A Proxy Indicator is Not What You Think It Is

Spread the love

I want to say a word about what a proxyindicator is. And isn’t.

I noticed that the term is not in some, perhaps many, dictionaries, so I guess this leaves me free to do what I want with it! But wait, the term “proxy” is of course in the dictionary. It is an ancient short version of the word “procuracy” which is the authority to act for another. Thus, a proxy vote. Proxyindicator (or proxy indicator) is a term widely used in climate science though it is used in many other fields as well to refer to a measurement that is indirect, or more accurately, that stands in for the direct measurement. For instance, the number of women in positions of authority in various national governments is a proxy indicator for perceptions of equality of women in those various nations.

I first heard the term in a class … the first class I ever took that had anything to do with human evolution. The instructor was John Barry, and he referred to Proxy Indicators of ancient climate. And it is in this manner that I’ve mostly heard and used the term since then. I quickly add, those in the know usually abbreviate it to the short form “proxy.”

And the word is misunderstood in a very important way that I’d like to address.

Consider temperature. Right now, I can see that the temperature is 73 degrees F because my gnome toolbar says so. I was just in the car and it said 77 degrees there, on the dashboard, and when I drove by the bank a moment ago it said 74 degrees. Our house thermometer (everyone has some sort of thermometer, right?) says 76, and projects it on the ceiling along with he time. The thermometer itself is outside on the porch in a plastic bag in the shade, and communicates with the laser projecting alarm clock via radio waves.

Those are all thermometers that give a direct measurement of the temperature, right? Of course, a temperature measurement related to paleoclimate would not be a daily measure, but perhaps an average over a year’s time. And a proxy indicator would be some indirect measurement of that, like which species of gerbil lives under the porch or which kind of plankton we might find in the aquarium.

In this sense, the temperature is measured directly with the various thermometers, and a proxy for temperature would be some ecological measure that might be quite general or even inaccurate. But a proxy could also be chemical. For instance, I could take the temperature every day for a year and average it out, then I could look at the isotopes of the biofilm that is constantly forming on the curb outside my house. In different temperatures, perhaps different isotopes act differently, so the average temperature from all my thermometers and the characteristics of the biofilm’s isotopic imposition would be linked using a model. The model would have to do with how chemicals work in different temperatures while the thermometers are giving a direct measurement of the actual, real temperature.

Right?

Well, no.

The proxy is in fact something giving an indirect measurement of temperature, via the ecological limits of a plant or animal or a chemical reaction that works differently in different temperatures. But this is also true of the direct measurement.

Well, very few thermometers use gerbils or plankton, but they do all use an indirect measurement. Most thermometers in the past used the expansion and contraction of a liquid, such as mercury or alcohol, in a tube. The temperature reading is not a reading of temperature at all. It is a measure of volume. The volume at any given moment is an indirect effect of the actual temperature in the same way that isotopic measurements are an indirect effect of actual temperature. Another kind of measuring device uses the differential expansion and contraction of different kinds of metals. A bi-metallic strip bends to different degrees depending on temperature, and that can be an indirect measurement of the actual temperature of the air. Such a thermometer measured bent-ness of metal, not temperature.

It is not the case that these mechanical proxies (liquid expansion or bi-metallic strips) are automatically more accurate than the other methods. Up at the cabin, I can estimate the annual temperature and humidity by noting that white pines are the predominant tree, pike, walleye and bass the predominant large fish, which frogs and toads are common, that there are loons but not tundra swans nesting, and so on and so forth. I could also estimate the annual temperature by observing the bear thingie. The bear thingie is a stake with a wooden bear on it, and the bear is holding a thermometer in one hand and a rain gage in the other, and there is a wind indicator sticking out of its head. It is outdoor Kabin-Kitsch and it is designed to give you basic weather data and be cute at the same time. When I use my knowledge of North American Ecology to estimate the annual temperature of the area around the cabin, I come up with about 40 degrees F. When I look at the climate data for the region I find that the average temperature is about 40 degrees F. The trees and fish and birds told me the temperature quite accurately. When I look at the bear holding the thermometer, I get a different number. According to the bear, the average temperature at the cabin over the year is 125 degrees F. Yes, it is true that the thermometer that bear is holding has been broken for a year. But it makes the point: It is not measuring the temperature. At the moment, it is measuring some very bent metal. When it was working it was measuring some less bent medal.

There is another kind of temperature indicator that is usually considered to be direct but is, again, not a direct measurement of temperature. It uses physics. Various materials conduct electricity. In most cases, the actual amount of conductivity depends on other factors like temperature. So, some materials can be used to estimate temperature by running electricity through them at a constant rate, then seeing how much gets through. The difference from time to time will be mainly due to temperature changes.

Now, I ask you, is a person who believes that physicists don’t’ really understand fission (and thus can’t use Potassium-Argon dating to estimate the age of old fossils) or who doesn’t think that mathematicians know how to make climate models (and can thus reject most climate related research) also going to believe that we understand the way electricity varies with temperature as it passes through a conductive substance? No, of course not! That would be entirely inconsistent. You can’t reject evolutionary science, climate science, vaccine science, or any other science you somehow find inconvenient or contrary to your beliefs, and math too boot, and then blithely accept the same methods of understanding the world when it comes to the thermometers in your clock, in your car, or in your bear.

All measurements of temperature are proxies. And whatever you were thinking about some scale of believability or applicability across different kinds of proxies may very well be incorrect.

Have you read the breakthrough novel of the year? When you are done with that, try:

In Search of Sungudogo by Greg Laden, now in Kindle or Paperback
*Please note:
Links to books and other items on this page and elsewhere on Greg Ladens' blog may send you to Amazon, where I am a registered affiliate. As an Amazon Associate I earn from qualifying purchases, which helps to fund this site.

Spread the love

20 thoughts on “A Proxy Indicator is Not What You Think It Is

  1. Interesting. I assume a related post would be advantages and disadvantages of different types of proxies, and there similarities/dissimilarities ’round the globe.

    On a lighter note:
    This:

    “Well, very few thermometers use gerbils”

    is one of, if not the most, amusing introductory portions of a sentence I’ve seen. I imagine one needs to look into some of the dingiest corners of the internet to find a gerbil-thermometer.

  2. Cool.
    Then warm again.

    And your blog just had an ad for the University I used to work for. That’s cool too. And then warm again…

  3. A 2006 paper by Yom-Tov and Geffen finds that minimum January temperature is negatively associated with a composite measure of Gerbillus dasyurus body size. In their model, minimum temperature explains 10% of the variability in gerbil body size among the specimens recovered from various environments (n=127). This is taking account rainfall, sex, maximum temperature, year, month, and the interaction between temperature and rainfall.

    It would take some work, but we probably could come up with a predictive model for annual average temperature based on gerbil size.

    Yoram Yom-Tov and Eli Geffen. 2006. Geographic Variation in Body Size: The Effects of Ambient Temperature and Precipitation. Oecologia, Vol. 148, No. 2. pp. 213-218.
    http://dx.doi.org/10.1007%2Fs00442-006-0364-9

  4. While your point is well taken, it seems to me you have glossed over the important difference between forward engineering and reverse engineering. When we design a thermometer, regardless of the physics it uses, we can calibrate it against other measures of temperature, test it in a variety of different environments, and build in checks of whether it is still functioning as expected. Using the results of all of that, we can determine an appropriate domain of use, deploy as desired for scope of coverage and redundancy, select the technology that best fits the desired measurements, and replace it with ever better thermometers as technology improves. Importantly, we can leave a trail documenting those decisions, so that future engineers and scientists have a leg up on evaluating the results.

    Reverse engineering is more difficult. The past cannot be replayed. No engineers have been instrumenting it, except for a very short recent period. The non-engineered proxies are haphazard and calibrated only in retrospect. You’re right, of course, that physics is physics, whether it is isotopes trapped in ice cores or mercury in a thermometer. But there is a real difference that occurs when people start focused measuring, designing devices for that purpose, and documenting the engineering course along the way.

  5. Russel, ancient proxy data is occasionally reproduced experimentally in the present; The chemical proxies (fractionation of isotopes) are not at all historical. The best ancient animal or plant proxies are running now and can be observed; The expansion/co traction of liquids and changes in metal are natural phenomena noted by people then applied to temperature measurement mostly ad hoc and in an experimental way. The forward engineering concept is a late 20th century one, most “direct” temperature measurements are 19th century technologies or older.

    So, I think you are imposing a sterile western heter-dilbertesque world view on a more complex, nuanced, and context-dependent reality.

  6. Greg Laden:
    The forward engineering concept is a late 20th century one, most “direct” temperature measurements are 19th century technologies or older.

    I think you’re significantly underestimating the understanding of engineering process in the 19th century. And even earlier.

    And significantly denigrating “experimental” by conflating it with “ad hoc”!

  7. “GL:
    Our house thermometer (everyone has some sort of thermometer, right?) says 76, and projects it on the ceiling along with he time.”
    Is this a proxy-indicator that you shop at Spencers?

    “GL:
    All measurements of temperature are proxies.”
    This is trivially true no? Even if you had a direct measure device, you’d’ve to read it, and that means that your reading is a proxy for the ‘real’ visual image, yadayadayada.

    “Russell:
    Importantly, we can leave a trail documenting those decisions, so that future engineers and scientists have a leg up on evaluating the results.”
    What does this have to do with absolutely-direct vs. indirect-and-therefore-proxy measurements?

    The only way to directly measure temperature is to measure the motion of all the molecules that make up a substance, and then find the statistical average of that motion. That’s the definition of temperature, the only way to make a direct measurement is probably thru the definition of the thing being measured.

    I have to wonder though, are typical-proxies called proxies because we recognize that they’re not measuring temperature in the past directly, OR is it because there’s an even bigger difference between going back in time and dipping a thermometer into the sea-surface over the course of a year and averaging that and calling that Sea-Surface-Temperature versus isotopic-measurements in fossil foraminifera, where whats being measured is temperature over a HUGE area over a HUGE span of time?

    Also, consider that oceanographers have multiple methods of measuring, say, salinity, and they’re often, rather than say ‘salinity was this’ say ‘salinity-as-determined-through-this-method was this” (indicating method via units, which are particular to different methods, conductance, g/kg, PSU (practical salinity units), etc). So thats in ‘live’ oceanography, but then in paleoceanography we’re calling the measurement ‘devices’ (say, again, foram test isotope ratios) “proxies”, to distinguish them from the ‘normal’ “live” measurements.

    And with that I think I met my quote-mark quota.

    “Gl:
    So, I think you are imposing a sterile western heter-dilbertesque world view on a more complex, nuanced, and context-dependent reality.”
    But that has nothing to do with your argument, which is that determinations that creationists find acceptable are equally theory-(greg)-laden as the ones they reject.

  8. Russell, I would never underestimate 19th century engineering. I did 19th century industrial archaeology too long to do that. But I am sticking to my guns: The dichotomy between “direct” and “proxy” is over-played and assumed to have more meaning than it does. It is not without meaning, but the truth is that a bi-metallic spring thermometer beining held by a bear is not automatically better than counting how many times a cricket chirps per minute.

  9. Schenk, first, I have to say that there are annual records where the forams show a SST estimate for summer over one year, tends of thousands of years ago. This is in comparison to, say, a satellite overhead using radiation from thousandths of meters away to estimate SST. The latter is daily but we end up using the measure averaged over the year in long term models anyway.

    But yes, the proxy term really is for “live” vs. “not live” as you suggest. My point is that denialists simply don’t believe proxies because they are indirect and use information that does not have a real time analog, but that disbelief is based on a false dichotomy.

  10. Russell, perhaps a better analogy would be with distance scales in astronomy. For stars that are sufficiently close, we can measure parallax from one side of Earth’s orbit to the other (the parsec is derived from this measurement technique). But stars that are too far away do not show a measurable parallax, so we have to estimate distances by other means. These other means involve so-called standard candles: Cepheid variables have a certain period-brightness relationship which lets us derive distances for nearby galaxies, and supernovae let us derive distances for more distant galaxies. In each case we calibrate a proposed distance measurement against one that is known to work at closer distances and then use it to calibrate one that is proposed for even longer distances. Today we would use the Hubble relation instead for extragalactic distances, but when I was in grad school the value of the Hubble constant was still a hot topic of debate (some thought it was ~50 km/s/Mpc, others thought it was ~100, where the actual value turned out to be around 72). But it was only after the value of the Hubble constant was settled that we could use this systematic approach.

    Likewise with temperature measurements. Mercury thermometers work by assuming the rate of volume expansion with temperature is constant. This is a good approximation for temperatures between -20 and +100 C but obviously doesn’t work below mercury’s freezing point (about -40 C) or above its boiling point, and there are circumstances within this range where you can’t use a mercury thermometer (e.g., biological systems, since mercury is toxic). So other means were devised to extend the measurable range and calibrated against mercury thermometers in the range where both are valid. For parts of the temperature range you need to cascade standards because phase transitions and other nonlinearities intervene. We now understand, thanks to quantum mechanics (a 20th century development), why these standards work, and we can devise a fully consistent systematic approach to measuring temperature. In the 19th century people knew this was possible but did not know how to do it.

  11. OK, now imma point out that in the late 60s and through the 70s, there were vast regions of the sea where proxies based on cores brought up by oceanic vessles gave us annual temperature records for thousands of years past in regions where we were not collecting equally good “direct” records for the present. The southern Indian Ocean and the Red Sea come to mind.

  12. I don’t mean to be argumentative, because I agree with the point made, about all measurements relying on some degree of indirection. And Greg describes that beautifully.

    I’m only interjecting that there is another dimension, that does have importance when reconstructing the natural past, in that all the measurements are of necessity happenstance rather than designed.

    Which gives me a thought. It would be neat to have a wall chart whose horizontal dimension is time into the past, logarithmic, and that lists vertically all (well, many) of the different techniques for measuring the age of stuff, and then draws a line for each showing the range to which it has been applied, with some kind of cross-lines for showing where different techniques are checked against each other. Anyone know of such?

  13. I don’t mean to be argumentative

    Oh yeah?

    I’m only interjecting that there is another dimension, that does have importance when reconstructing the natural past, in that all the measurements are of necessity happenstance rather than designed.

    There may be a difference between designed and happenstance, but, I think this is still partly a false dichotomy. How was the resistivity thermoeter invented? If engineers sat down and said “material science tell us that this could work, let’s design a thermometer” then the design was engineered. If engineers sat down and said “something is broken here, or unexpected” and discovered resistive-conductive interactions with temperature, then it was initially discovered.

    Then, when you go to build such a thermometer later, yes, it is designed, by my glancing around at trees and birds is not, it is what happens to be there.

    But in between we have people sitting around and saying … “We need a proxy for temperature in arid regions of Jordan because right now I’ve got dick”

    and somebody else goes

    “Are there springs there?

    “Yes”

    “Well, use the spaleotherms! I don’t know if it will work, it’s never been tried, but you could do yadayadayada”

    … 6 months later …

    “Hey, I brought back samples form my site in jordon, not only natural springs but also stuff from the foundation of an irrigation structure”

    “OK, let’s apply the science we thought up last many. I’ll design a sampling probe while that guy over there designs the protocol for the mass spec, while she works on the statically model”

    … ti ti ti ti ti ….

    Then suddenly we have a temperature proxy working for arid regions of Jordan.

    What I’m saying is that this scenario, which actually happened, is not much different from discovering that some material changes property when it cools or heats, and making a thermometer from that.

    So, I concede your dichotomy but I mess with your context.

    Which gives me a thought. It would be neat to have a wall chart whose horizontal dimension is time into the past, logarithmic, and that lists vertically all (well, many) of the different techniques for measuring the age of stuff, and then draws a line for each showing the range to which it has been applied, with some kind of cross-lines for showing where different techniques are checked against each other. Anyone know of such?

    I’ve seen those, used those, made those, but a) they never looked very good and b) I know of nothing current AND detailed. It would be a great project. I tried to get my intro to arch students to interdependently invent something like this last year but they did not manage it. Something like this might work best as a high res graphic that could be scanned in a browser or used in a PDF file that you would never actually print out.

    Could be an open source projec.

    A gnatt chart may be a good format to use?

  14. Could be an open source project.

    Especially for maintaining currency. Which could be done with something like a wiki backended by a database. It still would take some effort to turn into a nice graphic for showing the results, up to some point. If I were Warren Buffett, I’d throw a bit of money at it.

  15. My point is that denialists simply don’t believe proxies because they are indirect and use information that does not have a real time analog, but that disbelief is based on a false dichotomy.

    So therefore Steve McIntyre isn’t a denialist. He’s actually very skeptical of the use of certain proxies because of statistical issues in resolving that indirectness. Not because they’re indirect. (And you won’t delete my link to his site because he’s not a denialist.) Right?

  16. “Some people don’t trust the accuracy of certain paleotemperature proxies” somehow becomes “therefore they just don’t understand that all measurements are inherently indirect” and “a tree ring is as accurate a thermometer as a platinum RTD.”

    Is that the main gist of this post?

  17. @Greg…

    You find logic astounding? Why am I not surprised? But stay tuned, we might actually get to the point where you admit that it’s possible to be skeptical about the IPCC, or even “Climate Change” without being a denialist.

    Next step: There isn’t really any such thing as a skeptical atheist, it’s a contradiction in terms. The correct term is denialist. The only appropriate religious position for a skeptic is agnosticism.

    (Besides, the “Holy Spirit” most certainly does exist. The key question is whether it exists as anything more than a mass delusion. But even a mass delusion can have real world consequences and affect the real world.

    (Hows that for astounding logic?)

  18. AK, so far there have been a lot of hypotheses put forward testing the idea that temperature change across decades is local variation, short term variation, not real, caused by clouds, etc.

    Every one falsified.

    So far there have been a lot of hypotheses about various religious claims put forward, like about healing, prayer, sticks turning into snakes, etc.

    Every one falsified.

    Next step: No longer giving even a small voice to the denialists because it is nothing but a joke at this point. Not censorship, mind you. Just not longer bothering to listen.

Leave a Reply to Iain Cancel reply

Your email address will not be published. Required fields are marked *