Oscillation
In the comments on the previous essay, dougo gave me the perfect graph:
http://en.wikipedia.org/wiki/File:Mauna_Loa_Carbon_Dioxide-en.svg
Go take a look at it. Now completely ignore the labels on the X and Y axis and the numbers and the title. And look at it again.
What it shows are peaks and troughs, and a line of averaging through them that shows how the entire thing is trending upwards. Look a little closer. Some of the peaks and troughs are larger than others from year to year, and some are smaller. Some are sharper, and some are wider.
But we know the whole thing is trending upwards because we take an average over a length of time and put a dot in the middle for that time and use it to smooth the graph and make it look nice. That's just one way we could make that determination, however. We could do what's called 'peak to peak' comparisons or 'trough to trough' comparisons and both of those comparisons might tell us something subtly different. (They don't happen to for this graph, but it's possible to have a graph where the peaks get higher faster than the troughs and so different things are happening to high values than are happening to the average than are happening to the low values.)
However, if I was a bastard I could choose my dates such that the peak of one year would be exactly the same as the trough of another year, and so I could say (lying with statistics) that over the course of 9.5 years this value has not risen at all. It would be truthful, but it certainly would not show the whole picture.
Now, for the time we have been measuring, there has been an approximate 30-year cycle going on. It is related to the Pacific Decadal Oscillation, to the 11 year sunspot pattern, to larger patterns in sunspot minima and maxima, and several other things that we are just now learning about.
We know from newspapers around the globe that the years 1905-1910 were bitterly cold. We have reports of odd freezing deaths, frosts, deadly winters. And from there it gets hotter until 1940-1945 - with a very important frigidity break from the volcanic eruptions in 1931-1934 spewing sulfur dioxide etc into the air. From there it gets colder until in 1970 all the newspapers are talking about global cooling and predicting the next ice age. (I'll mention that due to another related cycle, we had a heck of a lot of hurricanes/typhoons around this time.) And then it gets hotter and hotter until 1998, or 2001, or 2003.
And we have very little data from those early times so we have to guess at how cold it really was in general by looking at secondary records like crop harvests, crop dates, vegetable strata data, the number of people freezing to death in major cities, and so forth. But as time goes on our numbers do get better.
As our numbers get better we have a better understanding of the comparisons and the averaging. I need a good graph to show it but when I talk about a 'yearly temperature' keep in mind that we are averaging zillions of different pieces of temperature data, interfacing them with those fudge factors and guesses, and trying to come out for a single value, a big dot in the center that we can plot on a graph.
And what the line really looks like is more like that little inset in dougo's graph, where you might have an 'average global temperature' for each day in the year and you can watch it go up and down some as summers and winters hit and weather patterns sweep across and make some places hotter and some places cooler. The real line is very jagged, up and down. But as we average it more and more, and make the average wider, we can see the trend. That's how we can see that the average in dougo's graph is steadily steadily rising.
But we have the problem of outliers. Two big outliers that we know about, that we have pretty good data on, are 1975 and 1998. 1975 was a bitterly cold year. It had at least two instances where 'blasts of icy arctic air' swept cold fronts down North America and stayed there for a long time - I remember being out of school for a whole week in the Midwest and it was much worse north of me. And similar things happened in other countries. The years around it were plenty cold, but not nearly so brutal. 1975 distorts the graph in its area. 1968-1974 and 1976-77 are all kind of similar in temperature but 1975 pulls the curve down and how you decide to average it in (one year, three year, five years) tells you where you think the oscillation was at its bottom - almost definitely before 1975.
In the same way, 1998 was a devastatingly hot year. And because of its heat, the Global Warming thing really turned on, propaganda-wise, because, man, you just could not deny that this year was notably hotter than the years before. But the years after that were cooler than 1998, and relatively steady, not rising nor falling except within statistical noise. Again, 1998 pulls our average-graph into a curve and where the top of that curve is depends on how you do the averaging. And after 2003, it's fairly clear that we are cooling. How dramatically depends on your style of measurement and your fudge factors of choice, but we're certainly not getting hotter and we've never gotten back up to 1998.
So now go back to dougo's graph. Here's the problem. In our recorded history of semi-adequate actual measurement with devices that give us numbers, we only have two peaks and two troughs.
And it's just not enough to say. It's not enough to make predictions over the long haul. We were definitely warming from 1975-1998, hell yes were we ever. But we're cooling now, as best we can tell.
And so the next problem is that people with agendas are not measuring peak-to-peak or trough-to-trough. They say, "We have dramatically warmed from 1905 to 1998." And while they do have a reason for starting at 1905 - it's one of the first years we have a lot of numbers for - they are measuring from a trough to a peak. Of course that shows warming. Stopping at 1998 is bastard-like, though. Everyone who's anyone knows that was a scary-hot year and you're just using that to make your numbers look good. Go to the current year if you want to do that. It's 2011. You have numbers until 2010. They're in your spreadsheet. I see them right there.
Our numbers can tell us a lot of things. They can tell us very strongly where we should be looking and when if we want to learn more - we need to be studying those icecaps and watching them come back, or not. But until we get more peaks and more troughs well-measured, we can't tell if the whole cycle is trending higher or lower. We just don't have enough data points to average that way.
(By the by, I don't know what the hottest year was in the 1940 era and how it compared to 1998; nor the coldest year in the 1905-1910 era and how it compared to 1975.)
http://en.wikipedia.org/wiki/File:Mauna_Loa_Carbon_Dioxide-en.svg
Go take a look at it. Now completely ignore the labels on the X and Y axis and the numbers and the title. And look at it again.
What it shows are peaks and troughs, and a line of averaging through them that shows how the entire thing is trending upwards. Look a little closer. Some of the peaks and troughs are larger than others from year to year, and some are smaller. Some are sharper, and some are wider.
But we know the whole thing is trending upwards because we take an average over a length of time and put a dot in the middle for that time and use it to smooth the graph and make it look nice. That's just one way we could make that determination, however. We could do what's called 'peak to peak' comparisons or 'trough to trough' comparisons and both of those comparisons might tell us something subtly different. (They don't happen to for this graph, but it's possible to have a graph where the peaks get higher faster than the troughs and so different things are happening to high values than are happening to the average than are happening to the low values.)
However, if I was a bastard I could choose my dates such that the peak of one year would be exactly the same as the trough of another year, and so I could say (lying with statistics) that over the course of 9.5 years this value has not risen at all. It would be truthful, but it certainly would not show the whole picture.
Now, for the time we have been measuring, there has been an approximate 30-year cycle going on. It is related to the Pacific Decadal Oscillation, to the 11 year sunspot pattern, to larger patterns in sunspot minima and maxima, and several other things that we are just now learning about.
We know from newspapers around the globe that the years 1905-1910 were bitterly cold. We have reports of odd freezing deaths, frosts, deadly winters. And from there it gets hotter until 1940-1945 - with a very important frigidity break from the volcanic eruptions in 1931-1934 spewing sulfur dioxide etc into the air. From there it gets colder until in 1970 all the newspapers are talking about global cooling and predicting the next ice age. (I'll mention that due to another related cycle, we had a heck of a lot of hurricanes/typhoons around this time.) And then it gets hotter and hotter until 1998, or 2001, or 2003.
And we have very little data from those early times so we have to guess at how cold it really was in general by looking at secondary records like crop harvests, crop dates, vegetable strata data, the number of people freezing to death in major cities, and so forth. But as time goes on our numbers do get better.
As our numbers get better we have a better understanding of the comparisons and the averaging. I need a good graph to show it but when I talk about a 'yearly temperature' keep in mind that we are averaging zillions of different pieces of temperature data, interfacing them with those fudge factors and guesses, and trying to come out for a single value, a big dot in the center that we can plot on a graph.
And what the line really looks like is more like that little inset in dougo's graph, where you might have an 'average global temperature' for each day in the year and you can watch it go up and down some as summers and winters hit and weather patterns sweep across and make some places hotter and some places cooler. The real line is very jagged, up and down. But as we average it more and more, and make the average wider, we can see the trend. That's how we can see that the average in dougo's graph is steadily steadily rising.
But we have the problem of outliers. Two big outliers that we know about, that we have pretty good data on, are 1975 and 1998. 1975 was a bitterly cold year. It had at least two instances where 'blasts of icy arctic air' swept cold fronts down North America and stayed there for a long time - I remember being out of school for a whole week in the Midwest and it was much worse north of me. And similar things happened in other countries. The years around it were plenty cold, but not nearly so brutal. 1975 distorts the graph in its area. 1968-1974 and 1976-77 are all kind of similar in temperature but 1975 pulls the curve down and how you decide to average it in (one year, three year, five years) tells you where you think the oscillation was at its bottom - almost definitely before 1975.
In the same way, 1998 was a devastatingly hot year. And because of its heat, the Global Warming thing really turned on, propaganda-wise, because, man, you just could not deny that this year was notably hotter than the years before. But the years after that were cooler than 1998, and relatively steady, not rising nor falling except within statistical noise. Again, 1998 pulls our average-graph into a curve and where the top of that curve is depends on how you do the averaging. And after 2003, it's fairly clear that we are cooling. How dramatically depends on your style of measurement and your fudge factors of choice, but we're certainly not getting hotter and we've never gotten back up to 1998.
So now go back to dougo's graph. Here's the problem. In our recorded history of semi-adequate actual measurement with devices that give us numbers, we only have two peaks and two troughs.
And it's just not enough to say. It's not enough to make predictions over the long haul. We were definitely warming from 1975-1998, hell yes were we ever. But we're cooling now, as best we can tell.
And so the next problem is that people with agendas are not measuring peak-to-peak or trough-to-trough. They say, "We have dramatically warmed from 1905 to 1998." And while they do have a reason for starting at 1905 - it's one of the first years we have a lot of numbers for - they are measuring from a trough to a peak. Of course that shows warming. Stopping at 1998 is bastard-like, though. Everyone who's anyone knows that was a scary-hot year and you're just using that to make your numbers look good. Go to the current year if you want to do that. It's 2011. You have numbers until 2010. They're in your spreadsheet. I see them right there.
Our numbers can tell us a lot of things. They can tell us very strongly where we should be looking and when if we want to learn more - we need to be studying those icecaps and watching them come back, or not. But until we get more peaks and more troughs well-measured, we can't tell if the whole cycle is trending higher or lower. We just don't have enough data points to average that way.
(By the by, I don't know what the hottest year was in the 1940 era and how it compared to 1998; nor the coldest year in the 1905-1910 era and how it compared to 1975.)