This gives credence to Dr. Roy Spencer’s population adjusted ISH surface temperature data for the USA
population density.
Impact of population density on the surface temperature and micro-climate
of Delhi
Javed Mallick1 and Atiqur Rahman2,* 1Department of Civil Engineering, King Khalid University, Abha, Kingdom of Saudi Arabia 2Remote Sensing and GIS Division, Department of Geography, Faculty of Natural Sciences, Jamia Millia Islamia University, New Delhi 110 025, India
Abstract:
Increasing urban surface temperature due to change of natural surfaces is one of the growing environmental problems in urban areas, especially in cities like Delhi. The present work is an attempt to assess the urban surface temperature in Delhi using remote sensing and GIS techniques. ASTER datasets of thermal bands were used to assess the land surface temperature (LST) using temperature emissivity separation technique. Ward-wise population density was calculated from the Census of India 2001 data to correlate the population density with LST. The study shows that surface temperature changes with the increase in the impervious surface area, which is related to the increase in the population density.
From the discussion:
Figure 4 shows the correlation between surface temperature and population density. A strong positive correlation between the two can be seen. The value of logarithmic regression (R2) is 0.748. The logarithmic regression equation between surface temperature and population density is Y = 1.059ln × Pop-density + 22.40. It means that with the increase in population density, surface temperature also increases. Furthermore, it is possible to predict night-time surface temperature on the basis of known population density.
….
Oke used the empirical method to represent the relationship between urban–rural temperatures as concentration on population. It is not easy to separate many contributors to the problem. However, one of the most noticeable and one that has proven to have an
extremely strong correlation with the UHI phenomenon or urban surface temperature is the population density of major cities32. The present study demonstrates a close
relationship between the population density, built-up area and surface temperature. The statistical analysis of nighttime surface temperature with population density indicates
that population growth tends to contribute to the urban surface temperature rise or UHI intensity and also to the micro-climate of Delhi.
In earlier times and even now, there are very few meteorological stations to record the surface temperature, and they may not be the true representative for the whole city. In such situations, LST derived from thermal satellite data are useful to study the variation of surface temperature over the entire city area that is an important parameter for micro-climate of urban areas. This study clearly shows spatial variation of LST over entire Delhi.
By superposing the population density map of Delhi over the surface temperature map, it can be clearly seen that high population density is one of the main contributing
factors for the high surface temperature, UHI intensity and also micro-climate of Delhi. To assess and address the issue of micro-climate and also to mitigate the impact of UHI on the city population, the outcome of such types of studies may be useful.
h/t to Dr. Willie Soon
paper here: MallickRahman12-June-ImpactofPopulationonTsfc+MicroClimate-NewDelhi
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
hey greenies, you better start planting some trees in your cities now instead of whining about rising CO2 levels.
Duh!
Unfortunately the relationship between LST and SAT is not straightforward.
So this and other studies have shown a good relationship between percent of impervious surface (ISA) and LST ( land surface temperature ), but the relationship between LST and SAT ( surface air temperature ) is not that clear, primarily because of horizontal advection.
Simply, you can have land that is adjacent to high percentage ISA and High population density that suffers the effect of upwind high LST. Think of a thermometer over grass that is downwind from a patch of concrete that has High LST ( at night when it gives up its retained heat ).
The relationship between Impervious surface percentage and population density is also confounded. It is regionally biased by building practices. This is one reason why Oke found different curves for the US, Europe, Asia and South America.
One curve doesnt fit all. and In some locals the curve is logistic.
This is easily seen by selecting areas that have High ISA, High LST, High SAT and no population. How do you get High ISA? High LST? High SAT and no population?
simple. Industrial sites. Population density figures are collected for residential population and not ambient population. So, if you have an airport or a refinery you will see High ISA, High LST, High SAT and no population, because census figures capture where people live not where they work. The effect of ambient population can be seen in Tokyo where the weekend UHI is less than the weekday UHI.
All very interesting. The point being there is no simplistic single factor adjustment you can apply to data to correct for UHI.
The other problem is that LST is not collected at the same times as Tmax and Tmin. Its utterly dependent on the over pass time. And its dependent on having the right synoptic conditions.
basically you can only collect the data when the conditions are correlated with the effect being its worst. Cloudless, windless, rain free days.
Nice to have it somewhat quantified but not exactly a shocker.
Clearly we can halt AGW by eliminating Indians? (sarc- my best software is Indian, and I would not personally eliminate anyone)
I don’t know where the idea of the log(pop) model comes from. It’s often bandied about but I never saw a reason proposed for the log relationship. Just a guess?
I the case of this particular graph I would estimate that T-T0=k.pop^2 would provide a better fit.
In any case it’s good to see some serious, objective science is finally getting done and getting published. In the last 6 months or so I have noted some sanity returning to the field.
Okay, is this about Delhi or New Delhi? Those are not the same thing, though in close proximity. 9,810,000 population vs. 294,000, respectively.
“All very interesting. The point being there is no simplistic single factor adjustment you can apply to data to correct for UHI.”
Yet we can build a model to replicate the ever changing climate?
Nice. Very nice.
for folks who are interested in some of the problems associated with the differences between LST and SAT you can look at this
http://ftp.cavs.msstate.edu/publications/docs/2006/03/3986GIScience_pub_2006.pdf
P. Solar
“I don’t know where the idea of the log(pop) model comes from. It’s often bandied about but I never saw a reason proposed for the log relationship. Just a guess?”
The log relationship is probably the wrong functional form. The easiest way to see this is
to look at their underlying data. The lowest population density they look at is 300 people per sq km. If you plug in a population density of 2 people, you get a delta temperature of 12 C or so.
functionally you can expect the relationship to be logistic. tending to zero at one end and limited at the high end. the log fits are just done for ease of computation. Other studies use 1/4 power functions,
The first person to do a log(pop) was Oke, but he quickly found out that the function didnt work if you considered other parts of the world. His first study was focused on America and Europe.
Studies in Asia and South America came up with different curves. He added a component to the model that represented regional wind since wind can drive UHI to zero if it is above 7 meters per second or so. Its been recognized that part of the reason while population curves differ for different regions is that building practices are different. Its the building material and density of the surface changes that drive LST. Population is a consequence of that. Population does effect the waste heat portion of the energy balance equation. For the most part today researchers are more focused on explaining temperature in terms of the land surface: ISA, Vegetation and there are very few who actually look at population density. Basically to get a really dense population figure you need lots of concrete and tall buildings. So, ISA explains a great deal of the variance.
the true impact of the humans can only be gauged by looking at the same surface with and without people. So, the weekend UHI effect in tokyo is a good place to start to understand how much of UHI is due to the buildings and how much is due to the people in those buildings.
or how much is due to the road and how much is due to the car on that road dumping hot exhaust into the air. Most researchers dont look at population and the science tended to move away from population because the other factors — surface changes made by and for humans– explained more of the differences in LST.
Hi S. Mosher,
My memory of New Delhi during monsoon seasons is there were not consistently “cloudless” … or “rain free days”. I would like to understand how the site’s monsoon makes the study’s observations skewed.
Moshpup’s first post is quite reasonable. Unfortunately all the studies and temperature series he works on, and with, completely ignore what he just explained to us!!! They simply assume that it all averages out and their adjustments aren’t affected!! Of course, without detailed investigation of each station and its environment they are showing more religious belief than empiricism.
HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA
The correlation will be with electricity consumption.
It’s well established that surface temperatures in India are largely driven by aerosol levels (outside the monsoon season). Increased population results in increased burning of coal and biofuels (mostly for cooking), increased aerosol levels, and decreased (especially daytime) temperatures.
However, this relationship breaks down beyond a certain population density – it’s not feasible to have open fires in apartment blocks – and electricity and gas become the main cooking fuel, aerosols decrease and temperatures increase.
This is substantially the reverse of the situation in the developed world.
So, the weekend UHI effect in tokyo is a good place to start to understand how much of UHI is due to the buildings and how much is due to the people in those buildings.
or how much is due to the road and how much is due to the car on that road dumping hot exhaust into the air.
The cars are dumping aerosols into the air. And Tokyo is warmer at the weekend than during the week. Clearly ‘cars dumping heat’ isn’t a significant factor.
http://www.springerlink.com/content/u41h6p5367580731/
Why not just heavily instrument a small, carefully selected compact urban center and get real empirical data instead of this guessing? A few hundred weather stations and a data collection system can’t be all that expensive or difficult and you would have all the information you need. It would be a better use of money than dumping it into faster computers or studying mosquito habits.
That line shape is an enzyme rate or a binding constant
http://en.wikipedia.org/wiki/Michaelis_menten
gringojay says:
August 31, 2012 at 2:47 pm (Edit)
Hi S. Mosher,
My memory of New Delhi during monsoon seasons is there were not consistently “cloudless” … or “rain free days”. I would like to understand how the site’s monsoon makes the study’s observations skewed.
########################
Simple. To do this type of study you look through all the samples ( for example using REVERB) and you find a cloudless day. Every image is marked by percentage of cloudiness.
here is REVERB.
http://reverb.echo.nasa.gov/reverb/#utf8=%E2%9C%93&spatial_map=satellite&spatial_type=rectangle
here is http://ftp.. but you need to understand the various products and bands
ftp://e4ftl01.cr.usgs.gov/
To get a good estimate the LST you need cloud free days. They picked one day in october ( october 7 2001 ) october is pretty good to find cloudfree and rain free days.
So basically you are seeing the results from 1 day.
http://weatherspark.com/averages/33934/10/New-Delhi-India.
As for skewing the results. The problem is that the synoptic conditions you need for for clear visibility of the surface are also those conditions which tend to show the highest UHI. So usually folks refer to this as UHImax. That is the worst you can see. Still it’s important to look at LST.
I would imagine that if I showed folks the LST images for surface stations some of them might object to what they see. It’s pretty easy to pull down an image for any station and look at the LST around it. Like I said if I showed people one day of data that went against their notions, I am quite sure they would rightly complain about that.
Maybe I’ll write that up, there is a group working on making MODIS data products easier to load into R. Right now its a pain in the ass. Lots of hand work, but very interesting to have at hand.
I’d think the “log-like” shape results from horizontal vs vertical buildings. As people start to gather into villages, they spread out horizontally, so the number of buildings, streets and blocks would be linear with population.
Beyond a certain density you HAVE to go multi-story. No choice. As pop increases above that inflection point, the number of buildings, streets, and blocks increases much more slowly.
Steven Mosher says:
August 31, 2012 at 5:44 pm
“I would imagine that if I showed folks the LST images for surface stations some of them might object to what they see. ”
========
Lots of things I don’t like to see, what did you mean by “object”.
Phillip
“The cars are dumping aerosols into the air. And Tokyo is warmer at the weekend than during the week. ”
The study:
“On Saturdays and holidays (Sundays and national holidays), δT* was lower than on weekdays by 0.2–0.25°C at Tokyo,”
As I recall the anthropogenic heat flux in Tokyo was 400W and 1590 depending on the season.
Or Phillip you could go to the tokyo government site and download “Guidelines for Heat Island
Control Measures” In that city they have heat maps for every ward.
Or you could look at a similar study done in Korea and week the weekend effect
http://journals.ametsoc.org/doi/pdf/10.1175/JAM2226.1
This paper is good because it shows you the effects of windspeed and cloudiness, as I recall the weekend differences are covered toward the end of the paper, but read the whole thing, there is a huge pile of reading to do in this area. some things that might surprise you.
“kuhnkat says:
August 31, 2012 at 3:28 pm (Edit)
Moshpup’s first post is quite reasonable. Unfortunately all the studies and temperature series he works on, and with, completely ignore what he just explained to us!!! They simply assume that it all averages out and their adjustments aren’t affected!! Of course, without detailed investigation of each station and its environment they are showing more religious belief than empiricism.”
Sorry katman. All of the work I’ve personally done is with raw daily data. So, you must be mistaking me for somebody who works with CRU data, GHCNM data, or GISS data.
It’s important to look at all sources and get your hands on everything. That gives you a leg up on people who just read about data.
Dennis Kuzara says:
August 31, 2012 at 5:08 pm (Edit)
Why not just heavily instrument a small, carefully selected compact urban center and get real empirical data instead of this guessing? A few hundred weather stations and a data collection system can’t be all that expensive or difficult and you would have all the information you need. It would be a better use of money than dumping it into faster computers or studying mosquito habits.
##################
that’s been done. The problem is you have a single sample and then you cant generalize.
Here is what is involved in instrumenting a city
http://pages.unibas.ch/geo/mcr/Projects/BUBBLE/
http://pages.unibas.ch/geo/mcr/Projects/BUBBLE/textpages/ov_frameset.en.htm
http://pages.unibas.ch/geo/mcr/Projects/BUBBLE/textpages/ov_frameset.en.htm
Kinda cool. You could compare
the rural sites
http://pages.unibas.ch/geo/mcr/dolueg/dbdoku/GEMP.en.html
with the urban sites.. and then look at how good the modelling was.
Even had a wind tunnel model of the city and tracer experiments.
Here Kuhnkat.
Some simple instructions
http://www.r-gis.net/?q=ModisDownload
Then once you have the data you have to do some nasty programming to pull out the data
for every site in the US ( for example ) takes a long time.
Steven Mosher: “All very interesting. The point being there is no simplistic single factor adjustment you can apply to data to correct for UHI.”
Certainly true, and a beautiful red herring on your part. The question is not whether or not there is a simple or complex relationship but whether there is a correlation. And, of course, you wisely note that there is such a correlation.
Am I to take you as making a legitimate broader argument that climate is too filled with confounding to make simple correctives not based on instrumentation? Or am I to take this blow hard obfuscation about what happens when the wind blows hard?