The first thing I want to say is that it is very complicated. The earth's climate is an extremely complicated system exhibiting nonlinear chaos and dominated by complex feedback mechanisms. So I guess the place to start is a general discussion of feedback.
All feedback can be classified as one of two types: positive or negative feedback. Positive feedback takes an input and amplifies it. Negative feedback takes an input and shrinks it. So in terms of climate change, a positive feedback would take an increase in temperature and make it larger. One example of a positive feedback in climate change is ice melt. Ice reflects most of the light that hits it, so sunlight hitting ice caps isn't converted into heat. When ice melts, less of the sunlight is reflected, so more of it is converted into heat, which should contribute to a further temperature increase. An example of a negative feedback (though not a very strong one) is the temperature dependence of the Stefan-Boltzmann law. The total power in radiation emitted by a hot object increases rapidly with temperature, so that a small amount of heating will increase the energy that leaves the earth through radiation, thus causing cooling (or, more precisely, counteracting some of the heating).
Some types of feedback are not understood. For example, cloud formation could be a very strong feedback, but we understand how they form poorly enough that we can't even say whether it would be a positive or negative feedback. Generally speaking, clouds (which are basically white when viewed from space) reflect more incoming sunlight than the ground below them, so that increased cloud cover will reduce the amount of sunlight that is converted to heat, leading to cooling. However, it's hard to say whether warming would result in increased or reduced cloud cover. Warming should increase the water vapor in the air, which should make it easier for clouds to form. But it's hard for clouds to form when the temperature is higher. With our poor understanding of cloud formation, it's hard to say which of these effects will dominate.
So on the whole, there's a complicated competition between positive and negative feedbacks, some of which are so poorly understood that we just can't model them. This isn't saying that climatologists are bad scientists. It's just that climatology is really, really hard. So it's no surprise that we have a very hard time finding models with predictive power in such a complicated system (more on that later).
Now we need to think about CO2. CO2 is actually a relatively weak greenhouse gas, by itself it can't do all that much to the earth's climate. However, it is a greenhouse gas, and it's silly to think that releasing it would not have a warming effect, though it might be very small. The trick is that in a feedback-dominate system, it's hard to say what the end result of even a very small effect will be. If positive feedbacks always strongly dominated under all conditions, then even the tiniest perturbation from CO2 would cause the earth to enter runaway heating, continuing to heat forever. Of course that's silly, because such a system would be so unstable that we would know it (or, rather, we wouldn't, because we couldn't be here), but it does serve as an illustration of what the real fear in climate change is. If positive feedbacks dominate over some range of conditions, then the earth will heat through that range. Conversely, if negative feedbacks dominate, then we'll have to push pretty hard (release a lot of CO2) for the earth to do anything very big. And of course reality could be anywhere along that spectrum. But the point is that we are releasing CO2, and that can only have a heating effect, but whether that effect is significant or not depends on lots of very complicated feedback mechanisms.
This brings us to modelling. I have some objections to the most politically charged model. In this model, global temperatures slowly increase for a while, then bend up and really go crazy, increasing very quickly (this has been called the "hockey stick" model because of the way the temperature suddenly curves up on a plot). I have two objections to this model, and three reasons for the first objection, one for the second:
- The model assumes no significant negative feedback (I think it includes some relatively minor ones, such as the Stefan-Boltzmann temperature dependence), which I think ludicrous because
- We would be in the case where positive feedbacks dominate, and some perturbation over the history of the earth would have driven us to ridiculously high or low temperatures already.
- There are other potential negative feedbacks, such as cloud formation (my instinct is to say that cloud formation should be a negative feedback, because you can counteract the increasing temperature by simply forming clouds at higher altitude, and you can't counteract the increased humidity), and ignoring these completely because we don't understand them instead of putting them in as error bars seems a pretty weaksauce way of handling things.
- It's very rare to find systems in nature that are positive feedback dominated over such a long range. They do exist, so this makes this reason for the objection very weak indeed, but things in nature almost always do end up in regions where negative feedbacks dominate, so I thought it worth making a note of this.
- The model has not been successful in making predictions. In fact, it's been predicting that we're about to turn the corner on the hockey stick for some two decades now, almost since its inception, and it hasn't happened yet. Its authority has come from the fact that it can be made to fit past data, but the truth is that if you add enough parameters you can make basically any model fit the past data. The test of a model such as this is in making predictions. If it can't successfully predict the future, then the model doesn't explain the phenomenon well enough to explain how it works, past, present, or future.
So I'm not willing to give much credence to this model or arguments built around this model. In fact, this system is so difficult and complex that I don't think that any model is particularly reliable yet: you're almost guessing when you choose a model (since you need to guess on some of the feedback terms), and if you're going to do little better than throwing darts at a board to choose your model, you may as well throw darts at a board for your temperature predictions. That's not to say that modelling is a complete waste of time (I am exaggerating a bit when I compare it to throwing darts at a board), as perhaps someday they'll find one that really does prove to be accurate, just that no model, to my knowledge, has shown itself better than so many others which make such vastly different predictions, so it seems senseless to say that we know very much from these models.
If we've yet to draw meaningful conclusions from modelling, what is left to us? Well, there is history and geology, anything that can tell us ancient temperatures. This gives us an idea. Looking back in time, we find that earth has often been basically stable (you might say "quasi-stable," meaning that it's not really stable, something eventually jarred it out of that mode, but it looks pretty stable for some time), but it's jumped between these quasi-stable modes very quickly. For example, we can look at ice ages. If you look at the Wikipedia article for "Quaternary glaciation" (the most recent ice age), you see all kind of graphs showing rapid changes between some relatively stable periods. More recently, there's the "little ice age" and "medieval warm period."
What does this say to me? It makes me think that the hockey-stick model is half right. You can explain these quasi-stable modes and mode hops in terms of positive and negative feedbacks: whenever negative feedbacks dominate, you're in a quasi-stable mode; whenever positive feedbacks dominate, you mode hop. Then history says to me that there are regimes in which positive feedbacks dominate, and others in which negative feedbacks dominate. The "hockey stick" could happen at a transition from the negative to the positive dominated region, and would correspond to the start of a mode hop. What the hockey stick misses is that geology and history say that the "hockey stick" levels out again, that you enter another negative feedback dominated region. It should be noted that climatologists are aware of this, and surely they limit the range over which they consider the "hockey stick" valid, but of course news reports and politicians aren't so careful about the limit of validity. Also, this kind of just confronts us with a different question: Will we mode hop, and if so, how large will the mode hop be? Major glaciations in the last ice age have led to variation as large as 8 C. But much smaller mode hops exist, on the order of a couple degrees C or smaller. The warmest temperatures I could find in the geological record in my short search (mostly just Wikipedia) were in the Devonian and Cambrian, 7 C above modern temperatures, though over shorter sections of those periods temperatures may have been even higher.
Me? I'll venture a guess. I think CO2 emission has precluded a mode hop to another glaciation, and may induce a smaller mode hop to higher temperature. I think that continued release of CO2 could result in a mean global temperature anomaly of 1-5 C, with 2-3 most likely (we're around .5 now, meaning that we're about .5 C higher than we were before we started emitting lots of CO2... I think 2-3 most likely because that seems to be on par with a lot of past mode hops). I do not think that the 10 C which is given as the doomsday scenario is reasonable, in part because that exceeds what I can find when we look back in history and geology. However, 5 C would have a significant effect (I think even 2-3 C would, but I can say 5 with a good deal of confidence), whether for good or ill on the whole, I cannot say. My initial point at the start, that the earth's climate is much too complicated for us to predict the future, holds for the effects of climate change as much as anything else (of course, there are some things we can say confidently could only be bad things for us, such as sea level rise).
This last point is important, too, and I want to hammer it home: I once attended a talk by my adviser's adviser (my "grandviser," if you will), who happens to be noted for his thoughts on this issue: Steve Chu. He put up evidence that warming is causing problems, but frankly, his evidence was awful, and I'm not particularly proud of him as a physicist for this one. His evidence, the only piece of evidence that he gave for climate change making things worse, was that costs of natural disasters had gone up. The top 10 most expensive natural disasters have all been in the last 50 years or so, he noted, and if you just look at data for natural disaster related insurance claims in the US, they've been steadily rising. The problem with this evidence is that he didn't do anything to control for increases in population or prosperity. If disasters cause twice as much damage in inflation-adjusted dollars, but there's twice as much property value around (also in inflation-adjusted dollars), then the disaster likely wasn't any more severe, it just had more stuff that it could destroy.
In thinking about this, there are a few good ways to control for this economic effect. One would be to track GDP and population changes in the areas affected by disasters. It may be difficult to amass all the data required to do this correctly, and it may be undermined by the fact that we've been getting better and better at engineering things to resist natural disasters, but it could help. A second method would be to compare earthquakes and weather-related disasters. Nobody believes that earthquakes are influenced by climate change, so increases in earthquake-related damages are probably a result of economic and demographic changes. This would allow a direct control for these factors. However, this method may be undermined by the fact that hurricanes and earthquakes don't always happen in the same places. If Florida has been growing in population and prosperity at a different rate from California, it may prevent direct comparison. Even so, some combination of these efforts may allow us to see what effect warming has on natural disasters.
Of course, I honestly don't think that hurricanes are the most significant question. If the mean global temperature rises by 3 C, two things may be more significant. First, ice that melts on land could lead to sea level rise, which would be a big deal for coastal cities. Now, if the sea levels were just a few feet higher and always had been, I don't think this would be a big deal: We'd simply have built our cities on the actual coast. However, there's certainly a transitional problem, and if there's significant sea level rise, it may be expensive for us to deal with.
Even more important, though, is arable land. In the long run, this is the main concern for climate change. We can find solutions for cities that would be slowly flooded, people can move inland or we can engineer good dike systems (though that's a solution we may need to be a little wary of... just look at New Orleans and Hurricane Katrina). It's expensive in transition, but in the long run, there's a solution. However, changes in temperatures lead to changes in wind patterns. This is another thing that I think is very complicated (which I think is the main reason Chu didn't present anything about this: he probably would have presented it if we had a good idea of how arable land will change), but changing wind patterns lead to changing rainfall. In the very long run, the total amount of arable land could change, as the changing wind and rain patterns could change fertile soil to desert, and vice versa. For all I know, the Sahara could become a great breadbasket, or the Amazon could become a desert (those seem far-fetched to me, but I have to admit just how deep my ignorance on this is). For all I know, on the whole, climate change could make the world a better place on the whole by increasing the total arable land, thus allowing us to grow more and better food for everyone. Or, on the other hand, it could mean that we have to work harder to produce the food we need. But I will say this: If arable land changes significantly, even if the total increases, it's certainly a transitional problem. After all, if arable land changes significantly, where it is will also change. It's hard for me to believe that the Sahara Desert will suddenly get lots of rain, but if it does, those kinds of changing global wind patterns won't just leave every breadbasket in the world untouched. It'll take us time to figure things out, and in the meantime, there will be drought. It's even been suggested that this is currently a factor in the drought in the Western United States (droughts like this, and even worse, have happened in the past, well before the industrial revolution, but it is possible that climate change has made this one worse than it might have otherwise been).
This has been a very long post. So, in summary:
-The climate is very complicated.
-Exactly what will happen is hard to say, exactly how significant it will be is hard to say, but warming from CO2 emissions seems like a pretty sure bet.
-Climate change may have either positive or negative effects in the most important, long-term sense, depending on complicated issues involving wind patterns and arable land, but if its effects are significant, they will certainly be negative in the short term.
This last point is important, too, and I want to hammer it home: I once attended a talk by my adviser's adviser (my "grandviser," if you will), who happens to be noted for his thoughts on this issue: Steve Chu. He put up evidence that warming is causing problems, but frankly, his evidence was awful, and I'm not particularly proud of him as a physicist for this one. His evidence, the only piece of evidence that he gave for climate change making things worse, was that costs of natural disasters had gone up. The top 10 most expensive natural disasters have all been in the last 50 years or so, he noted, and if you just look at data for natural disaster related insurance claims in the US, they've been steadily rising. The problem with this evidence is that he didn't do anything to control for increases in population or prosperity. If disasters cause twice as much damage in inflation-adjusted dollars, but there's twice as much property value around (also in inflation-adjusted dollars), then the disaster likely wasn't any more severe, it just had more stuff that it could destroy.
In thinking about this, there are a few good ways to control for this economic effect. One would be to track GDP and population changes in the areas affected by disasters. It may be difficult to amass all the data required to do this correctly, and it may be undermined by the fact that we've been getting better and better at engineering things to resist natural disasters, but it could help. A second method would be to compare earthquakes and weather-related disasters. Nobody believes that earthquakes are influenced by climate change, so increases in earthquake-related damages are probably a result of economic and demographic changes. This would allow a direct control for these factors. However, this method may be undermined by the fact that hurricanes and earthquakes don't always happen in the same places. If Florida has been growing in population and prosperity at a different rate from California, it may prevent direct comparison. Even so, some combination of these efforts may allow us to see what effect warming has on natural disasters.
Of course, I honestly don't think that hurricanes are the most significant question. If the mean global temperature rises by 3 C, two things may be more significant. First, ice that melts on land could lead to sea level rise, which would be a big deal for coastal cities. Now, if the sea levels were just a few feet higher and always had been, I don't think this would be a big deal: We'd simply have built our cities on the actual coast. However, there's certainly a transitional problem, and if there's significant sea level rise, it may be expensive for us to deal with.
Even more important, though, is arable land. In the long run, this is the main concern for climate change. We can find solutions for cities that would be slowly flooded, people can move inland or we can engineer good dike systems (though that's a solution we may need to be a little wary of... just look at New Orleans and Hurricane Katrina). It's expensive in transition, but in the long run, there's a solution. However, changes in temperatures lead to changes in wind patterns. This is another thing that I think is very complicated (which I think is the main reason Chu didn't present anything about this: he probably would have presented it if we had a good idea of how arable land will change), but changing wind patterns lead to changing rainfall. In the very long run, the total amount of arable land could change, as the changing wind and rain patterns could change fertile soil to desert, and vice versa. For all I know, the Sahara could become a great breadbasket, or the Amazon could become a desert (those seem far-fetched to me, but I have to admit just how deep my ignorance on this is). For all I know, on the whole, climate change could make the world a better place on the whole by increasing the total arable land, thus allowing us to grow more and better food for everyone. Or, on the other hand, it could mean that we have to work harder to produce the food we need. But I will say this: If arable land changes significantly, even if the total increases, it's certainly a transitional problem. After all, if arable land changes significantly, where it is will also change. It's hard for me to believe that the Sahara Desert will suddenly get lots of rain, but if it does, those kinds of changing global wind patterns won't just leave every breadbasket in the world untouched. It'll take us time to figure things out, and in the meantime, there will be drought. It's even been suggested that this is currently a factor in the drought in the Western United States (droughts like this, and even worse, have happened in the past, well before the industrial revolution, but it is possible that climate change has made this one worse than it might have otherwise been).
This has been a very long post. So, in summary:
-The climate is very complicated.
-Exactly what will happen is hard to say, exactly how significant it will be is hard to say, but warming from CO2 emissions seems like a pretty sure bet.
-Climate change may have either positive or negative effects in the most important, long-term sense, depending on complicated issues involving wind patterns and arable land, but if its effects are significant, they will certainly be negative in the short term.
No comments:
Post a Comment