Green bits
Next we look at population. There are a few good sources here: first is a 5 minute source for 2005 and after a 2.5 minute source for the year 2000
Next, we will look at daytime LST or land surface temperature.
And next we look at Night time LSt when UHI is supposed to be the highest
In case you are wondering what that long green patch of warmth is, it’s water.
A couple things are clear. A reader named sunshine suggested that perhaps cars from the city infected the airport with UHI. Looking at the data, the differences in temperature would seem more likely to derive from fewer clouds over the city, more built surfaces, and the higher population that created those surfaces. The surfaces hold heat and they appear to retard the formation of clouds, leading to more sunshine in the city than at the airport. And, it seems the nearby water is morel likely to modulate the temperature at the airport than anything else.
Funny. First “sunshine” thought it was “depopulation”. Then he surmised it was cars. In all of this there was no attempt to to verify a thesis, just random thoughts. Sunshine, appears to have his head where there is none. In any case trying to find a cause for cooling trends or a cause for warming trends, trying to find a UHI signal, isn’t a simple task of throwing up random thoughts. We know the causes. 20 years after Oke’s essay, the causes remain the same. The weight of the causes is critical. more on that later
Steven Mosher's Blog 
I thought my thesis was well explained.
“warming counties grew much, much faster than the country as a whole, while cooling counties grew slower than the country as a whole.”
http://sunshinehours.wordpress.com/2012/03/17/county-population-statistics-and-coolingwarming-stations-since-1900/
I think the fact that I consider different explanations for so many cooling stations is a credit to my curiosity.
You seem stuck on one explanation.
Ha sunshine
you forget about ur car thesis. forgot to check the stats. and you are stuck on ur thesis. stations cam cool for many reasons. first do the stats right. then examime each series for station moves. then you cam look at land use changes. then local population. local. not county.
Another interesting evaluation in that it points to water as a key component in UHI. When considering rural versus urban temperatures this should be taken into account, especially as changes in agriculture have resulted in some areas in crops of low transpiration rate being substituted for ones with higher values and related to yield. There have also been changes in water use in cities for many reasons but especially in parks and gardens.
Interestingly, at least to me, is that whilst AGW “theory” posits a positive feedback from increased transfer of water from surface to atmosphere the Class A pan evaporation units that are used widely to measure crop water requirements for irrigation do not support this claim, rather the claim that climate sensitivity is overstated. See:
Click to access nc-ess-pan-evap.pdf
Also of interest is the claim by Salby that annual fluctuations in the atmospheric levels of carbon dioxide measured at Mona Loa can be linked to changes in surface water as measured by satellites and to temperature anomalies that reflect ENSO, PDO etc.See: YouTube for video.
water and UHI. Yes, the factor that matters most is the moisture differential between the urban and the rural, Oke and Grimmond.
“you forget about ur car thesis”
Are you suggesting a county that grew by over 600% (the average warming county) from 1900 to 2010 does not have more cars in the county? And more blacktop etc etc?
“forgot to check the stats”
The stats that show warming counties population grew faster than cooling counties?
It is interesting to zero in on at the county level and see stations turn from warming to cooling in large groups.
Sorry I’ve tuned in late. Is there a posting which describes and summarizes the purpose and rationale for this effort? Can you point me to a link?
Thanks,
– Jan
Bruce, you seem to have lost the bubble. Your car thesis was that the cars from Dubuque “somehow” had an effect on the airport temperatures. You forgot several things
1. The airport is cooler than the city
2. The airport thermometer is located far away from any car traffic
3. hot air rises
4. The airport is cooler because it doesnt have the all those features which cause UHI
a) it has a skyview of 1
b) building height is low
c) population is 15 people per sq km versus the city which is 1200 people per sq km
d) Built area. the airport has a low percentange ( <5% ) of built area whereas the city
has 90% +
Your chart of cooling stations NEGLECTS to do the proper statistical test for trends.
Further you neglected final QC and you forgot to check for station moves.
” The airport is cooler than the city”
I think a valid thesis would be that if the population of a city drops (as it did in Dubuque) and the region de-industrializes somewhat (as it appears to have done), then UHI could drop meaning that the airport could get colder as the UHI bubble around Dubuque shrinks.
“The airport thermometer is located far away from any car traffic”
People walk to the airport? No roads at all?
” hot air rises”
Yes.
Aside from all that, have you noticed the weather station 2 miles north of Dubuque is also cooling? It seemed to have started cooling when Dubuque’s population started shrinking.
Look at any satillite views of the UHI or any empirical studies of UHI. I gave you studies that show the extent. The effect falls off with distance from the city. In some cases very rapidly.
Look at the actual plots of temperature at the airport relative to its close surroundings.
Look at any data on road temperature and how quickly it falls off to zero effect. You can find that data as well ( snow removal planning, for example ).
The bottom line is this. UHI exists. If you look for it in really big cities on perfect days for UHI you will find it. The question is not MAX UHI. the question is average UHI over all conditions and locations. The highest that has been measured is .12C per decade from the 70s on. That’s the worst case of the worst enviroment. Biggest most dense cities. 10000 people per sq km or more. huge building, millions of cars.
By your logic, Dubuque at 1200 people per sq km would be less. fewer people, shorter buildings, fewer cars.
By your logic the airport at 15 people per sq km would be even less. “could” there be some? sure, but far below .12C per decade. Too low to even measure because the uncertainty in trends is too large.
Before you go looking for an effect you have to understand the size of effect you are looking for AND the underlying noise in the measure.
Do a power test.
“Over the course of 12 years, between 1987 and 1999, the
mean nighttime surface temperature heat island of Houston
increased 0.82F0.10 K in magnitude. It increased in area
170F30 km2 using the Gaussian method of area determination,
and 650F60 km2 using the 1 K threshold method. It
is curious to note that the growth of UHI, both in magnitude
and spatial extent (using the Gaussian method of determination),
scales roughly with the increase in population
(extrapolated to 1987 levels), at approximately 30%.”
Click to access UHI-houston.pdf
“Further you neglected final QC and you forgot to check for station moves.”
BEST says about the data I am using:
“Same as “Single-valued” except that all values flagged as bad via the quality control processes have been removed. This dataset is recommended for users that require relatively clean data and want seasonality to be preserved.”
http://berkeleyearth.org/data/
Are they lying?
I know Houston is no Dubuque, but an increase in UHI of 170 sq. km to 650 sq km that scaled with population gives us some idea of what UHI is REALLY capable.
.82F in 12 years is not negligible.
The inference is that if Houston’s population had dropped, the heat bubble would have shrunk is also worth considering.
Bruce. Please read the paper. Whenever you see a satellite measure of UHI understand that they are not measuring the same UHI that we are talking about
They are using LST or the land surface temperature. This is different than the AIR SURFACE TEMPERATURE. In the literature people now refer
to the increased heat in the actual surface as SUHI. Next, of necessity, these measure are WORST CASE. Why? because they have to be taken
on cloudless days, days when SUHI and UHI are worst. Next, SUHI >> UHI. simply the heating you see at the surface is always greater than
the airport temperature 2 meters above it. For global temperature we are concerned with SAT ( surface air temperature ) not LST
If you spent more time trying to understand the problem you’d be better off
“The major urban core of Manchester city centre is clearly discernable with a heat island extending nine to ten kilometres from the centre.”
Click to access urban_heat_island_-_manchester.pdf
“That the temperatures were warmer at an urban
weather station than a rural one comes as no surprise.
Yet, the results shown here corroborate the findings
of Lupo et al. (2003b) that even a city of modest
size (Columbia, Missouri, population was ∼84 000 in
2000) can have a measurable heat island. Indeed, the
temperature differences between the urban core and
KCOU discovered by Lupo et al. (2003b) of 1.1 to
1.7 ◦C compare well with those shown here of 1.6
to 1.8 ◦C.
https://mospace.umsystem.edu/xmlui/bitstream/handle/10355/2471/FurtherStudiesHeatIsland.pdf?sequence=1
Please Bruce, spare me the school children studies. please read. 9 to 10 kilometers from the CENTER. What you want to do is look at the actual size of the urban center
( typically ISA will be 75-100% here ) then comes the outer core ( 25-75%). And yes, I’ve read that one before. I’ve read all the studies that try to quantity the
fall off with distance. WHy? very simple. To define a rural site I want to make sure that I am outside the periphery of the city. 5-10km away frm the EDGE of the city
not the CENTER of the city. why? cause 10 km from the center of the city can STILL be urban. doh.
UHI falls off with distance. you just cited another paper that shows that. Have a look at the map. have a look at the actual GIS data. then get back to me.
Here is what you need to do. Find the FUNCTION that tell you how fast the UHI falls off. in some cases it drops to zero quickly. like a kilometer or 2.
If cities of 84,000 can have 1.7C UHI, that mocks your claim of .017 or whatever miniscule amount you say.
The other key point in the Houston study is that UHI extent and magnitude climbs with population. That means UHI should drop if the population drops as happened to Dubuque. As my analysis shows, many counties have shrunk in population in the same region that so many stations show cooling.
@sunshinehours1,
“As my analysis shows, many counties have shrunk in population in the same region that so many stations show cooling.” So? That’s not a cause.
Uh, still haven’t gotten a comment about what this is all about, but I know that some temperature sensors are located in places like National Parks, which tend to be cooler than ambient.
Also, it seems to me, not really knowing much about the project here, that a supposition, as I understand it, that some sensor positions are biasing the results could be answered definitively by 0.632+ bootstrap or leave-one-out or leave-group-out cross validation. It seems the stratification technique pursued here is laudable, but is doing things the hard way.
Still, I’m probably not familiar enough with the goals here to really be qualified to comment.
” So? That’s not a cause.”
Shrinking population = shrinking UHI.
If thats not a cause, what is a cause?
Simply because population decreases does not necessarily mean infrastructure goes away. Infrastructure (schools, apartments, shops, etc) are built in response to population increase, but these don’t fade away when population decreases. Buildings still need residual heat lest pipes freeze. Even if infrastructure does go away, there should be a lag between population decrease and infrastructure dismantling.
Also, your statement “shrinking population = shrinking UHI” needs QUANTITATIVE backing. How much does it go down? What’s the correlation? Is it statistically significant? (R built-ins “acf”, “ccf”, and “pacf” have their own definition of statistically significant. Have you checked these against that?) Does this happen just for one city or urban area or many? If for many, how does it vary? If many have you done bootstrapping to see the variability? or leave-one-out cross-validation? What’s the lag between UHI and shrinking population? Is there in fact a deterministic change? What time period was this observed for? How many distinct observations? What’s the scatter?
“Simply because population decreases does not necessarily mean infrastructure goes away.”
Tell that to Detroit.
http://www.theseekerbooks.com/detroit/neighborhoods.html
http://zfein.com/photography/detroit/index.html
Anecdotal evidence is not quantitative evidence.
“Between 1950 and 2000, the US population grew by 86%, but the population of Kansas shrunk by 21%.
This is driven by young people rejecting farming for city life, the impact of technology on the yield of the land and the financial challenges of farming in the 21st century.
Across the Midwest, 89% of cities have fewer than 3,000 people, and in Kansas alone, there are more than 6,000 ghost towns.
These communities are often made up of only a handful of houses, miles apart, and many settlements have been abandoned completely.”
If you take the individual station data for the United States and plot average temperature against elevation, latitude and population by individual states (see the series I wrote at Bit Tooth Energy where South Carolina temperatures sort of gives the flavor of the results I found) you will find that the Time of Observation corrected data gives better correlation than does the data after it has been massaged. I found it relatively easy to find local population sizes around the stations for each location from the web (if you are interested, the data is available by state on Excel spreadsheets).
In general the correlation of average temperature with population size around the individual stations correlates with the log of population (yes I know that this is not a new finding, merely a corroboration of an old one) though, with changes in population over time, there is a question as to how many years the temperature should be averaged to correlate with current populations. But it is a start – hope it helps.
How much correlation? And, regarding “…the official temperature trend for the state over the past 115 years shows an increase in total of about 0.44 deg F per century, although in similar mode to Georgia there is that odd dip that ends in 1965…” at the cited post, why is that fit with a line? The series clearly has periodicities. That means R-squared is meaningless, since it assumes the data are i.i.d. and they are not. (Also, for the “Average temperature for the USHCN stations in South Carolina, using the time of observation corrected raw data” plot, it’s interesting no comment was made, or at least nothing was puzzled at, since R^2 is tiny.) To see, either plot the residuals on a lag plot, or do a QQ plot against a Gaussian. Are the data behind these plots available for download? (If not, why not?)
Commenting here rather than there, because the primary issue is here, and rather than recap the results there here, a reference was placed.
I hate to add to the random thoughts but when I lived in Melbourne, Australia I made the regular observation that the city was surrounded by a ring of cloudy skies while the city had clear skies. How do I change that random thought into something more of a provable observation? Mosher I like your MODIS cloud image how do I generate the same for Melbourne?
Thanks