Guest Commentary by Zeke Hausfather and Matthew Menne (NOAA)
The impact of urban heat islands (UHI) on temperature trends has long been a contentious area, with some studies finding no effect of urbanization on large-scale temperature trend and others finding large effects in certain regions. The issue has reached particular prominence on the blogs, with some claiming that the majority of the warming in the U.S. (or even the world) over the past century can be attributed to urbanization. We therefore set out to undertake a thorough examination of UHI in the Conterminous United States (CONUS), examining multiple ‘urban’ proxies, different methods of analysis, and temperature series with differing degrees of homogenization and urban-specific corrections (e.g. the GISTEMP nightlight method; Hansen et al, 2010). The paper reporting our results has just been published in the Journal of Geophysical Research.
In our paper (Hausfather et al, 2013) (pdf, alt. site), we found that urban-correlated biases account for between 14 and 21% of the rise in unadjusted minimum temperatures since 1895 and 6 to 9% since 1960. Homogenization of the monthly temperature data via NCDC’s Pairwise Homogenization Algorithm (PHA) removes the majority of this apparent urban bias, especially over the last 50 to 80 years. Moreover, results from the PHA using all available station data and using only data from stations classified as rural are broadly consistent, which provides strong evidence that the reduction of the urban warming signal by homogenization is a consequence of the real elimination of an urban warming bias present in the raw data rather than a consequence of simply forcing agreement between urban and rural station trends through a ‘spreading’ of the urban signal to series from nearby stations.
Homogenization is a somewhat complex term for a conceptually simple idea. Climate variations tend not to be purely local so changes in temperatures over long time spans (longer than a month) will be highly spatially correlated. Any major changes over time in individual stations that are not reflected in nearby stations are likely due to local (rather than regional) effects such as station moves, instrument changes, time of observation changes, or even such things as a tree growing over the thermometer stand. By removing any artifacts of individual station records not shared with other stations in their region, we can get a more accurate estimate of regional climate changes.
The conterminous United States (CONUS) has some of the most dense, publicly available digital surface temperature data in the world with over 7000 Cooperative Observer (Coop) stations reporting daily maximum and minimum temperature. This provides a unique resource to compare subsets of stations with various characteristics (e.g. urban form, sensor types, etc.) without suffering bias due to differing spatial coverage, a factor that often complicates global-scale studies of UHI. The Coop Program also maintains accurate station location data (roughly 30 meter accuracy), which allows for the accurate indexing of Coop stations against high-resolution spatial datasets that are useful for identifying urban and rural areas.
We use four datasets to classify stations as urban or rural, all with 1 km spatial resolution:
- Satellite nightlights – how bright a specific location is at night as observed from space.
- Impermeable surfaces (ISA) – what percent of the area is covered in concrete or similar materials.
- GRUMP – an urban boundary database using administrative borders and other factors (including nightlights) produced by Columbia University.
- Population growth – 1930 to 2000 population growth data interpolated to kilometer resolution using U.S. Census data.
We also used two different methods to compare urban and rural stations: a station pairing method, where we looked at all possible permutations of urban and rural stations within 100 miles (160 km) of each other for each urban proxy, and a spatial gridding method where we used a grid-based approach to calculate CONUS temperatures separately using only urban and rural stations and compared the results. For the station pairing method, we imposed additional restrictions that the pairs must both have the same instrument type, to avoid accidentally conflating bias due to urban-correlated differences in the frequency transition from liquid-in-glass thermometers to electric MMTS instruments with actual urban-related warming.
Finally, we examined six different versions of U.S. temperature data separately for both maximum and minimum temperatures:
- Raw station data with no adjustments
- Station data with only time-of-observation bias (TOBs) adjustments
- Station data with both TOBs and full PHA homogenization
- Station data with TOBs, full PHA homogenization, and GISTEMP satellite nightlight-based corrections
- Station data with both TOBs and rural-only PHA homogenization.
- Station data with both TOBs and urban-only PHA homogenization.
We created estimates of urban-rural differences for each of the four temperature proxies, two analysis methods, six temperature datasets, and maximum and minimum temperatures for a total of 96 different combinations.
As shown in Figure 2 from our paper, there are significant differences in the warming rate of urban and rural stations in the raw (and TOBs-adjusted) data that are largely eliminated by homogenization, even when that homogenization is limited to using only rural stations (to avoid the possibility of ‘spreading’ the urban signal).
This can also be seen in the figure below (from our paper’s supplementary information), which shows urban-rural differences over the 1895-2010 period using the spatial gridding method:
We conclude that homogenization does a good job at removing urban-correlated biases subsequent to 1930. Prior to that date, there are significantly fewer stations available in the network with which to detect breakpoints or localized trend biases, and homogenization does less well (though the newly released USHCN version 2.5 does substantially better than version 2.0). In general, there might be a need for additional urban-specific adjustments like those performed in NASA’s GISTEMP for areas and/or periods of time in which station density is sparse, but they are rather unnecessary for the post-1930s CONUS data. The simple take-away is that while UHI and other urban-correlated biases are real (and can have a big effect), current methods of detecting and correcting localized breakpoints are generally effective in removing that bias. Blog claims that UHI explains any substantial fraction of the recent warming in the US are just not supported by the data.
In case people are interested in playing around with our data or code and replicating our approach, all relevant materials to conduct the analysis (as well as versions of the code in both STATA and Java) are available on the NOAA NCDC FTP server. More detail on the specific methods used can be found in our paper, and our supplementary materials contain more detailed tests to ensure that homogenization was properly removing urban-correlated biases. This paper is also somewhat interesting as it arose out of a a blog post back in 2010, and represents a productive collaboration between a number of climate bloggers (Troy Masters, Ron Broberg, David Jones, and Zeke) with climate scientists at NCDC (Matt Menne and Claude Williams).
References
- D.E. Parker, "Large-scale warming is not urban", Nature, vol. 432, pp. 290-290, 2004. http://dx.doi.org/10.1038/432290a
- X. Yang, Y. Hou, and B. Chen, "Observed surface warming induced by urbanization in east China", Journal of Geophysical Research, vol. 116, 2011. http://dx.doi.org/10.1029/2010JD015452
- J. Hansen, R. Ruedy, M. Sato, and K. Lo, "GLOBAL SURFACE TEMPERATURE CHANGE", Reviews of Geophysics, vol. 48, 2010. http://dx.doi.org/10.1029/2010RG000345
- Z. Hausfather, M.J. Menne, C.N. Williams, T. Masters, R. Broberg, and D. Jones, "Quantifying the effect of urbanization on U.S. Historical Climatology Network temperature records", Journal of Geophysical Research: Atmospheres, vol. 118, pp. 481-494, 2013. http://dx.doi.org/10.1029/2012JD018509
Nice article! I’ve been waiting for an analysis like this for some time, now.
Zeke,
Can you explain why there was a greater increase in maximum temperatures in the urban stations compared to the rural stations. I read the paper, but could not find an explanation (although I may have missed it).
(For those who have already seen this, please bear with me — I’ve added a few new features, and I’ve packaged this up to make it easier to use).
For those who are interested, I bundled a very crude global-average temperature averaging algorithm with a Google Map front-end, and wrapped everything up in a portable, easy-to-use VirtualBox virtual-machine (VM) appliance file.
It can be downloaded from: http://tinyurl.com/NASA-HANSEN (Easy-to-remember short-url pointing to a Google-document file).
I didn’t plan to do this originally; it just sort of “evolved” as I experimented around with the data, Google-Maps/javascript/etc. (Thanx to Nick Stokes for the original javascript/html which provided me a nice starting point).
The VM file is a big download, unfortunately (about 1GB), but it was the easiest way for me eliminate hassles involving browser/software setup/configuration without going back and doing a proper software design from scratch ;)
With the VM file, you just download it, import it into VirtualBox (available for free at http://www.virtualbox.org), hit the VirtualBox “start” button, and go. It should run on any newer Mac (OSX 10.5 or newer) or Windows PC/Laptop (2GB or more memory strongly recommended). It boots up a Linux virtual machine which then automatically launches everything.
The basic algorithm is *very* simple-minded — just anomaly gridding/averaging, with grid-cells big enough that I didn’t have to worry about computing interpolated values for empty grid-cells.
I start with 20×20 degree cells at the Equator and adjust the longitude dimensions to keep the grid-cells *approximately* constant in area as you go N/S from the Equator. (This kept the grid-cells from shrinking in area — like I said, I wanted to minimize the number of empty grid-cells).
What you can do with this is “roll your own” global average temperature results by clicking on station icons on a Google Map display.
There are also javascript controls (thanks, Nick) that allow you to filter the displayed stations by data record length, start/end years, rural/urban status, etc.
Among other things, this makes it easier to identify the “long record” stations needed to get good results going back to the late 1800’s.
You can also generate “station batch” results (i.e., all stations, all rural stations, all stations with at least XXX years of data, etc.)
Raw and adjusted data results can be displayed together or individually (with the NASA/GISS results also displayed for comparison purposes).
The results are displayed in very simple GnuPlot window (nothing fancy).
The upper plot in the window shows the temperature results computed from GHCN V3 raw and adjusted data via the simple agorithm. In addition, the NASA/GISS “meteorological stations” index is plotted for comparison purposes.
The lower plot shows the number of selected stations that actually reported data for any given year. (The number is fractional; if a particular station reported data for 6 months in a given year, I count it as “half a station” for that year).
This will allow you to correlate the quality of the temperature results for any given time period with the actual number of stations that actually reported data for that period.
Play around with it a bit, and you will find how amazingly easy it is to confirm the NASA/GISS global-average results — just a few dozen stations scattered around the world will do it. Rural stations, urban stations, raw or homogenized data — all combos will produce largely similar global-average results.
There’s nothing of any real scientific interest here — it should be considered a demo tool that can be used to shoot down attacks made on the global temperature record by the usual suspects. What it does show very nicely is that you can confirm NASA/GISS with data from a very small number of stations run through a *very* rudimentary global-temperature algorithm.
And it’s simple enough for middle-school students to use.
[Response: Very nice. We did something similar (but even simpler) when it was being insinuated that the temperature trends were suspect, back when all those UEA emails were stolen. One only needs about 30 records, globally spaced, to get the global temperature history. This is because there is a spatial scale (roughly a Rossby radius) over which temperatures are going to be highly correlated for fundamental reasons of atmospheric dynamics. Anyone claiming significant issues with the temperature records has always been barking up very small tree.–eric]
Dan H.,
To be perfectly honest, we aren’t sure why the urban-correlated max biases are as large as they are in the TOBs data in recent years. It is interesting to note that it shows up much more prominently in the station pairing approach than the spatial gridding method, which makes sense given that station pairing tries to control for a max cooling bias due to MMTS transitions that is somewhat urban-correlated (e.g. airport locations and similar sites didn’t have the transition, while rural sites were slightly more likely to). Max biases are also significantly smaller over the century-scale period than the last 60 years (see Figs. 1 and 3 in our paper). That said, there definitely is some follow-up work to do in examining urban-correlated max biases in recent years in more detail.
First I just want to thank Gavin for maintaining one of my favorite sites on the www. Definitely advancing my progress on the Junior Climate Scientist merit badge ;-]
I am so excited to have an interesting and relevant comment to add to the discussion. Guy S. Callendar’s groundbreaking paper: Callendar, G. S. (1938), The artificial production of carbon dioxide and its influence on temperature. Q.J.R. Meteorol. Soc., 64: 223–240. doi: 10.1002/qj.49706427503 is a fascinating read. Probably the most noteworthy aspect of this paper is a graph (Fig. 2) which by my careful measurements predicts a rise in global average temperature of 0.59C from the 1938 level if atmospheric carbon dioxide were to reach 400 ppm. Callendar thought that would happen around 2100 so he gets no crystal ball award but it seems that his scientific prediction is about as accurate as any. Rather remarkable coming from a time when published papers included hand drawn and lettered graphs.
I have never seen anyone mention Callender’s addressing of the urban heat island effect in his calculations of the then current global average temperature. Callendar was explicitly aware of this possible error and addressed it by comparing isolated “Best Exposures” stations to those in small towns and those in large towns. Averaging the temperatue rise since the earliest days of systematic monitoring for these three categories of stations found (respectively) rises of 0.23C, 0.19C and 0.21C. Callendar concluded, “This shows that no secular increase of temperature, due to ‘city influence,’ has occurred at these city stations, in spite of the great increase of population in the immediate neighbourhood during the period under consideration.”
So, including the BEST study and the new Hausfather paper, how many times has the urban heat island effect been addressed?
[Response: A quick search of AGU journals reveals at least 100 publications. This has been addressed many times. Nothing against Zeke’s excellent new work, but the first-order answer — and the only one that matters if the question is “what is the mean trend in temperature in the last 100 years — has been known pretty well for a decade or two. Nothing really substantial has changed. –eric]
I might have missed something but do you prove that urban sites are adjusted down rather than rural up?
Rikard avfuktare,
Effectively yes. We do a homogenization run using only rural stations to homogenize all stations (excluding urban stations from the homogenization process) and get effectively the same results. Our supplementary materials go into a bit more detail in examining this.
Is there any adjustments made for sites near reservoirs?
I’d imagine these could have large local effects in rural areas.
John Cartmill,
The PHA looks at different series between stations and their neighbors over a window of time. Things that cause absolute temperature differences shouldn’t really impact the results, unless a reservoir was created near an existing station, in which case it might result in a break point that is detectable.
“Is there any adjustments made for sites near reservoirs?
I’d imagine these could have large local effects in rural areas.”
There is a very intersting case of this in the Orland record of GISS. Have a look at the record pre and post adjustment and then consider when the dam was built there.
At one point I started to collect metadata on all the dams of the world to see if the phenomena was repeated at other locations. Hmm, I need to go back and do that someday.
Zeke,
Thank you for your response. It would be interesting to see a follow up regarding max temperatures. Gordon asked how often the UHI has been addressed, and I would say quite often in the U.S., owing partly to the preponderant temperature data available. How well would you say that temperature data in other parts of the globe have accounted for the UHI (Europe excluded)?
Questions I’m asking myself, will get to poking after them as time allows. Or, of course, answers and pointers welcome.
Has anyone correlated day by day or hour by hour measurements of other trace gases, particles, and humidity with temperature? Ruled out an urban smog-heating effect for example?
What other molecules are known to be concentrated around cities (‘point’ of origin) from which they’d diffuse or blow away, thinning out with distance?
How does air temperature correlate with sky temperature, what you get pointing an infrared thermometer at the sky, on those inversion-layer days and nights?
I’m glad to see the study looking at how much pavement is in place. Is that thinking about how much rainfall evaporates into the local air (thinking of clouds of steam rising off hot pavement after rainstorms) — as well as how much rain goes into the ground locally available for plants, vs. how much is drained off to storage in tanks or to the sewage plants in the suburbs?
John Cartmill.
The effect you are looking for can be seen in the Orland station in the US.
A dam was built in the early 20th century. At one point I started to look at GIS data on damns and resevoirs, probably need to finish that.
Rikard avfuktare:
“I might have missed something but do you prove that urban sites are adjusted down rather than rural up?”
To follow up on Zeke’s response, I think the figure you are looking for is Figure 9 (and this important issue is examined more in the SI as well). As you can see, if you adjust the USHCN stations by *only* urban CONUS stations, then the urban stations (obviously) have a tendency to adjust the rural stations upwards. But as should be apparent in that figure, homogenizing using the full set of CONUS stations (as the USHCN v2 does) produces a time series that is very similar to homogenization using *only* rural stations.
If you are interested in seeing the time series when only the most rural stations are used (both for homogenization and actual station data, with ISA < 1%, which is sparse enough early that it misses many of the breaks), I have put up a figure in my post:
http://troyca.wordpress.com/2013/02/13/our-paper-on-uhi-in-ushcn-is-now-published/
To the individual who is still wondering about whether UHI has been properly accounted for in the global-average temperature results published by NASA/NOAA/etc., all I have to say is, why don’t you roll up your sleeves and crunch the temperature data yourself?
All of the data and software/data-crunching tools needed to do that are freely available on the web and are just a few mouse-clicks away. You guys have been going on and on about this for *years* now. If this is such a pressing issue for you, then why haven’t you bothered to perform any of your own data analysis in all that time? This is not rocket-science — it’s #*&!ing averaging, for Pete’s sake.
BTW, here’s what I was able to produce with the VM appliance (that I posted about above) and a bit of mouse-clicking:
Global-average results for 30 *rural* stations (none of them in Northern Europe or the Continental USA) right here: http://img69.imageshack.us/img69/675/screenshot20130213at806.png
I took me nearly as long to make a screenshot image of the results and upload it to ImageShack as it did to generate the results.
There is a bit of a departure between the NASA results (red) and the raw data results (green) prior to about 1930 — but then look at the number of selected stations that reported data for those years — 20 or fewer.
Challenge to “skeptics” — fire up the VM file per the instructions and pick any 50 stations (rural or urban) scattered around the world. I am highly confident that you will *not* be able to generate results with a long-term temperature trend that contradicts the NASA global temperature trend (unless you cherry-pick “short record” stations that result in poor global coverage for significant time periods).
This is probably a silly layman question on UHI, but aren’t those stations simply measuring actual temperatures, i.e., couldn’t growing urbanization simply be regarded as another anthropogenic forcing? Why is it even necessary to adjust for UHI if what those stations are measuring is the actual temperature?
[Response: The main issue here is that if one were to use an urban station to represent the temperature change over a much larger area than that represented by the urban area itself, then there would be bias. A second order thing one might also be interested in what it would be without the albedo changes of cities. But the first order thing is more important: what to do if there are lots of stations per unit area in cities, and fewer in the countryside. In general, this is the case, hence the need for these sorts of calculations. –eric]
Tangential to this silly question, I’ve often wished realclimate.org had a section where us climate science-loving laymen could post questions. I’m perfectly content to google and wiki away to advance my climate knowledge. I read the books (Hansen, Mann, Alley, etc., etc.), peruse the best climate websites (and occasionally WUWT or Morano’s site, painful as it is, under the know-thine-enemy premise. I even pick up old Science, Scientific American and Nature mags at my local library for a dime when they have climate articles, and read those articles over and over until I grasp most of what’s in ’em to a satisfactory degree. However, occasionally a question pops into my head that I can’t answer with a moderate amount of research and reading, and I’d love to simply be able to pose such questions online and get an answer, or at least some direction toward an answer, such as a link to a paper (preferrably NOT one behind the paywall). Have y’all ever thought of adding such a feature to realclimate.org. I realize you are all quite busy fending off denialists and doing the occasional bit of research, but for you folks, the kinds of questions I’m thinking of would be relatively easy and not take up much of your time (in theory anyway).
Love your website. I used to “waste” my time doing difficult sudoku puzzles, but once I “discovered” the fascinating and challenging world of climatology, I realized it was somewhat akin to the most challenging sudoku puzzle ever, and one that I’d never, much to my delight, ever complete.
Thanking you in advance.
Jim
Hank,
There are some papers that address the aerosol issue for UHI, but the largest effects are the area of land surface that has been transformed: It hits the albedo, the surface roughness, the emmisivity, the heat storage.
Below 10% ISA you wont see an effect unless that 10% happens to be in the footprint of the sensor ( say within 100x of the sensor height )
In my books that urban heat, as well as heat generated by producing electricity through fossil fuels, even nukes, away from cities, driving ICE cars, etc is all anthropogenic environmental harms, in addition to the human-enhance GH effect warming. It can all combine to cause heat deaths and other problems….not to mention harms from the concomitant local and regional pollution produced when producing GHGs. There is the synthetic or holistic outcome, beyond just the outcomes of each taken individually.
Has anyone done a study of the overall effect of all this combined together?
I’m thinking if we do a cost-benefits analysis (somehow adding in costs on into the future for 100 years — tho it should be 100,000 years), and do it in terms a human lives and well-being scale instead of $$$, it seems to me the costs would far outweight the benefits (of using fossil fuels and even of nuclear power — for which one must also throw in the harms to and deaths of tribal peoples in uranium areas — Niger, the U.S. Southwest, the Bennett Freeze area, etc.).
Or, if one insisted it be in $$$, then figure it in terms of hypothetical lawsuit settlements for each person harmed or killed or having his/her life shortened (even figuring that men are worth more than women, rich more than the poor, and adults more than children or the elderly….as mentioned in A CIVIL ACTION).
How might UHI impacts be altered if urban albedo changes from ‘ white roofs’ initiatives were amplified by equally deliberate brightening of urban and suburban water surfaces?
Eric,
I agree that there was little doubt that global UHI is not going to be that large (after all, some existing series like GISTemp already do explicit UHI corrections and don’t find immensely different results). That said, I wouldn’t necessarily classify it as a settled issue. While Parker found relatively little effect, other studies have found larger regional effects (e.g. Jones in China). We actually ended up finding a larger effect in the raw data than we had initially expected (e.g. 14% to 21% of the century-scale trend). The fact it is effectively identified and corrected by automated homogenization is good to know, as these approaches are being increasingly applied to global temperature data (in GHCN v3, for example). When station networks are sufficiently dense to detect anomalous local breakpoints or trends via pairwise comparison, additional explicit UHI corrections (like GISTemp’s nightlights) may not be needed.
[Response: Zeke, thanks for the response. I agree, it is not “settled”. But then, nothing ever is. The important point is that while some use the lack of total perfection as an excuse to act as if we don’t know anything, we have known for a very long time that the impact is not big — that is, not big enough to change the question (has the earth warmed up in response to CO2?) or the answer (yes). I would furthermore wager that no amount of refinement about UHI is going to make a meaningful dent in our ability to estimate climate sensitivity (equilibrium or otherwise). Further improving UHI calculations is important, but its relevance to the big picture has been exaggerated to a ridiculous degree in the blogosphere. To borrow from Kuhn, this is “normal science”: no revolution is to be expected here. I have read your paper, and I think it is a solid and very worthy contribution.–eric]
> brightening … water surfaces
Have any actual experiments been done? Maybe by someone who owns a swimming pool, or a farm pond? I’d be very curious how such a layer of microbubbles affects various creatures that use the water surface, from human swimmers to water striders to mosquito larvae to migratory waterfowl.
I recall hearing a story about a college town police chief, back in the 1950s, phoning the college science department to ask what they knew about the layer of white foam on the city reservoir, and why the ducks that landed in that water were sinking and drowning. A container of some very powerful detergent was missing from the supply cabinet.
Hank, what are you talking about ? Certainly not the phenomena discussed in this month’s Physics Today , or the paper that appeared in Climatic Change last year.
Eric
‘A quick search of AGU journals reveals at least 100 publications. This has been addressed many times. Nothing against Zeke’s excellent new work, but the first-order answer — and the only one that matters if the question is “what is the mean trend in temperature in the last 100 years — has been known pretty well for a decade or two. Nothing really substantial has changed. –eric]
I think your quick search might be a bit broad. With regard to studies of UHI in the US, there are less than a handful and none of them satisfactory. I won’t belabor the details of each one, but I think its fair to say that Zeke’s paper represents the most comprehensive study of UHI in the US record that has been done.
Nobody who worked on this projected expected ‘revolutionary’ results and responses like “we knew that already’ are not really on point or charitable.
It’s good science, why not leave it at that. It advances our understanding, why not leave it at that? They achieved what they set out to do. Why not thank them for the contribution and keep your opinions about the grander meaning to yourself? [edit] here you have an opportunity to give nothing but praise to Zeke, and the other guys who took the advice given. They took up the challenge and did their own damn science. And its damn good science. You take an opportunity for showing other citizens how they can contribute and you play a stupid game of well, its not revolutionary.. its nothing in the grander picture. Talking with Zeke throughout the course of this project I held out hope that folks here would showcase this as prime example of how citizens can make contributions. You pooped on it. why?
[Response: The irony of your repeating tired old insults, while both completely misinterpreting my statement, and accusing us of playing “stupid games” is pretty ironic. Meanwhile, I stand by my point: Zeke’s work is excellent, but its significance is small. The latter point is relevant because there are clearly many that think that the UHI effect could be “game changing” if only it were accounted for. (It not always their fault, of course, thanks to the huge amount of speculation on UHI promoted by the various self-appointed auditors in the blogosphere.) Kudos to Zeke for breaking out of that mold and looking at the facts. You might try following his lead.–eric]
Mosher,
I don’t think Eric intended to be dismissive. He just pointed out that, while interesting, the paper doesn’t significantly change out best estimates of temperature records. Not being particularly groundbreaking is one of the downsides of finding a small UHI effect in the homogenized data :-).
(As an aside, I always wonder why some folks accuse scientists of following a consensus. Finding results the fly in the face of common knowledge is much more interesting than just confirming things that most people already believed, and is arguably easier to publish).
Thanks Russell, no, I’m just not sufficiently well educated yet.
I know there are big differences between your methods and the old detergent-foam stuff — just not very clear on what’s going to be different.
Do mosquito larvae successfully get their breathing tubes up through a layer of microbubbles, or can they wait out a period til the bubbles go away?
Does a layer of microbubbles change the environment in interesting ways for water striders, diving spiders, tadpoles, or kids swimming in that water?
What happens to viruses and bacteria in the surface layer?
Does a layer of microbubbles make a sound-reflective layer difference?
(and can I get a microbubble-generating pump to pump fire retarding foam? I’d guess whatever you use would be less nasty chemically than what’s used by firefighters now and comparably effective in cooling and smothering a fire or temporarily blocking ignition?)
(I’m doing botanical restoration on a forest fire site that a fire crew defended back in the late 1980s, as their camp was on that location — and the stumps they sprayed to stop them burning are still surrounded by a circle of dead ground, while the surrounding area is growing — forest service botanist said yeah, it’s something in their firefighting spray foam stuff, nobody knows…..)
Oops, pardon the tangent. Leave those questions for later somewhere else, I’ll keep watching that subject.
> keep your opinions about the grander meaning to yourself?
Any factor — UHI, sunspots, cosmic rays, whale poop — after it’s been weighed against the overall change that’s happening, isn’t opinion.
“You pooped on it. why?”
Eric:
“I have read your paper, and I think it is a solid and very worthy contribution.”
My, that’s some remarkably unstinky poop eric laid on it …
25:
What “layer of microbubbles” Hank?
Inverse clouds of micron sized microbubles are subject to Stokes law, and no more stratify than cloud droplets. They typically don’t rise fast enough to reach the surface before dissipating by gas solution.
Mosher why do you try and spin things so it seems like people are being nasty?
” Nothing against Zeke’s excellent new work…”
Eric says it was excellent and you claim he pooped on it.
Please don’t derail an intelligent discussion with your own hypersensitivity.
> don’t rise …
By ‘layer’ I meant — wherever they are, in the water. Thinking of having ships create them — below the waterline and above the keel?
Nev’mind. I’ll look for something with pictures.
http://adsabs.harvard.edu/abs/2012EGUGA..14.7007P indicates people who can do the work are thinking about the kinds of questions that occurred to me. That’ll do. It’ll work out. Like anything else, it’s a huge subject growing fast.
Hank- here’s the correct link to the Physics Today piece on Arctic albedo loss, and water brightening as a potential means of addressing it .
I’m not exactly clear how giving someone a guest post and expecting them not to freak out when subjected to normal scientific rigor is anything but offering respect and praise.
Only in an Alice in Wonderland universe can one find this kind of double standard regarded as honest scrutiny. Consider Mann and Cuccinelli (or Phil Jones and … or any other fake skeptic fake heroics) before you talk of persecution.
For modern temperature calculations, is there any reason to mix urban and rural temperatures at all? The post’s technique elevates rural above urban, which, while reasonable, might not be as good (nor as easy) as just defining stations and area as rural or urban and keeping them separate for all purposes. Sum by area in the end. This retains the, as others pointed out, quite valid UHI influence instead of making the error of adjusting it out. It ain’t error. It’s real warming.
I have a question that kind of spins off from. From what I can tell, a very rough estimate of the area of world wide roads is somewhere between 10% and 17% of “normal” icecaps (given that icecaps, even with global warming are dynamic systems). I’ve read discussions of “cool roofs” or “white roofs” for years. Would a “white roads’ or “cool roads” program that painted roads a light color compensate for about the same square mileage of icecap loss, thus mitigating one of the worrisome feedbacks a tiny bit? Not a net profit mitigation like “cool roofs” but possibly cheap for what accomplished.
An obvious question is whether this work points to any correction in existing data sets like GIStemp that already use homogenization and correct for UHI. At least I think it does. A google search on ‘GIStemp homogenization’ results in a lot of pages with the following when I attempt to reach them
[Response: The GISS website is undergoing a technical upgrade and a lot of the interactive stuff is still offline. It should be back up soon. – gavin]
What are the results if one looks only at inland cities and inland rural areas? Population movement might be a confrounding factor. USA urban areas are increasingly near oceans, which probably have a more moderating effect on temperatures nearby than in inland areas.
Zeke,
I’m curious why you only included census data up to 2000 when 2010 census numbers (which show a 9.7% population increase over 2000) are available.
Doesn’t it get frustrating having to whack moles repeatedly?
I’ve been barking up the Arctic tree for about three years now, so have got out of touch with wider issues such as UHI. But the last time I looked at it, some four years ago, I concluded it was being adequately accounted for and moved on with my life.
> why … spin things so it seems like people are being nasty?
He wants to make sure it’s clear the paper’s treated like real science, rather than like blog science?
This is how it works: you put your model out there in the coliseum, and a bunch of guys in white coats kick the shit out of it. If it’s still alive when the dust clears, your brainchild receives conditional acceptance. It does not get rejected. This time.
“He wants to make sure it’s clear the paper’s treated like real science, rather than like blog science?”
I didn’t think Mosher liked real science. Have you ever seen his definition of ‘Lukewarmer’. It’s rather ummmm variable.
Good work.
Marcos,
The NOAA dataset with 1 km resolution census data is the output of a model using the highest available resolution census data (which is of variable sizes, but tends to be a tad bigger than 1 km in less populated areas). As the Census doesn’t publish an official 1 km gridded dataset, and the NOAA data only goes through 2000, we were not able to include 2010 population density/growth in the analysis. That said, given the time periods involved I doubt it would have significantly affected the results.
As a cab driver on the night shift I see temperature differentials that seem to average between 1 and 2 degrees cooler on the outskirts of my city of over 1 million people than in the core of the city. Why are the adjustments in this article so small compared to my real life experience?
[Response: The UHI effect can indeed be much larger in practice that what is being addressed here — that is, cities are very warm compared with non-urban surroundings. But most of the weather stations are not downtown. They are out at the airport, or in suburban (previously countryside) areas, and they like. They are affected by the urban warmth effect, but generally not that much. Does that help? –eric]
Adrian Smits:
“As a cab driver on the night shift I see temperature differentials that seem to average between 1 and 2 degrees cooler on the outskirts of my city of over 1 million people than in the core of the city. Why are the adjustments in this article so small compared to my real life experience?”
Because that’s what the data shows.
If you have a long-term time-series of measurements with accurate lat/long tags taken with reliable instruments, I’m sure somewhere, someone will be willing to incorporate those measurements into the datasets for your cities.
On the other hand, anectdotal evidence is pretty much useless. For trend analysis, of course, you would have had to have taken a lot of measurements around much of your city for a long period of time for any impact on trend to have been detected.
Adrian:
In addition to Eric’s inline response, keep in mind that errors in the long-term trend would require that the difference between the core and outskirts be changing. The 2013 downtown temperature is being compared to the 1970 downtown temperature, and the 2013 rural temperature is being compared to the 1970 rural temperature. (Well, actually, each to the corresponding mean for the base period, usually a thirty-year period). It’s not enough for the two locations to be different – they have to have different trends. The numbers in the article refer to the trends, not the absolute temperature.
As a follow up to Bob Loblaw #47:
Indeed there have been a number of papers in this area. A couple on London:
http://onlinelibrary.wiley.com/doi/10.1002/wea.679/abstract (paywalled, first page free)
http://onlinelibrary.wiley.com/doi/10.1002/wea.432/abstract (this one is OA)
The latter papers findings were noted to also hold for a number of Australian cities:
http://onlinelibrary.wiley.com/doi/10.1002/joc.3530/abstract (again, paywalled I’m afraid).
Basically sites in urban centres that are and always have been developed and where sprawl has not appreciably occurred can exhibit trend behaviour strikingly similar to surrounding rural stations. Equally, there are a huge number of papers suggesting that for areas experiencing very rapid development urban influences can dwarf the regional trend on a site specific basis in the raw record. So, for long-term regional trend characterization it is not black and white in the raw data. Then there is the issue of whether the adjusted / homogenized records account for this which is where the paper that is the subject of this post comes in.
The division between “urban” and “rural” in the linked paper is arbitrary.
I would recommend looking at indicators such as: “minimum temperature” instead of average temperature. Also focus more on temperature minimums in winter season.
Some questions I would like to see answered are”:
– when is the effect of Urban heat island greatest? (radiation versus advection)
– what is the UHI effect on snow cover (max snow depth, days with snow): my hypothesis is that it is even more remarkable than on night minimum temperature..
– How far beyond city edge does the urban heat island reach? Is there difference such as windward / leeward side of the city?
And now an important consideration: What if most (if not all) of the “rural” stations are in fact affected by urban heat island? Just consider the increase in built-up areas and asphalt in “rural” areas. For example roads, or industrial zones next to highway in an otherwise rural area. For the specific use-case of U.S I would recommend only using the stations in natural areas (national forest, national park) as truly rural.
One example:
Station Praha-Klementinum with measurements already in the 19th century. Maximum-temperature records are being broken at this station on a yearly basis sometimes several days in a year. However: The last time a daily minimum record was broken was back in the nineties. The last time a daily minimum record in winter was broken goes even further in history, to the 1980s. Station is in center of Prague, a fast-growing city by built-up area. I would be interested in the following relationship: – pick a station in city-center
– find how the distance of this station to the edge of the continuous built-up area changed (or look how the total built-up area of the city where the station is located had grown)
– I suspect, that you will find a strong correlation between the built-up area growth and the “minimum temperature in winter in city center” growth.
[Response: Have you actually read the paper? I’m guessing not. In any case, calling their method “arbitrary” is not a very compelling way to engage in a discussion about it.–eric]
For modern temperature calculations, is there any reason to mix urban and rural temperatures at all?