.

Monday, May 16, 2016

Bob Tisdale's trick of hiding the data, revisited

Sou | 12:40 AM Feel free to comment!
I've written previously about the trick Bob Tisdale uses to hide the fact that observed temperatures are now very close to modeled projections. He uses the same trick every time he posts his update to global mean surface temperatures (as archived here). It's an obvious trick, which should fool no-one except people who aren't familiar with charts and disinformer tricks, and those who want to deny global warming. I shouldn't need to repeat what he does, but I will, briefly. The wonder is that Bob keeps repeating his trick, even though he knows it's deceptive for two reasons:
  1. Firstly it's deceitful because Bob chops off much of the past two and a half years of data, by plotting a 61 month moving average.
  2. Secondly it's wrong because he doesn't let his readers know that the CMIP5 data he's using has only estimated forcings since 2005. The actual forcings had a net effect lower than what the CMIP5 models were based on.
Here is a plot of monthly CMIP5 multi-model mean, using the original forcings estimated in the IPCC's latest report from 2013. I've also plotted the monthly observations, which in part because of the current El Niño are currently way hotter than the multi-model mean.

Figure 1 | Monthly global mean surface temperature, observations and CMIP5 multi-model mean. Data sources: GISS NASA and KNMI Climate Explorer

Contrast Figure 1 with what Bob Tisdale did. He plotted a moving 61 month average, which means that his end points are cut off. In this case he's chopped off much of 30 months of data, or two and a half years of observations.

Figure 2 | Chart of 61 month moving average of HadCRUT4 and CMIP5 multi-model mean. Using a moving average chops off two and a half years of data at each end. Source: WUWT

Contrast Bob's chart in Figure 2 above with a plot of GISTemp, to April, together with CMIP5. The chart below is just showing the plot of LOESS smoothing of the monthly data, the same data as is plotted in Figure 1 up top.

Figure 3 | LOESS smooth of monthly global mean surface temperature, observations and CMIP5 multi-model mean. Using a LOESS smooth doesn't chop off any data, so gives a better picture of the end points. Data sources: GISS NASA and KNMI Climate Explorer

Bear in mind that the forcings in the models since 2005 were overestimated. Stefan Rahmstorf has updated the CMIP5 vs observations from Mann 2016 with 2015 observations. This CMIP5 plot below includes adjustments for observed forcings, which shows that the multi-model mean is very close to what is observed.

Figure 4 | Global mean surface temperature and CMIP5. The CMIP5 multi-model mean has been updated with recent forcings and GISTemp data includes 2015 observations. Source: Stefan Rahmstorf, updated from Mann15 article in Nature's Scientific Reports.

Bob Tisdale also put up a chart of the difference between the CMIP5 multi-model mean and HadCRUT4 observations. I've animated his chart to highlight some of the deception and how his own chart betrays him:

Figure 5 | Monthly model mean (CMIP5) minus monthly observations (HadCRUT). The red line is the 61 month moving average, so chops two and a half years of data from the end points. Source: WUWT

As you can see, even though Bob chopped off much of 30 months of data with his 61 month filter, he superimposed the running mean over the monthly difference between HadCRUT and CMIP5. That's why the final data point on the far right shows that the observations are more than 0.2 C warmer than the multi-model mean. Put another way, the model projections are more than 0.2 C colder than the observations in March 2016. (HadCRUT4 data only goes to March. I don't know why Bob uses HadCRUT4. If he used GISTemp he'd have been able to include April data.)



So when Bob wrote that climate models "are diverging rapidly from reality", what he should have said was that the models are running colder than observations. Then he could have explained that one of the reasons they are running cold, despite having recent forcings too high, is El Niño.

When Bob wrote: "The difference now between models and data is worst-case, comparable to the difference at about 1910" he was wrong. What he should have written was that two and a half years ago the difference was comparable to the difference at about 1910, but right now the models are colder than observations. The actual temperature is hotter than the model projections.

When Bob wrote: "It’s very easy to see the recent record-high global surface temperatures have had a tiny impact on the difference between models and observations." he was plain wrong. That's what he wants his readers to think, which is why he discards the last two and a half years of data. The recent global surface temperatures have had little impact on his 61 month filter, because the data is given less weight at the end points. The recent global temperatures are above the multi-model mean. The world is hotter than projected. (That's in part because of the current El Niño.) Looking at two datasets:
  • HadCRUT4 as at March is more than 0.2 C hotter than the multi-model mean. 
  • With GISTemp, the recent global temperatures using the January to April 2016 average, are 0.3 C above the CMIP5 multi-model mean.


From the WUWT comments


There are only three thoughts so far. I wasn't going to bother, however the first comment is a treat. It could be a Poe. I don't recall coming across News First Site ©2006 before. I particularly liked the capitalised Palm Trees that grew in Canada tens of thousands of years ago. The "very cold time" ice age cometh comment is fairly standard. Saying that he or she "formed a Corp" and became a CEO instead of a meteorologist has to be a good thing for the world at large. I suspect his "Corp" if it exists could be a one-person show, and a rather lowly one - though who am I to say?
May 15, 2016 at 5:22 am
Very interesting article as always with whattsupwiththat. This is perhaps my favorite on WordPress. They always have interesting stories with facts & of course credits to news links etc.

As for climate, I’ve done study since I was actually 8 years old. I attended College & was going to become a meteorologist but I ended up as forming a Corp & becoming a CEO instead. I still kept study going on climate. I’ve learned there are many cycles of climate. About every 30 years it cools then it warms. Then every few hundred years very cold or very warm. We are due now for a very cold time. Last was early 1800s and it comes every 200 years or so. Larger warming and cooling is 500-1000 years then there are huge changes over tens of thousands of years where an Ice Age or warming so harsh that Great Lakes form, Palm Trees grow on a year round Canadian open Ocean Coast. The only thing caused by humans from my experience is City Heat Island which is reversible. Humans do cause pollution though. I always find these climate articles interesting though from this part of WordPress. They do a great job over here.

Tom is surprised to find that it was quite a bit colder than now, back in 1998:
May 15, 2016 at 5:28 am
Why is the 1998 El Nino spike so muted in the present graphs, such as figure 1?

Bill Illis predicts that in April 2017, the temperature will be 0.4 to 0.5 C colder than now.
May 15, 2016 at 5:39 am
Just noting that temps appear to have peaked around mid-late-February. The Nino 3.4 index peaked in mid-November so that makes for a 3 month lag once again. I guess I have been pushing this lag period idea for years now but this is just another data point to add to the broken record. And this lag period never stops. There is always a 3 month lag to whatever the ENSO conditions are on a continuous on-going basis.
When we get a moderate La Nina later this year, temps will be down by about 0.4C/0.5C by this time next year.
Bill could be right on a monthly basis, however on an annual basis his numbers would most likely be an overestimate of cooling if there is a La Niña this year, and more so if there isn't. The chart below is just from 1997 onwards. The biggest drop in annual surface temperature was from 1998 to 1999, when there was a drop of 0.22 C. The 1998-2001 La Niña was considered moderate, after a very strong El Niño in 1997-98. The 2010-2012 La Niña was regarded as strong, or moderate in terms of sea surface temperature. It came after a weak to moderate El Niño. The drop in temperature from 2010 to 2011 was only 0.12 C.


Figure 6 | Global annual average surface temperature anomaly from the 1951-80 mean. The chart shows El Niño years as brighter orange, La Niña years as blue, neutral years as grey, and years that had both La Nina and El Niño as mustardy green. Data sources: GISS NASA and the Australian Bureau of Meteorology.

From the HotWhopper archives

No comments: