Guest essay by Barry Wise
Christopher Monckton has pointed out that a trend of zero or smaller can be calculated stretching back over 18 years, but critics have pointed out that this encompasses the super el Niño of 1998 and so biases the trend downward while the overall temperature is still rising. Of course they also don’t mention that the el Niño biases the trend upward when the trend is viewed in it’s entirety. Now Lord Monckton has identified a valid point and so do his critics. What if we look at how the trends vary over differing lengths of time? Can Lord Monckton be validly accused of cherry picking or are his critics nit picking? This article will attempt to show a broader view of how the trends have varied over time, both from the beginning of the the record and from the end.
To begin, at what point do we say that a trend of a given length makes sense in terms of whether it’s an indication of future global temperatures or just a statistical anomaly? Obviously the longer the better, or so one might surmise, but then that’s assuming that the data represents a linear trend. The more data you have the less any additional point will affect the overall trend. Given that, one might expect the trend to oscillate around a given value with a reduced amplitude as it zeroed in on the actual trend.
So what actually happens? For this exercise I’ll use RSS lower troposphere data since that is what Lord Monckton used. Here is the RSS data for the full period of the data from 1979 to the present with the 1998 el Niño shaded in grey and showing the least-squares linear-regression trend line for the entire data set. The el Niño time period is based on data from El Niño and La Niña Years and Intensities. The slope of the line is approximately 1.2 K/century which, as it turns out, is below the low end of the IPCC projection (1.9 to 4.2K/century) but is consistent with a doubling of CO2 with no secondary effects.
Let’s look at how the trend has varied starting from the earliest data. I’ll start with a minimum length of ten years (an arbitrary length but shorter lengths give widely varying trends that make it hard to read the graph) and I’ll increase the length in one month increments. The trends are plotted based on the end date used for the data of each trend, with the full data shown below as a reference. The shaded area brackets the 1998 el Niño so we can see where it enters into the computation of the trends. Notice how much the trends vary prior to the el Niño. Everything from about 1.6 down to .4 K/century. This would indicate that data that is less than 20 years (at a minimum) is unreliable to discern the trend over a longer term. Notice too how the trends peak with the el Niño but immediately start tailing off. By the end of the data the trend is at the 1.2 K/century we calculated before and the change in the trend is flattening out. Also of note is the rapid rise of the trends when the el Niño occurred.
Now let’s look at how the trends changes as we increase the length in one month increments starting at the present and working backwards to see how Lord Monckton’s 18 year 8 month value fits into the changes that occur as we vary the length of the data. In a similar fashion to the previous chart, I started with a minimum data length of ten years. Notice that not only are there negative trends where the el Niño data is included but there are also negative trends prior to that data. Additionally, the trends prior to including the el Niño are even more pronounced, longer and extend back to June 2000 which is over 15 years. Also of note is that this includes the 2010 el Niño which, by it’s relative location, should bias the trends in a positive direction at these lengths. At no point does the trend exceed .5 K/century for the data after the 1998 el Niño. Just prior to the el Niño the trends are approximately .7 K/century.
This article has just been my attempt to show a broader view of how the temperature trends have evolved. I make no claim to whether Lord Monckton or his critics are correct. In summary we now have over 35 years of satellite data with over 15 years post 1998 el Niño showing little, no or even negative trends at that length. The data prior to the el Niño also shows trends that are at or below that the IPCC has promoted, not to mention the entire record. While some have critiqued Lord Monckton’s trend because of the inclusion of the super el Niño I would question their consistency because I haven’t seen a similar complaint based on the much larger effect that it had on the trend from the beginning of the record. Given the lack of a positive trend post el Niño, it would appear that there was a step that occurred in the earth’s atmosphere’s temperature. The present el Niño is being touted as being another massive one. Will it too show a step?
Is the Y Axis in the bottom image for Fig 2. correct? It looks like it should be K not K/Century. In Fig 3 it is just K.
Gas thermodynamics is not something I normally comment on.
This article from phys.org published yesterday may be of some interest:
Scientists experimentally demonstrate 140-year-old prediction: A gas in perpetual non-equilibrium
October 19, 2015 by Lisa Zyga report
“Now for the first time, physicists at JILA, the National Institute of Standards and Technology, and the University of Colorado at Boulder have experimentally realized a three-dimensional cloud of gas that never reaches thermal equilibrium,”
Read more at: http://phys.org/news/2015-10-scientists-experimentally-year-old-gas-perpetual.html#jCp
http://phys.org/news/2015-10-scientists-experimentally-year-old-gas-perpetual.html
Evince, in replying to my question you say,
The keyword in Monckton’s question is FAR. “How far back back in time …”. Going back from 2014 there are more than a few years when the temperature was higher than 2014. But we are not looking for annual temperatures. We are looking for a TREND in temperature. The longest zero trend we can find is 18 years and 8 months. Hope this helps 🙂
Yes, it does help. I hadn’t thought of it like that. Thanks a lot.
MCT.
Some debate around Monckton. We all now that the IPCC and alarmists cut the curve at the peak of 98 El Ninjo and user this all the way until 2013, with no updates. These days the do not show temperaturer curves in their propaganda.
I do not consider that enough attention is being made of the fact that there are two 9not one0 pauses in the RSS data.
In a theory which requires that whenever there is a rise in CO2 there must always be a corresponding increase in temperature (because of the increased forcing caused by the increased level of CO2 in the atmosphere), a period when there is no corresponding rise in temperature potentially contradicts and invalidates the theory.
Of course, Earth’s climate is complex, it is a dynamic system never in equilibrium with a number of lags such that there is much variation in temperatures on a short timescale which can mask the effect/signal of rising CO2. So there is obviously an issue as to how long must a lack of warming go on for, or how long must a period of cooling go on for before it invalidates the CO2 warming theorem. Santer initially suggested that a period of 15 years would be problematic, later he revised that to 17 years.
But in the RSS record there are two periods of about 17 years duration where there is no statistically significant warming trend. the first runs from 1980 (the start of the data set) through to 1996 (ie., up to the run up to the Super El Nino of 1997/8), and the second pause runs from that event (say 1999) to date.
One pause of more than 15 years duration would be potentially problematic for the basic tenet of the CO2 warming theory, but two pauses of this duration are more than twice as problematic. The chances of there being two such pauses running closely on from one another is remote.
Of course, it may be that 15 years is too short a duration to contradict the theory because of the constant out of equilibrium dynamics of the system, but that said, I do consider that sceptics should emphasise the fact that we are looking at two pauses not just one pause in the RSS data.
Whatever, the position may be, all one can say is that the signal to CO2 (if any at all) is so small that it cannot be eeked out from the noisy signal of natural variation such that it cannot be detected by our best measuring and most sophisticated equipment within the limitations of that equipment. If the limitation of that equipment and resulting error margins thereto are small, then climate sensitivity must likewise be small (if any at all). if on the other hand the limitations of that equipment and error bounds are large, the climate sensitivity to CO2 could likewise be large (although of course, it could well be non existent).
The satellite data set is extremely problematic to the CO2 warming theorem since there is zero first order correlation between CO2 levels and temperature in that data set. As I say, I consider that the first period when there was no warming (1980 to 19960 to be as problematic as the second period when there has been no warming (the infamous pause).
If you insist that the last 18yrs of the RSS record is meaningful then so is the first 18yrs.
So see this….
http://woodfortrees.org/plot/rss/from:1978/to:1997/trend/plot/rss/plot/rss/from:1997/to:2015.8/trend/plot/rss/trend
OK
So we have a mean trend-line for the whole record (purple). … the sensible option.
We have the trend (blue) for the last 18yrs.
We have the trend (red) for the first 18yrs.
Now for there to have been a true hiatus then the red line when extended forward in time should be ABOVE the blue. Correct? Or else we are still “warming” above the initial trend.
Do that.
Err, it doesn’t even cross the blue line until ~2025!
I do not intend this analysis to do anything other than to highlight the absurdity of calling the last 18yrs on the RSS record a “hiatus”.
If you wanted to argue anything by cherry-picking it out to confirm your bias then actually a warming step is more correct – that still remains above the initial long-term trend.
Clue: the big Nino of 97/98 biases things horrendously – a known satellite sensing anomaly.
Which is why UAH (which version?) as well as RSS do NOT measure surface temps meaningfully.
No one claims that the satellite data set measures surface temperature. It does not. It measures atmospheric temperature. but according to the CO2 theory, the atmosphere must warm, and it is this that causes the surface to respond by surface warming.
It is not cherry picking to analysis the data set as a whole. So the question is, what does this data set inform, when looked at as a whole.
You are correct that the 1997/8 Super El Nino sticks out like a sore thumb. It is the only significant warming to be seen in the satellite data set. The satellite data set reveals a one off warming in and around that event. Since the 1997/8 Super El Nino “horrendously” biases matters, it is interesting to ponder upon what the data set might have shown had that Super El Nino not taken place.
Your plot shows three trend lines.
First, the purple line is a straight line linear trend fit, and the data does not suggest such linear response, and I would suggest that such a line is not a reasonable interpretation of the data, and what it is informing.
Second, the red line deals with the period prior to the 1997/8 El Nino, and it is apt to look at that period since as noted at the outset the 1997/8 Super El Nino sticks out like a sore thumb. The red line shows very slight warming, but not statistically significant warming.
Third, the blue line deals with the period post the 1997/8 Super El Nino. It shows slight cooling, but not significantly significant cooling.
The data set when looked at as a whole shows initially a no statistically significant trend period lasting approximately 16 or so years. Followed by significant rapid one off warming coincident upon the 1997/8 Super El Nino. Followed by a second period of no statistically significant trend lasting approximately 16 or so years.
What is the significance of this? Well that is moot, since it is not clear whether a period of no warming lasting 5 or 10 or 15 or 20 years (or whatever) is really significant or not, but much depends upon the claimed levels of Climate Sensitivity to CO2. The higher the claimed sensitivity, the more relevant periods without warming become. Although, the significance of such periods may be moot as far as the CO2 theorem is concerned, I would suggest that it is highly relevant as far as the models are concerned, such periods are very important for testing their worth.
First, at no time prior to about 2013 was there any claim that the models projected/predicted periods of 15 years (or so) without warming. Now it is claimed that some models do project/predict this, although no detail is given, and in particular no detail as to the level of CO2 during such period; obviously the forcing is less when CO2 levels are at say 320 to 340 ppm, than it is when CO2 is at 380 to 400ppm. It is therefore important to know at what level of CO2, some models project/predict periods of 15 years (or so) without warming. Is this early in the 20th century, mid 20th century, late 20th century, or the 21st century?
Second, whilst we are told that some models project/predict periods of 15 years (or so) without warming, how many models project/predict two such periods closely following on from one another.
Svalgaard may have just confirmed all the Electric universe theories he never had. 🙂