Monckton, RSS, and no warming since September 1996
While browsing a climate change article on Huffington Post, I noticed a global warming denier using a Watts Up With That post by Lord Monckton as "evidence" that global temperatures haven't changed since September 1996. In it, Monckton uses least squares regression to show that satellite data from RSS is flat (trend: -0.0001394ºC per year) between September 1996 and January 2014.
Monckton then interprets the data to show no global warming in 17 years 5 months. And he's correct—IF you ignore what we know about RSS AND you ignore the other temperature data sets. The problem is what he ignores.
First off, back in July 2011 none other than Roy Spencer noted that RSS shows false cooling since AD 2000. The problem is that the satellites the RSS team uses have run out of fuel and cannot maintain stable orbits. This introduces a false cooling into the data, as the satellite data is based on measuring the same part of the Earth at the same time each day. If the satellites measure slightly different areas at a given time, this daily (diurnal) drift will bias the data. Every satellite data set has some false cooling due to diurnal drift and each team attempts to correct for that drift to factor it out. The problem is that the RSS satellites are drifting faster than the computer algorithms expect them to be, meaning that not all of the drift is factored out—and therefore the RSS temperature estimate is biased low. This by itself invalidates Monckton's claim, as he uses a dataset that shows false cooling.
Second, RSS is the only dataset to show no warming since September 1996. The rest all show various degrees of warming—even UAH satellite data compiled by Roy Spencer and John Christy. As always, I'm correcting for autocorrelation with ARMA to get the correct p-values.
UAH: Trend: +0.011294ºC per year, p-value = 0.1713
GISS: Trend: +0.009928ºC per year, p-value = 0.0171
HadCRUT4: Trend: +0.009238ºC per year, p-value = 0.1046
I'll add NOAA data and the new Cowtan-Way coverage corrected HadCRUT4 when I get the updated data but I think it's obvious why Monckton choose RSS despite its known false cooling. If he'd chosen any other satellite or surface dataset, it wouldn't have fit his argument as the warming is obvious. Since he didn't correct for autocorrelation, his analysis would have shown that all of the trends were statistically significant. Monckton's argument is an blatant example of cherry-picking. He deliberately cherry-picked a dataset and then cherry-picked his start point to get the answer he wanted.
Despite all this, Monckton's analysis and interpretation was lapped up uncritically by the denizens of WUWT. Monckton's post is a prime example of the blatant, deliberate misinformation peddled on that site as well as a sample of the type of unthinking readers WUWT attracts. And none of it reflects well on Watts.
 |
| Least squares regression trend in RSS satellite data |
Monckton then interprets the data to show no global warming in 17 years 5 months. And he's correct—IF you ignore what we know about RSS AND you ignore the other temperature data sets. The problem is what he ignores.
First off, back in July 2011 none other than Roy Spencer noted that RSS shows false cooling since AD 2000. The problem is that the satellites the RSS team uses have run out of fuel and cannot maintain stable orbits. This introduces a false cooling into the data, as the satellite data is based on measuring the same part of the Earth at the same time each day. If the satellites measure slightly different areas at a given time, this daily (diurnal) drift will bias the data. Every satellite data set has some false cooling due to diurnal drift and each team attempts to correct for that drift to factor it out. The problem is that the RSS satellites are drifting faster than the computer algorithms expect them to be, meaning that not all of the drift is factored out—and therefore the RSS temperature estimate is biased low. This by itself invalidates Monckton's claim, as he uses a dataset that shows false cooling.
Second, RSS is the only dataset to show no warming since September 1996. The rest all show various degrees of warming—even UAH satellite data compiled by Roy Spencer and John Christy. As always, I'm correcting for autocorrelation with ARMA to get the correct p-values.
UAH: Trend: +0.011294ºC per year, p-value = 0.1713
GISS: Trend: +0.009928ºC per year, p-value = 0.0171
HadCRUT4: Trend: +0.009238ºC per year, p-value = 0.1046
I'll add NOAA data and the new Cowtan-Way coverage corrected HadCRUT4 when I get the updated data but I think it's obvious why Monckton choose RSS despite its known false cooling. If he'd chosen any other satellite or surface dataset, it wouldn't have fit his argument as the warming is obvious. Since he didn't correct for autocorrelation, his analysis would have shown that all of the trends were statistically significant. Monckton's argument is an blatant example of cherry-picking. He deliberately cherry-picked a dataset and then cherry-picked his start point to get the answer he wanted.
Despite all this, Monckton's analysis and interpretation was lapped up uncritically by the denizens of WUWT. Monckton's post is a prime example of the blatant, deliberate misinformation peddled on that site as well as a sample of the type of unthinking readers WUWT attracts. And none of it reflects well on Watts.
One wonders why the RSS analysis continues to be published. It seems like everyone knows it's no longer valid (and I'll bet that "everyone" includes Monckton).
ReplyDeleteI understood that RSS have corrected for the orbital decay.
ReplyDeleteSurely Monckton has aways used the RSS figures?
PS I do believe some warming is CO2 related, but RSS at the moment keeps reminding us that the climate is not totally understood. (cf its very slight negative trend over the appropriate period compared to the other datasets argument of around 1.0C per century over the same period). Clearly something so far not included in the models has impacted the last 18 years. This is also shown by the fact that ENSO neutral years have exhibited the fastest rate of growth in your analysis
Yes, they do correct for orbital decay. The issue that Spencer pointed out is that orbital decay in the satellites RSS uses has increased as they have run out of fuel. The algorithm that RSS uses to correct for orbital decay is not able to compensate for that increased orbital decay. That would lead to the cooling that RSS exhibits whereas every single other global temperature data set shows warming.
Delete"RSS at the moment keeps reminding us that the climate is not totally understood. (cf its very slight negative trend over the appropriate period compared to the other datasets argument of around 1.0C per century over the same period)."
No. The only lesson to be had is that RSS may simply be wrong due to increased orbital drift.
"Clearly something so far not included in the models has impacted the last 18 years."
This statement shows substantial misconceptions about both climate models and the observed temperature records, far too much to correct in a comment.