The Wager and Ocean Heat Content

Since I was playing with data to see if the 2015 climate wager a fair bet, I thought I may as well include the Ocean Heat Content (OHC) data. I linked to the data on Lucia's site were she was wading in on the tiff between Tisdale and Tamino.

The OHC data is probably some of the most important for determining the change in climate. It is also some of the worst data available. Tisdale honed in on some of the more resent data, from 2003, when the ARGO floats started measuring the temperature, salinity at depths of 2 kilometers. I have never played with this data before, but assumed that ARGO must have made some improvements. The ARGO data is far from perfect, but better than what came before.

In the chart above you can see that the OHC data is pretty noisy. Around 2001, there appears to be a flattening of the trend and less noise. The noise decreases from 2001 to about 2004, then stabilizes somewhat. The change in the noise is so much I can't even guess if there is the start of a real trend or if before 2001 the data was just pure crap. It is probably a little of both, but I don't think you can say with confidence anything but OHC has increased. I plotted the quarterly data in blue with annual in red.

Since I am playing with the wager, I took the same time frame, 2001 to present that I did with the other data. I stuffed a few short regression lines on the chart above, then made a new chart starting in 2011.

This shows there may be a real flattening trend, but that warming might increase until 2015. There are not enough data points in the quarterly data to do much predicting. The data looks odd though. I would expect seasonal fluctuations in the OHC, but not that much. Looking at the quarterly data the northern hemisphere winter has most of the peaks. That is not unexpected, the southern hemisphere has more ocean area. The angle of the sun though still favors the northern hemisphere. So I chopped out the first and second quarter data then made a new plot.

With a short time frame, noisy data and not many data points, this probably means nothing, but it is a little odd that here the possible trend turns negative. A negative trend that better matches the ENSO/PDO is what I would expect. With the limited data, comparing the first half of the year to the last half could show anything. Still that looks to be a fairly significant change in the slopes. Does it mean anything? Dunno. From a wager stand point, it reenforces my opinion that the wager is a toss up, but that is my gaming point of view not a very scientific point of view.

It does get me thinking though. Most scientists are looking for lags based on fix time periods. I think it is more reasonable that lags are likely related to threshold values. Fixed time lags, perfect correlations and dominate specific natural drivers are not what I would expect in a somewhat chaotic system. Natural variation can be amplified, but I doubt that a tight phased locked loop type of control is in charge of the amplification. A little extra CO2 or land use change or dry/wet annual change or touch of an increase/decrease in solar could shift the threshold. Throw in a few oscillation shifts and before long you have a complex puzzle.

When I was picking on Nicola, he was looking for lags that matched the small changes in solar energy and estimated about 60% natural forcing. That may be true in the past, even using the slightly outdated reconstructions, but it does not have to apply to the future. Wagering wisely requires knowing probabilities, but winning against players that also know the odds, requires the use of tells, hunches, patterns and gut instinct. You will never win every hand, but knowing your opponent tilts the odds in your favor, until he shifts his pattern.

UPDATE: After more eyeballing and even resorting to ttests, the data before the middle of 2003, sucks. It is not all that great after 2003 either. The ocean heat content has been dead flat since 2004 for all practical purposes. Mention the rise in OHC though and you may get better odds.

Climate Wagering - Have the Aliens Landed?

Because of a simple bet, I found a few data bases easily on line to use my own simple methods determine how fair I felt the bet to be and how confident I felt about wagering (i.e. how much I may wager myself). Since the only thing at risk is a little of my money, I can use unproven methods, make large assumptions and cherry pick the data I wish for my analysis. Should I have found the data compelling, I would have tidied up my analysis and used more standard methods to verify what is had found to justify a larger wager. The greater my confidence, the more I would bet.

So how is this different than the real world? There is no difference. A climate scientist or group of climate scientists find that there is very high risk which means a high cost to reduce the risk. The people financing the bet want to be confident enough to justify the wager, so they want more standard analysis done before writing the check. Not understanding this is odd, one might think alien behavior.

The stacked regression analysis I may have created (the same thing can be done various ways)could be considered "novel". That is a big leap, but it is just for an example. A variety of statistical analysis methods used by climate scientists are "novel". It is reasonable to me that if the climate scientist wants backers for his bet he would be willing to allow more in depth analysis of his data and methods.

This is a major sore point in the climate change debate. People expected to chip in on the financing want confirmation. The methods used by some of the scientists is novel enough that other statisticians and scientists will not sign off on the "skill" of the analysis. The reviewing statisticians and scientists need the actual codes for the analysis and data used in the analysis to reproduce the results.

For some reason, the scientists won't or can't provide the actual data and the code as used in the analysis to the reviewers. Since the reviewers can't reproduce the results, they are not willing to chip in on the wager. That's life in the high stakes climate change betting game.

Most climate scientists are human (There are a few suspected aliens, one has indicated he may be Venusian), and as all humans are fallible, make mistakes. Human scientists know that and search for their own mistakes and will grudgingly admit their mistakes should someone else find them. Alien scientists believe they are infallible, which may be true on their home world. Scientists that are alien or heavily influenced by alien cultures, don't understand the human logic of intense analysis before going all in.

Alien logic dictates that if an analysis method is less than optimum (remember they are infallible,) that increasing the complexity of the analysis improves its accuracy, though it reduces its reproducibility. The alien influence on Wall Street where Maximum Overlap Discrete Wavelet Transforms are used to determine that the longer your money is in a hedge fund the more risk you assume, illustrates the beauty of complex analysis.

Humans know that reduction in complexity is required to convince financiers to pony up the cash, if the potential financiers are wealthy, and that increasing complexity is useful in convincing the less savvy masses (pension funds), to chip in on the wager. This is the, "if you can't dazzle them with your brilliance, baffle them with bullshit" truism.

Determining if a scientist is alien or human may be a new game, "Is he/she Alien?" Any game should have reasonable rules. Perhaps we can take a test scientist for trip through the analytical gauntlet to get an idea of the basic rules.

Nicola Scafetta, is thought by a few to be of alien origin. He has a published, peer reviewed paper that used data from the ACRIM and CRU with methods outline in his paper which he suggests are clearly explained in this book. On Nicola's website he lists his publications. One of his published papers is actually written in English so that is a good one for humans to review. Note that his papers in an alien language do not necessarily prove he is an alien.

The book he referenced costs money. Savvy financiers will not invest a dime until there is a reasonable expectation of a return on that investment. So we will keep the game simple with rule number one - Shell out cash only when you expect a return on the investment.

Since we humans are on the lazy side (that is not a dig, lazy is an efficient use of mental and physical energy), Rule two - look at the easiest stuff first. The CRU data is easily download and can be imported into any spread sheet. Once on a spread sheet, we can export it to any program we wish.

The CRU data Nicola linked to is our first clue. Humans know that humans make mistakes and the more steps a human takes the more mistakes they are likely to make. The CRU data has no notes. Those may be on a previous page, but the link was to a page that had no notes. The data on that page appears to be monthly temperatures with annual averages below in the next row. Mixing data by rows complicates the use of the data. Also the data appears to be space separated values, but the number of spaces are inconsistent for the apparent monthly data further complicating the use of the data. There are 13 columns in the apparent monthly data. One of those columns may be a monthly average or they number of months may be based on an alien calendar. A quick check indicates the 13th column is likely to be an average of the monthly temperature anomaly. The data that appeared in the alternate rows then may be actual temperature rather than anomaly. Since I don't care, I am trashing the alternating rows and just using what appears to be anomaly data. I would assume the anomaly data is global average anomaly, but since I am human and therefore lazy, I am not going to waste my time trying to find out for sure. It is my money Nicola wants, if he wants it bad enough he will make it easier for me to justify the investment. I expended the least amount of effort possible to organize the data he linked for my analysis.

If I didn't mess up, this may be the data that Nicola used. If I did mess up, I don't really care. I am not selling this stuff, Nicola is, I just want to see if it is worth the investment.

The ACRIM data is somewhere on the page Nicola linked. Right now I am pretty sure that Nicola is not a very good salesman. The more work I have to do, the less likely I will invest. If Nicola was really hungry to sell, he would simplify the effort for potential buyers. Since I just happen to have the ACRIM TSI data and a few other TSI records/reconstructions from Leif Svalgard's website, I will use them instead of fighting with ACRIM site which lists data products but won't let me down load them.

The solar TSI data has a mean of roughly 1366. I adjusted the DORA set so that it has the same mean value for the same range (I trimmed the Hadcru set a few years so I am using the same time periods for the DORA set.) Since the ARCIM set is short, I adjusted its mean, but let it float a little above the mean of the others. Both TSI sets were scaled with 0.25 multiplier to roughly match the amplitude of the temperature set.

Above is a chart with the data collected. Scaling can be changed as needed based on Nicola's paper to see what it looks like. I used the DORA TSI set because I want something recent and unbiased. If need be, I can add the Lean at al 2000 or Lean et al 1995, but since both have been revised, I would rather avoid their use.


In Nicola's abstract, he states that they find that at least 60% of the warming since 1970 is due to natural effects created by the orbital cycles of planets, mainly Jupiter and Saturn. The planet's gravitational forces tend to tug the Sun around, changing its impact on climate. This impact is seen in decadal and bi decadal cycles. In the introduction Nicola lists a variety of oscillations that appear to some what match the orbital cycles and even mentions the Chinese calendar which most of us have seen parts of while enjoying Chinese buffets. A pretty convincing argument that there is a lot of stuff going on, but I will pause momentarily to order Chinese.

Before spending much time trying to reproduce his work, I would like to see if it is worth the effort. Nicola's paper was accepted in April of 2010. The most recent TSI reconstruction he listed was Solanki 2007 with Lean 2005, 2000 and 1995 and Hoyt 1997. Without creative methods, there is not much correlation between the HADCRU temperature series and the TSI measurement (ACRIM) and reconstruction (DORA.)

The Solanki 2007 is not on this chart using Svalgard's data. The Wang reconstruction should be the same or similar to the Lean 2005. It would appear that since the TSI reconstructions Nicola used are a little out dated that his conclusions may be over estimated. This causes this potential investor to think:

1. The newer data will likely change the results.

2. Since the code for Nicola's paper is not available, I would have to reinvent the wheel to first replicate his code, then update the data used in his analysis to see if newer TSI reconstructions changed the results.

Conclusion: Nicola probably is not an alien because he exhibits the human trait of laziness by not providing working code and as used data sets. However, the complexity of his analysis does tend to lean toward alien behavior. This possible investment will be tabled until the proposal is revised. Time to eat Chinese.

More Analyzing the Wager to Death

The wager based on the 2015 average GISS surface temperature versus the 2008 surface temperature is interesting. The overly simple regression analysis, (or eyeballing) leads me to believe that the wager is pretty fair and is virtually a coin toss. The originator of the wager of course feels that it is a sucker bet, or he would not have made it.

In my opinion, the timing and intensity of the ENSO quasi-cycle makes or breaks the bet. The surface temperature response to the ENSO/PDO quasi-cycles does tend to favor the originator of the wager. The atmosphere responds differently to the ENSO/PDO than many would expect.

Sensible heating (warming) has a quicker temperature response than sensible cooling as it approaches the latent cooling threshold. So while the GISS temperature shows warming quickly, cooling takes a longer time to happen. There is some correlation between the GISS temperature and the ENSO/PDO, but that is super imposed on a general warming trend. Part of that trend is due to land use/CO2 and part is the natural way the atmosphere responds to sensible cooling. In order to cool, the ENSO/PDO cooling impact has to be enough to cause enough latent atmospheric cooling. Since the Hurricane ACE is low, that part of the latent cooling is not available (which is fine by me).

The difference in the rates of warming and cooling is pretty clearly shown in the temperature record if you compare the slopes of the warming trends to the cooling trends. That means that a lag in cooling relative to ENSO/PDO is somewhat likely, more sever winter storms and chaotic spring storms take longer to get rid of the moisture than a kick butt hurricane season. This makes predicting future climate more complicated and the wager more fun.

The above chart is the GISS global surface temperature in the burgundy color that may look black with the period from 1950 to 1982 in thin line and the rest in thick line. The regression for each period is shown in the same color with the varying thickness. In the orangeish color is the ENSO Data from NASA and in the light blue is the PDO index data from the University of Washington. I have inserted an orange dashed mean line for the ENSO which is pretty close to the PDO index. The chart is pretty busy and I could have picked better colors, but the information is there.

Prior to 1982, there was a better correlation between temperature and the ocean oscillations. After 1982, the temperature trends keeps on heading up with some correlation with ENSO/PDO super imposed on the warming temperature trend. The correlation is not all that great because there are more pieces to the puzzle, but ENSO and or PDO do appear to be players in the puzzle of reasonable significance. It is a wager analysis, so I can take a few liberties.

The more the ENSO/PDO trends falls below the ENSO mean, the more the impact the ENSO and/or PDO appear to have on the temperature trend. So there is a cooling lag, not dependent on time, but dependent on a magic temperature threshold that can change with the overall impact of various forcings both natural and man made. Since the ENSO/PDO values are at the lowest point since 1982 and on a par with the values prior to 1982, there is a reasonable likelihood that the GISS temperature trend can flatten and possibly start cooling if the ENSO/PDO trend continues below the mean of the overall trends.

I call the temperature threshold magic, because I have no idea what its real value is in the current state of the oscillations. I am pretty confident that if the temperature trend starts to cool, that the cooling slope will be less than or equal to the slope of the last cooling trend.

In this chart from before, I weighted the analysis to the near term, because of the wager. The blue line is the regression of GISS global monthly from 2001 and the red line is the regression for the wager start date in 2008. Warming wagers will base their confidence on these trends (pseudo trends if you prefer). The Orange line is the regression starting in 2009, March. The orange pseudo trend was cherry picked to find the most resent regression with roughly the best R squared value. That should be the maximum cooling pseudo trend possible with current conditions. The two dashed trends are the more recent pseudo trends which I use to estimate the more realistic near term path of a cooling trend should it actually develop. The green dashed line is the win line for the guys betting on 2015 average being below 2008 average.

All this boils down to the ENSO/PDO staying at or below the magic temperature threshold. As a reminder, I am not a climate scientist, statistician or psychic, I just enjoy a friendly wager on occasion. Anyway, the climate science professionals may get a kick out of this analysis because when wagering is involved, cherry picking is allowed.

Analyzing the Wager to Death

The great thing about a wager, is even if you are not involved, you can waste time, cheery pick, try crazy stuff to rationalize the outcome, just do about anything and it is perfectly acceptable irrational behavior. So I am analyzing the wager to death!

The stack cluster thing I am using is not predictive, it is just a ballpark look a things. The timing of the wager fairly short term though. It is will the temperature recorded by GISS in 2015 be above or below the average of 2008. The data is noisy, this La Nina will fade, a new El Nino will start and the period of the bet may be in the middle of a new La Nina. It pretty much is a coin toss.

Just to try to get a little better picture, I changed the length of the individual stacked regressions to 24 months. Since I am more concerned with the future, I made a chart of the most recent regressions with overall linear regressions from 2001 and 2008 to the end of the data in March of 2011. The most recent 2 year regressions indicate a down turn. That is of course subject to change. So I added one more regression starting at March 2009 to March 2011.

The main clustering was positive, so temperatures should rise, but with 2015 just around the corner, I biased the stack for the closer term. At least that is what I think. Because the RSS and UAH data for April are already out, I know that there is an up turn as the La Nina fades. The orange 2009 to 2011 line should be a lower limit for the next few years. Hansen's GISS data tends to be the warmest of the surface averages and GISS on occasion makes new adjustments to their data. By 2015, I would not be surprised if GISS adjusts a little to get more in line with the other guys. The bet was against the 2008 average which is already published, so I am thinking 0.5 with remain the target, even if it is adjusted down a touch.

After all this playing around, it still looks like a toss up with the winner being the one guessing the start of the next La Nina. I may waste some more time and download the ENSO to see how that may come into play.

Cherry Picking, Trends and Climate Catfights - For Your BS Dectector

For those of you that don't know, I have a basic bullshit detector that I occasionally add to. There is a lot of bullshit, so the detector has a lot of room to grow. I have been meaning to add a climate science filter for the BS detector, but it is difficult to simplify things enough for most folks to understand. Since Tamino and Bob Tisdale are in a cheery picking cat fight, now is as good a time as any.

A climate trend is generally accepted to be 30 years. That is because there is weather and there is climate. Thirty year trends make it somewhat easy to see past trends but it makes it damn hard to predict trends. Even past trends can be somewhat controversial. It is difficult to assign with complete confidence the factor that forced the trend. The 1910 to 1940 trend is one that is often compared to the 1980 to 2010 trend (you can fudge a few years either way as long as the period is at least thirty years long).

Since I have been accused of cherry picking, I try to follow the rules, still if I mention the 1910 to 1940 period, "CHERRY PICKING!!" is the first thing I hear on most of the warmer sites. Tamino did his mathematical magic a year or so ago and came up with what should be the minimum time period for a statistically significant trend. His math indicated that roughly 14.7 years is the minimum and that even then it depends, if it can be statistically significant. I am no statistically wizard, but what he did made sense.

That is why I picked 15 years as the regression length for my 15 30 stacked regression method. With each regression 15 years long and 30 years of regressions it makes it a little harder to call it cherry picking. It is right at the edge though.

The minimum thirty year trends cuts both ways. If the actual measured trend falls outside of the modeled trend after thirty years, that makes people question the validity of the models. That makes the modelers reconsider their models and change the uncertainty of their model accuracy. The length of the period also makes it virtually impossible to predict trends. Waiting until after something happens to predict it happened doesn't impress most folks.

Wagering on climate changes takes cherry picking out of the picture. Had it not been for the wager that the 2015 temperature will be higher than the 2008 temperature I would never even thought of using such a short period of time. If it had not been for the cherry picking issue, I would not have created my stacked 15 30 regression method. Just like the bullshit detector, the stacked regression method is a work in progress.

The spat between Tamino and Tisdale is over use of an "apparent" trend in the ARGO ocean heat content data starting in 2003. That is only eight years of data, so it is not a trend, it can be called an apparent trend developing, but not a trend. Since Mr. Tisdale did not phrase his post answer as a question, he can't win the trend thing. "What could be a developing trend starting in 2003...", would have avoid the trend argument but since cherry picking is more ideological than rational, there is no way to avoid that, other than a friendly wager.

Tamino is pretty math savvy, but is not infallible mathematically. He has face planted trying to defend an inappropriate use of a particular statistical method in the past, being human, he is likely to screw up again. We are all human. Wagering is a fun way to reduce the tension of the cat fighting. I am not talking mega buck wagering, that is a fool's way to attempt to win an argument. If someone fails to take his outrageous wager, a fool thinks he can crow. There is always someone with more money and more savvy on any particular subject. Unless you have the nuts, don't bet big and if you do have the nuts bet wisely to get the most out of it.

I am posting this only so I can add the link to the BS detector. I doubt that Mr. Tisdale and Tamino will make a wager, but if they do, I will try to use the stacked regression method to estimate the odds.

Wagering on Climate Change

Wagering on climate change seems to be growing. At least offering wagers, I don't know how many takers are out there. Bishop Hill mentioned on his blog a new offered wager of one thousand pounds based on the GISS global data being below the 2008 average of 0.5 in the year 2015.

I modified my stack regressions for the wager using GISS monthly data from 2001 to present which is March 2011. I have a thick purple line which is the regression for the whole period and a thick greenish colored line which is the regression for 2008 to present. Of course I have all my stacked regression lines. Since the object of wagering is to predict an outcome, I have modified the stack by including the most recent regressions as thicker dashed lines. The wager is if the GISS 2015 average will be above or below 0.50, 50 on the chart. This looks to be a fair wager.

It is fair because the possible paths average just above the wager temperature but the most recent paths are below the wager temperature. Based on the temperature track record from 2001 it is close to a coin toss. This type of wagering is a good opportunity to sucker people in to being over confident in their predictive skills. I would not take this wager without negotiating the odds. 2:1 odds may be pushing it, but after a few courage inducing cocktails, may be worth the bar tab. So a 500 payout above 0.5 with 1000 and the push going to guys betting on below 0.5 sounds like fun, 5:4 is doable.

Remember, predictions are hard, especially predicting the future. The stacked regression method does give you a little more to work with, I think.