Friday, May 17, 2013

Testing the Science of Global Warming

Once upon a time, I studied physics and chemistry (chemistry was my original major in college). As aspiring scientists, we did more than just read facts written by experienced and learned scientists, we made our own observations, and formed hypothesis, designed and conducted experiments, recorded our results and concluded whether our hypothesis was true or not.

There was another thing that we did in our lab reports. We recorded areas where our experiments may have had errors. Some of these were routine.  Things like test tubes might not have been sterile, variances in room temperature and so forth.  Some were very specific to what we were doing.  Not being able to reuse a sample for our tests was always frustrating. And if you looked hard enough on any experiment, you could find areas where measurements or results could have been influenced by an outside factor. If you couldn't find these, then you weren't looking hard enough. And we lost points if we didn't list these in our report.

Here is how this works. If we designed an experiment to study the average speed of marathon runners.  And we decide to do this, starting two thirds of the way through the race.  Can we get data to do this?

Lets see, we can pick a runner, and retrieve his check in time at the various check points along the course.  This will give us the historical time and distance references.  Then we can graph the results and show a nice flat line with little change in speed between the checkpoints.

Now the race isn't over yet.  We have miles left to go. And the University has given us another tool other than a calculator and a stop watch. We have a radar gun so that we can get the exact speed of a runner at a given moment in time.  This is great for us because we can get more samples of speed instead of having to rely on check point time records.

So we start to setup at a variety of places.  They just happen to be next to water stations along the route.  And we find an alarming trend. All of the runners are slowing down in all of our new reading.  Even the fast runners are slowing and it looks like that if this trend continues all of the runners will be finishing the marathon at a walk.

We even graph our data and show it to people with a nice downward hockey stick shape.  We have a slowing trend across the entire set of competitors.

But is there something wrong with this method?  Aren't getting more samples of speed making our experiment more accurate?

Obviously two things are wrong.  First, we have changed our method of reading speed. Changing this method makes our readings uncomparable and my professors would have thrown out my results and made be go back and do the test again.  Second, there is a bit of lazyness in everyone, and setting up next to the water stations may have seemed like a smart move for the speed taker, but it adds an environmental factor that is negatively influencing runners speed.  That is slowing down to get water, or being slowed by others that are getting water.

I was curious.  So I looked about and found that to get an atmospheric C02 level for times in the past, we go to ice core samples.  We then crush them and use gas chromatography to find the composition of the atmosphere.  Okay, stay with me.

I then looked up how to get CO2 readings from the atmosphere today.  Guess what.  They don't involve getting ice cores.  They are taking direct measurements.  Yes, this gives us more samples, and can give us better results of the CO2 concentration, but the comparative measures should not be related. (stopwatch and radar gun comparison)

Why does this matter?  Aren't we still talking about how much CO2 is in the atmosphere?  Shouldn't testing the air trapped in the ice give you a good enough reading?  Nope, remember the Marathon checkpoints and the radar gun measurements, where we put the guys with the radar right next to the water.  Our ice core samples come from areas of the world with ice.  (I know that sounds obvious.)  So it stands to reason that the air trapped in them comes from those areas.

But where are we getting our current temperatures?  Why from all over the world, replies NOAA, NASA, and other organizations.  And this sounds good until you realize that "all over the world" includes the observatory on Mauna Loa, a volcano in Hawaii.  (No ice, and volcanoes have a habit of emitting lots of CO2.)

This is why we examine our experiments and see what could skew our results.  Changing the method of collecting measurements, and the locations of the measurements can skew the results. It makes the data hard to understand. And it gives us reasons to doubt the results and conclusions based on looking at this mixed method data.

When we are looking at something important like our climate change, it is necessary to stick to like data, and like methods and not mix and match apples and oranges. The failure to do this casts dirt on the claims of the scientists that present them. Its why I keep giving the global warming predictions a failing grade.

In fairness, climate is something that we should seriously study. And it matters if there is a warming trend. But before we start making claims that we can prove something, we need to get our science in order.


  1. First of all, it's unfair to compare chemistry with climate science, especially when you're talking about paleoclimate. It would be like comparing cell biology with paleontology, or particle physics with cosmotology. Sciences that attempt to understand the past are, by nature, limited. However, it doesn't mean that the data found by these scientists is worthless.

    Next, it must be said that ice cores aren't the only way to get historical readings of temperature or CO2, although out of all the methods it's the most accurate. Tree rings, sediment cores, magnesium/calcium ratios in marine sedimentary layers, number of stomata in fossilized leaves, and fossilized pollen studies among others. While none of these is perfect (especially compared to modern measurement methods) having these multiple lines of evidence agree gives us greater confidence to accept the conclusions they lead to.

    Finally, let's address your analogy. What it seems like you're trying to say is this: Let's assume that climate temperature has always had fluctuations (just like runners speeds have always had fluctuations). The only reason we're measuring higher temperatures, even though we have more accurate measurements, is that we happen to be at a high point in the normal temperature fluctuations (just like we're measuring runners to be slower only because we're happening to measure runners when they're at their slowest moments in their normal speed fluctuations).

    As far as the climate goes, you're right--the climate has gone through fluctuations of temperature. Of course, we only know this because of studying ice cores. So you can either dismiss ice core data of temperature fluctuations as being too inaccurate to base modern theories on, or you can use ice core data to support your idea that climate temperature has always had fluctuations and we're just looking at a high point of those fluctuations.

    On what are you basing the assumption that the climate has gone through fluctuations?

    1. First, it is perfectly fair to compair good scienftific method to bad scientific method.

      Second, It doesn't matter that paleoclimetologists use tree rings too. The point is they don't use tree rings and ice core samples to measure CO2 levels today. When you change the method of your measurement you introduce a huge source of error and inacuaracy into the experiment.

      Finally I am not saying we are not having warming. I am saying that the science that is being presented to me on the matter starts with a long period of data that is relatively flat coming from ice cores, and the a huge spike that is comming from direct surface mesurements. I know if this was happening in my lab, I would toss the data and find another way to get measurements that can be taken reliably accross the timeframe.

      I don't have to be cornered int these choices. I don't have to dismiss one or the other. I would, however like to see us show a graph of only ice core samples. Is there a reason we can no longer get theses samples? Did the earth warm so much that all of the ice is gone?

    2. My point about comparing experimental sciences with historical sciences is to point out that while the experimental sciences are able to get very accurate measurements (generally because they're dealing with fewer variables and can do repeatable experiments), it doesn't mean that the conclusions of those in the historical sciences are useless. The conclusions are based on what data we have and generally fit into the larger theories of their respective fields.

      I don't see why you're concluding that a difference in measurement tools leads to difference in data. If I use a mercury blub thermometer and a thermoresistor electric thermometer, they'll still show the same temperature as long as they're calibrated properly. We do have ice core samples that overlap recorded temperature and they match up, so why should we think that ice core temperatures from times before modern recording would be so completely incorrect as to be worthless?

      From what I could find, the most recent ice core measurements are from about 1980. The reason we don't have any more recent measurements from ice cores is because the snow takes time to compact into ice trapping the air bubbles to measure. In the most recent ice core samples, we see the same increase of temperature (starting soon after the industrial revolution) that we've seen in our recent measurements. By merging together all the data (which overlap and the different measurement techniques fit where they overlap) we can produce a record of temperature that's as accurate as we can make it.

      And it turns out that the measurement does have a dramatic increase near the end. The ice core show measurements show the start of this rise, and modern measurements, which overlap with the ice core measurements, show the continuation of that increase in temperature. But it's not an unexplainable increase--certainly not to the point that the measurements should be brought into question, as there seems to be a likely correlation with the increase of CO2 (an known greenhouse gas) due to human activity. Is that the only possible solution? Possibly not, but it's the one that best fits the data so far.

  2. When I was measuring samples in the lab, one of the things I did was to use the same scale for all of my samples in the same experament. I didn't do this because I was superstious. Or out of convenience. I did this to eliminate a possible source of errors.

    When it got down to weighing substances by milligrams, the scales didn't always agree about the weight of the same sample. Its good science to take samples and record data with as few external variation as possible. Thus is we are tracking something like air temperature, we should first pick a location that doesn't have much wiind and sun so as to not contaminate out data.

    Then we should take all our readings from this location so that we don't introduce new changes. Graphing this data becomes useful?

    But if we start out taking CO2 level readings in the frozen wilderness of antarctica. And the move our observations to a smokestack or volcanoe, we have corupted our data.

    Even if I were to agree that paleoclimatology was a more difficult form of science than chemistry, that is no excuse for abandoning the proper way to gather data.