Daily Archives: May 21, 2013

Worst Person In The World: Emma Way of the UK

Emma Way (formerly @EmmaWay20, but she has deleted her account) was driving down the road and turned into oncoming bike riders who were in a race. She hit one of them, knocking him into the woods. He’s OK.

She tweeted abut how proud she is of having done that, making the claim that this was OK because bikers don’t have the right of way and don’t pay road taxes.

The police picked up on her tweet and tweeted her a suggestion that she report the incident and DM them back. Others picked up on her tweet and scanned her social networking streams and found all sorts of other horrid things. Apparently she was into doing selfies of herself (obviously) tailgating other vehicles and driving at excessive speeds.

Emma Way is a horrible, terrible, awful person. Is she redeemable? I don’t know. Maybe you found this blog post because you are googling her and she’s a job applicant. If you hire her and find out that she’s reformed, post a comment and let everyone know!

(And she will be looking for a job. Her current employer appears to be poised to fire her.)

The UK does not have a road tax.

Understanding Storms and Global Warming: A Quaint Parable

A quaint New England rocky creek

Imagine standing next to Parable Creek, an imaginary rocky brook in New England. The water is rushing past you from left to right, around the rocks that emerge tall above the surface of the stream, mounding over the top of those that are lower down. The deepest parts of the steam are relatively flat but show ripples that belie the presence of other rocks and sunken branches that are well below the water line.

While you are observing a young boy of about 11 years old comes along, carrying his fishing pole. “Hey mister, how’s it going?” he says, as he steps into the stream. “I’m going fishing over there,” he says pointing in the direction of a mill pond a mile or so away. As he crosses the stream you notice that whenever he puts is foot down, some of the water mounds up on the upstream side as it rushes by him. He continues across the stream and climbs the opposite bank, running off to his destination. You wish him good luck with his fishing and return to your observations.

You can see large eddies here and there that seem to persist though they may change shape or grow or shrink a little. Smaller eddies, mini vortexes, form in certain parts of the stream, and rush down slope only to disappear as the water crashed into an obstruction. Every here and there there is a splash caused by the rushing water hitting a rock or branch just the right way.

Now, imagine that you are a compulsive data collecting scientist standing next to the rocky brook with nothing else to do for a while. So you start measuring things. Every where you see a mound of water built up in the current alongside a rock, that is a bit of kinetic energy (water moving) converted momentarily in to potential energy (water rising against gravity). So you estimate the number of mounds and their collective mass. This is a measurement of one form of energy in the stream.

You make a prediction. If the amount of water coming down this stream increases for a while, the total energy of the stream will increase, and this will be visible as an increase in total potential energy in the mounds you’ve been measuring.

Coincidentally it has been raining heavily upstream and just as you have formulated your hypothesis you see the water rising. Aha! A chance to test your prediction. At first, your hypothesis seems supported. As the water rises, the relative height of the mounds increases, and some new mounds form. You take some quick measurements, by eye, and note that the total potential energy stored in water mounds has increased, presumably as an effect of more overall energy in the stream. You gain confidence in your theory and congratulate yourself on your brilliance.

But then, as the water level continues to rise something different happens. More and more of the stream is now above the obstructing rocks. Therefore, there is less conversion of kinetic to potential energy. Most of the mounds disappear and the overall surface of the stream is much smoother. You take a new set of measurements and estimate that the total potential energy stored in water formed into mounds is an order of magnitude LOWER than your original measurement. Apparently, you think, something is wrong with this stream.

Just then a troop of Brownies comes along. The little girls want to cross the stream to take a short cut to their picnic grounds. They ask, “Hey, Mister, do you think it is safe to cross this stream?”

You had a nice theory linking total energy and a specific observation, which seemed to be confirmed by some of your research. The total energy of stream flow is linked to the total mound-i-ness of the stream’s surface. Now, the stream’s surface is smoother than it was before. Therefore, the total energy of the stream is at the low end of its known variation. A while back you saw a small boy cross the stream with no problem. Clearly, it is safe to cross now.

So, you say, “Actually, I’m sure it is quite safe. Go ahead and cross, and have a nice day!”

The brownies jump happily into the stream and start wade through the water. Half way across the stream, one by one but over just a few seconds of time of time, they are carried away by the water and drown.

“Hmmmm,” you think. “Maybe I had that wrong.”

Rivers Of Air

Air flowing over the surface of the land is a bit like water running down a stream or river. The air interacts with the ground (especially things like mountains). There are different layers, mounds, streams, and eddies of air that interact with each other. The overall form of movement is shaped by the spin of the earth, the tendency of warm air to form in certain areas (i.e., near the equator, or over water during winter and over land during summer, etc.) which causes the air to pile up and spill into nearby eddies. There are all sorts of ways in which batches of air interact, and when you thrown in differential amounts of moisture in different air masses, and things like night vs. day, and so on, you get the surface of Parable Creek. Metaphorically. In real life, we call the The Weather.

When the total energy in the system of air movement changes the way those crazy zany air masses move and what sorts of weather form can also change. For example, there is in total more energy on the hemisphere (north vs south) that is sticking its face towards the sun. It seems that one result of this is that the hemisphere with more energy (the summer end of the earth, as it were) has hurricanes, severe thunder storms, tornadoes, and so on while the hemisphere with less energy has less of that stuff.

However, a tornado is like a small eddy in the stream, and a hurricane like a large eddy, and a line of thunderstorms like the outer edge of one the mounds and the rainstorms are like the splashes at the edge of the log and so on and so forth. As Parable Creek’s level rises, exactly which phenomena are predominant changes, even as the total ability of the stream to wash away Brownies increases to the level where it can also wash away Girl Scouts and eventually Brawny Construction Workers and Bikers. Having said that, while a rocky stream converts to a large and deep river by adding a LOT of water, which may have a smooth surface despite the total energy of the river being orders of magnitude greater than Parable Creek’s energy, the system of air movement is not likely to become smoother owing to various limitations in the system.

Too Much Variability

You can’t measure the energy in the stream by only looking at one of the many phenomena that are the manifestations of that energy. In order to understand the relationship between global warming and storminess, it is minimally necessary to measure all of the storminess and find some way to combine it.

I remember when I first moved to Minnesota. That summer we had numerous straight line wind events of the sort never seen before. Maplewood, a community near where I lived famous for it’s tree lined streets lost almost all of its trees in one storm. That same storm also took out most of the stock of most of the new car companies in that town, famous for its numerous car lots. The cars were pitted with hail stones. Every single home for about three miles along a street right near where I lived had it’s vinyl or aluminum siding drilled with hundreds of holes and dents from large hail stones being driven by a 60–100 mile per hour wind. It was one of the worst weather years in Minnesota, with insurance companies practically going bankrupt.

There were only a few tornadoes in the area that year.

The next year there were hardly any straight line wind storms of the magnitude just described. But that is the year of the Saint Peter tornado. It was one of the largest tornado events ever; It was a twister that lifted and dropped a couple of times, so ‘nato-pedants divide it into multiple events, but that’s absurd. It was an F3 and F4 event, and it tracked for 67 miles and was up to one and a half mile wide.

There were a lot of tornadoes that year.

The atmosphere over central and southern Minnesota had a lot of energy those two years, for whatever reason. If we want to understand the total energy, and its effects on life and property, we would be doing a disservice to our pursuit of understanding if we failed to consider both straight line winds and tornadoes together (though obviously also separately).

The Big Picture

Weather comes in bands. The biggest and most obvious band is the Intertropical Convergence Zone, a band of thunderstorms that rings the entire planet and is pretty much always active. Another band is the arid band that rings the earth; actually there are two of them, one in the Northern Hemisphere and one in the Southern Hemisphere. Almost every major desert on the earth is in one of those bands. In fact, any desert that is not in one of those bands has to explain itself, and the excuse is usually a mountain rain shadow. Conversely, any wettish areas in those bands also have some ‘splainin to do. The southeastern US is in the Northern Hemisphere’s arid band, but the Gulf of Mexico keeps that region pretty wet much of the year.

Severe weather is also patterned in these bands, to some extent. Hurricanes form in the bands just north or south of the Intertropical Convergence Zone. Tornadoes tend to be confined to subtropical and southern temperate bands away from the equator. In a sense, one could say that most tornadoes that are not spinoffs from hurricanes occur in a certain band either north or south of the equator, and if we are going to count tornado activity, measure its total intensity, etc., we should be looking more globally at those zones, not just parts of those zones.

This of course applies to national borders as well. Tornadoes occur in the US but also in Canada, but the most easily available tornado data for North America is always presented as US tornadoes. Also, years are tricky. Events that span Jan 1st are hard to track if we count things by calendar years.

Some have been harping about the “tornado drought of 2012” as evidence that there is not an increase in tornadoes owing to global warming. Well, there are very few US tornadoes in January, but the January with the most tornadoes ever (in our records) was January 2012. Also, Canada had a lot of tornadoes in 2012. Has anyone looked to see what the combined US and Canadian count would be? And, how do you count a Canadian Tornado? The very fact that a tornado forms 1000 miles north from where most occur has something to do with the nature and distribution of atmospheric energy across the plant’s surface. I’m not making a specific claim about the distribution of tornadoes across time and space. I am saying, rather, that counting tornadoes within an arbitrary boundary in space (or time) can be misleading.

Then, there is the problem we have with all of these storm types, especially tornadoes and hurricanes, of how to actually measure them. Even using standard severity scales, tornadoes can be very different from each other in ways that are not counted in the usual statistics. An F3 tornado that is extra wide and stays on the ground for 100 miles involved significantly more energy than an F4 that formed momentarily and disappeared. Indeed, the different kinds of tornadoes (funnel vs. wedge, for example) really may be very different (but closely related) weather phenomena that should be examined separately at the very same time we combine vastly different storm types to measure and understand at a larger, global scale.

Tornadoes are not a good canary, in the canary in a coal mine sense. But they are obviously important. When we see people stating clearly and plainly that we need not be concerned about the frequency of tornadoes increasing with global warming, we should ask why they are saying that. We should be concerned with increasing storminess … there is almost no way that is not going to happen, and likely, it already has. If tornadoes are part of that increase storminess, we may want to get smart about it fast. For instance, we might want to take seriously the problem of schools and workplaces, where people tend to concentrate, having actual storm shelters instead of just hallways that some administrators says is a storm shelter, for protection when a big tornado comes along. Don’t you think?

See also this post which more directly addresses the question of tornadoes and global warming.

Photo Credit: Hamed Saber via Compfight cc

Are there more tornadoes because of global warming?

There are good reasons to believe that global warming leads to more storminess, but the exact nature of that transition is unclear and hard to measure. Part of the reason for this difficulty is that a given type of storm may become more likely under certain conditions caused by climate change, while a different kind of storm may become less likely, with the “storminess” overall increasing but doing so indifferent ways across time. Also, the most severe, and thus possibly the most important, weather events are infrequent so it is difficult to see changes over time with any statistical confidence. I address many of these issues here and here.

Looking at the raw data, it is clear that there are “more tornadoes” over time in the US. Have a look at this graph:

Annual number of tornadoes for the period 1916-1995; the dashed line connecting solid circles shows the raw data, the red heavy solid line is the result of smoothing. Also shown in the green light solid line is the number of tornado days (i.e., days with one or more tornadoes) per year.
Annual number of tornadoes for the period 1916-1995; the dashed line connecting solid circles shows the raw data, the red heavy solid line is the result of smoothing. Also shown in the green light solid line is the number of tornado days (i.e., days with one or more tornadoes) per year.

At first glance, his graph makes it look like there are a lot more tornadoes, but there is a strong effect of observer error; earlier tornadoes were simply missed much of the time, so the big increase you see here, while it may reflect an underlying increase in number of tornadoes, is not reliable and cant’ be taken as evidence. However the later years shown here, from 1950-something to the 1990s, seems to show an increase that could be taken as meaningfull

However, when people speak of tornadoes they often show this graph as evidence that there are not more of them over time:

Looks like the number of tornadoes does not go up over time.
Looks like the number of tornadoes does not go up over time.

Looking only at this graph it looks like the number of tornadoes per year in the US is pretty variable but not increasing, as one would expect if global warming was causing more of them.

There is a problem with this graph, however. Actually, a couple of problems (other than those pointed out here). The main problem is that the most frequent tornadoes are left off this graph. If we look at F0 grade tornadoes, not included here, we see that they have actually increased in frequency over time. If we include ALL tornadoes, and not just the kinds that don’t seem to increase in frequency over time, we get this graph:

Huh.  Maybe the number of tornadoes DOES increase over time!
Huh. Maybe the number of tornadoes DOES increase over time!

Compare the scales of the last two graphs. It turns out that the number of tornadoes at the smaller end of the scale goes up quite a bit. It might be hard to see. The upper graph goes up to 900, the lower graph goes up to 1900. So, if we add all the data instead of just select data, we get many hundreds more tornadoes per year.

The proportion of tornadoes that are F0 increases over time as shown here:


… and the overall distribution of tornadoes by strength changes over time as shown in this very cool graph:

It isn't just the F0 tornadoes changing over time.  The overall pattern of tornadoes shifts with time.
It isn’t just the F0 tornadoes changing over time. The overall pattern of tornadoes shifts with time.

As I point out here, one of the contributing factors to variation over time in tornado frequency is the fact that we have somewhat arbitrary boundaries in which we measure them. For instance, the US-Canada border provides an arbitrary line across our data set. By not counting all North American tornadoes the same way, we may be adding unnecessary variability to the data. To demonstrate this, have a look at this graph showing tornado frequency per year in France and Germany, two countries that are right next to each other:

Frequency of tornadoes in France and Germany ... seems to be uncorrelated.
Frequency of tornadoes in France and Germany … seems to be uncorrelated.

This shows a few things. For one thing, they don’t have too many tornadoes in that part of the world. For another thing, there is an increase in overall frequency over time, and this is not because of lack of reporting. The reporting problem in the US is partly because the western and central states were relatively empty in the old days, and also more technology was available for spotting tornadoes later. But the European and US data have the same shape over a similar time span, but France and Germany do not have the missing observations owing to vast unoccupied (sort of) territories.

But the main thing I want to demonstrate with this graph is the fact that dividing a largish area of land up into arbitrary units can cause your data go go all flooey. Increased variability in data owing to partitioning is a well known phenomenon and this is what it looks like.

Another part of the problem is that the largest storms, which may be the most important ones, have a great deal of variation in their occurrence. Compare any of the graphs above of all tornadoes or all excluding the F0 tornadoes of this graph of just the largest storms:

Pay attention to the vertical scale, but note that there is a lot of variation over time in these large events.  This kind of data almost has too much variability to track change over time meaningfully
Pay attention to the vertical scale, but note that there is a lot of variation over time in these large events. This kind of data almost has too much variability to track change over time meaningfully

Not only is there a lot of variation in numbers of tornadoes at the larger end of the scale, but I suspect there is a lot of variability among the tornadoes in each class in terms of overall energy represented. An F4 tornado that lasts five minutes compared to an F4 tornado that lasts 20 minutes are hugely different, but this is not reflected in this sort of data.

Here is a graph showing the amount of storm damagein adjusted dollars over time in the US (pink) with average temperature (blue). Clearly, the total amount of damage goes up, and probably for a number of reasons including there being more stuff to damage, but also, likely overall increases in storminess including hurricanes, tornadoes, severe thunderstorms, etc.

More storm damage over time
More storm damage over time

Here is another graph that shows something similar:

Increasing bad stuff over time.
Increasing bad stuff over time.

There are many who do not want to link increases in severe weather to global warming. They are probably wrong. Global warming seems to increase severe weather overall. The best way to deny this is to cherry pick the data by ignoring variability across space, leaving out entire categories of storms, or focusing on just some kinds of storms. I suspect the size and severity of tornadoes at the larger end is increasing now, but did not start increasing until recently; time will tell if this is right. But overall tornadoes are so variable across time and space that they are not a reliable canary, as it were. But overall storminess seems to be on the increase, in accordance with expectations from the basis physics of climate, under warming conditions.

Photo Credit: Vvillamon via Compfight cc

Does parasite load really matter?

In behavioral biology there is a fair amount of attention to individual quality, which may be determined by genes or parasite load or energy balance, or some interaction among these (and other) factors. Individual quality is honestly indicated by some trait or behavior; a large bright thing hanging of your head, a long bout of complex and energetic dancing, or a very loud complicated song, may be impossible to achieve in an individual with insufficient energy or some sort of disease. Therefore, other individuals looking to choose a mate can observe the traits or behaviors and do what the old guy in the cave said: “Choose wisely.”

Here is one of the nicest demonstrations of the relationship between parasite load and reproduction that I’ve seen in a while. And, as is so often the case, we gain valuable knowledge by closely observing great tits.

Photo Credit: OneTrack via Compfight cc