Arguments
Recent Comments
Prev 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 Next
Comments 25201 to 25250:
-
Tom Curtis at 14:19 PM on 21 March 2016Heat from the Earth’s interior does not control climate
pjcarson2015 @50, you claim the rate of release of geothermal energy varies significantly due to earthquakes and volcanism, but provide no numbers. Numbers, however, are easy to find. For example, the largest geologically recent eruption is that of Lake Toba in Indonesia, 74 thousand years ago. Here is a comparison of ejected lava for Lake Toba compared to other geologically recent volcanoes:
The Lake Toba erruption was so large that its global impacts may nearly have caused the extinction of the human race. It errupted with a force of 3.3 x 10^18 Joules, equivalent to 9.15 on the Richter scale. Despite that, globally and annually averaged, it represents a forcing of just 0.0002 W/m^2. That is only a 0.2% increase in background geothermal flux, or 0.005% of the forcing from a doubling of CO2.
Of course, for a change in geothermal flux to significantly effect climate, a 0.2% change for just one year isn't going to do it. Rather, you would need significantly more than that increase year in and year out over the long term. So your theory is, not that one Lake Toba erruption equivalent has occurred, but that multiple such eruptions are occuring year in and year out, and we are simply not noticing.
Alternatively you might consider the flux to come from increased earthquake activity, and that we are experiencing significantly greater than twenty times the normal rate of magnitude 9+ earthquakes year in, year out without noticing.
Of course, magnitude 9 earthquakes are fairly noticable, so you may prefer a increase in a lower magnitude quake. But it takes 52,400 magnitude 6 earthquakes to release the same amount of energy as a Lake Toba erruption (magnitude 9.15 equivalent). So for a 0.2% increase in the background rate, you have to imagine that the rate of magnitude 6 earthquakes has increased, year in, year out, by 52,400 per annum over their normal background rate. Unfortunately, the background rate of magnitude 6 earthquakes measured over the twentieth century is 100 to 150 per year. Ergo your theory requires the rate to have risen to that level from a prior rate of -52,275 magnitude 6 earthquakes a year.
Quite frankly, all these alternative proposals are ridiculous.
And what is even more riduculous given that for the change in geothermal flux to be significant relative to the change in heatflow due to the enhanced greenhouse effect, you need a change in the background flux, not of 0.02% but of 10,000%. Again, we would have noticed. Indeed, the measurements which determined the current background rate would have picked it up. The greatest increase in background rate of geothermal flux you can concievably argue for is the current rate of flux of 0.09 W/m^2. And that is insignificant compared to the change due to the change in greenhouse gases, and relies on the unphysical assumption that there was no geothermal flux prior to about 1850.
In short, that you provide no numbers in support of your contention is no accident. That is because if you look at the numbers, you see immediately that your theory cannot be true.
-
pjcarson2015 at 10:23 AM on 21 March 2016Heat from the Earth’s interior does not control climate
Planet Earth does its own (analog) calculations which it always gets right. It doesn’t have error bars. (Man is playing catch up in the calculation stakes.) Earth’s input and output energy always balance to zero when the Earth’s temperature is at equilibrium. As I wrote previously, this is the circumstance when Earth’s 47 TW interior heat comes into significance.
As you say, the rate of Earth’s heat production is a very slowly changing property – too slow to be of any significance in these discussions – but its rate of release does vary significantly, principally with earthquakes and volcanism, both of which known to change.
Moderator Response:[PS] Please note the Comments Policy statement on "No sloganeering". If you wish to make assertions about "Planet Earth does its own calculations", or that geothermal rate can change in a way that could be significant to climate, then you must back your assertions with data and/or papers. You beliefs fly in the face of known data and laws of physics. Please provide the basis for such beliefs.
-
Tom Curtis at 06:53 AM on 21 March 2016Tracking the 2°C Limit - February 2016
Rob Honeycutt @16, thanks for the headsup.
Mann's tweet says:
"Feb GISTEMP +1.35C 1951-80 baseline is ~2C relative to true pre-industrial baseline"
Citing the GISTEMP global anomally as he does, that should indicate that he is talking about global temperatures, but it turns out he is not.
Specifically, when you follow up to the referenced article in the Huffington Post, he makes use of information cited in two other articles, one peer reviewed, and one from Scientific American, both of which deal exclusively with Northern Hemisphere temperatures. Indeed, he is quite clear about that in the Huffington Post article.
In the end, the data he shows is used in this graph from the Scientific American article:
This is derived from a combination of temperature data, including BEST NH land back to 1750, forcing data and an energy balance model. Exactly how these are combined to produce the preindustrial temperature is unclear to me. He does provide Matlab code, but as I do not have Matlab, I am unable to read the code. He writes (Huffington Post):
"There are a number of things to note from Fig. 3. First of all, using the more appropriate 1750-1850 pre-industrial baseline, we see that the Northern Hemisphere average temperature (gray squiggly curve) has already warmed nearly 1.2C. Temperatures have exceeded 1C above pre-industrial levels for most of the past decade. So 2015 obviously won’t be the first time this has happened, despite press reports to the contrary."
The 1.2 C figure for the NH is significantly less than the 2001-2011 mean for BEST NH land relative to a 1750-1779 baseline, which is 1.67 C. That is no surprise as a land only anomaly will be higher than an ocean/land anomaly. The difference shows, however, that Mann has not simply taken the BEST land only value, although he may have taken the straight 1750 value, which was high relative to the 1750-1779 mean. I suspect, however, he has used a much better method than that and would not dispute his figure.
What I do dispute is taking the NH as an analogue for the globe. There are good reasons for doing so in the cited articles, specifically the much better historical and paleo records for the NH. But the 2 C target is for global temperatures. Because the NH is approximatley 2/3rds land, it is expected to have a higher anomally than the whole globe at 2 C, and thus 2 C NH is not equivalent to 2 C globally. Suggesting that the February 2016 temperature gives us an idea of what 2 C is like is misleading. We still have significant warming to go before we crack 2 C globally, and thus actually reach the 2 C limit.
-
Tom Curtis at 05:49 AM on 21 March 2016Heat from the Earth’s interior does not control climate
pjcarson2015 @48, had you read the original paper from which that diagram it taken, you would notice that the error margins in the absolute values of the "two large quantities" was so large as to not narrow down the difference at all (and would have indicated a much larger difference taken at face value). Consequently the authors used reanalysis products to narrow down difference in the values. More recently, the principle of conservation of energy has been used in conjunction with the much better known increase in surface heat content (mostly in the form of Ocean Heat Content) to constrain the difference between the values.
More importantly, the 47 TerraWatts is a product of radioactive decay and tidal stresses. The tidal energy is also the product of tidal stresses. Absent a major change in the relative orbits of the Earth, Moon and Sun (which has not occurred at any time over the last 4.5 billion years), and/or radical change over time of the fundamental physical constants, these cannot increase over the long term. Indeed, both will gradually decrease over the long term as the amount of radioactive material decreases, and as the Moon gradually moves further from the Earth (reducing tidal stress). In the geologically short term, tidal stresses will change at a rate proportional to continental drift as different configurations of continents induce more or less tidal friction. At the moment the northward migragion of Africa, India, Australia and South America are (very gradually) reducing tidal stress, not increasing it, and the slowness of the process will not result in changes detectable on a centenial scale. In like manner, there has been no new formation of large ignious provinces, or rapid acceleration of continental drift that would be necessary to increase heat flow from the interior (but not heat generation in the interior that necessarilly decreases).
In short, theories that the rapid recent rise in temperatures is primarilly due to factors other than the increase in greenhouse gases suffer the same impediment as another famous 'theory':
-
Tom Curtis at 05:35 AM on 21 March 2016GWPF throws out centuries of physics, climate scientists laugh, conservative media fawns
I have responded to richpender on the moderator suggested thread. I'll note here that the very existence of that thread, not to mention papers like this one, and the host of papers on the water vapour feedback effect falsify richpender's claim that "All the attention is given to CO2". It may well be true that he "...see[s] no discussion of [the effect of water vapour] on global warming", but that is not because it is not copiously discussed by the scientific community.
-
Tom Curtis at 05:27 AM on 21 March 2016Water vapor is the most powerful greenhouse gas
richpender elsewhere suggests that there is 80 times by volume more H2O and CO2 in the atmosphere.
According to Trenberth and Smith (2005) the total dry mass of the atmosphere is 5.1352 +/-0.0003 x 10^18 kg, while the mean mass of H2O in the atmosphere is 1.27 x 10^16 kg. According to the CDIAC FAQ on Climate Change, each ppmv of CO2 in the atmosphere has a mass of 2.13 Gigatonnes. The most recent IPCC report (AR5) has corrected that to a value of 2.12 Gigatonnes, or 2.12 x 10^12 Kgs. That means with 400 ppmv of CO2, there are approximately 848 x 10^12 Kgs CO2 in the atmosphere. In other words, by mass, there is approximately 1.5 times as much H2O in the atmosphere as there is water. CO2 is heavier than H2O, however.
To run the comparison of the ratio by volume, however, we need to convert from Mass to molar units. Apparently water vapour has a molar mass of 18.02 Kg per kilomole, while CO2 has a molar mass of 44.01 Kg per kilomole. It follows that there are 705.8 x 10^15 moles of H2O in the atmosphere; but only 19.3 x 10^15 moles of CO2 at 400 ppmv. The ratio of H2O to CO2 by volume is therefore 36.6, just under half of richpender's estimate.
I am not sure why richpender got the estimate wrong. Taking a median value on surface humidity, the ratio is approximately 53 moles of H2O per mole of CO2 - a value which does not include clouds (unlike Trenberth's figure). Perhaps richpender has used an area weighted mean of humidity, which would give a larger ration at the surface. Crucially to the strength of the greenhouse effect, what such comparisons nearly always get wrong is using only surface values. Because the volume of H2O in the atmosphere is closely tied to temperature, the H2O concentration falls rapidly with altitude, as seen in this graph by Science of Doom:
(Note that the graph only shows about 320 ppmv for CO2, not the current 400 ppmv).
Because of the rapid decline of H2O concentration with altitude, in the upper troposphere where the greenhouse effect has its major impact, CO2 has a far greater impact than would be expected from surface calculations alone. Indeed, CO2 contributes 20% of the total greenhouse effect, compared to 50% by water vapour and 25% by clouds. The ratio of the instantanous impact of water vapour (plus clouds) to CO2 for the total greenhouse effect is not 80 to 1, not 36.6 to 1, but 3.75 to 1.
That is the instantanious impact only. Because the concentration of H2O in the atmosphere is highly temperature dependent, if the CO2 were eliminated from the atmosphere, the amount of H2O in the atmosphere would fall drastically due to the reduced temperature. That in turn would lead to a further reduction in atmospheric H2O, and a further fall in temperature. The end of the process would leave very little H2O in the atmosphere, and a total greenhouse effect a very small fraction of current values.
Conversely, increasing CO2 increases the amount of H2O in the atmosphere, amplifying the impact of that increase in CO2. This is called the water vapour feedback.
-
John Hartz at 03:22 AM on 21 March 2016Sanders, Clinton, Rubio, and Kasich answer climate debate questions
Suggested supplemental reading:
John Kasich is no better than Donald Trump on climate change by Rebecca Leber, Grist, Mar 15, 2016
-
richpender at 01:14 AM on 21 March 2016GWPF throws out centuries of physics, climate scientists laugh, conservative media fawns
Water vapor is a global warming gas, yet I see no discussion of its effect on global warming. All the attention is given to CO2, yet there is more then 80 times (by volume) H2O versus CO2. You would think that H2O would play some role in atmospheric temperature rises.
Moderator Response:[TD] See the post "Explaining How the Water Vapor Greenhouse Effect Works." First read the Basic tabbed pane, then the Intermediate one. In future, to find discussion of a topic, enter search terms in the Search field that is at the top left of every page. Or peruse the list of myths/arguments below that Search field.
-
pjcarson2015 at 15:43 PM on 20 March 2016Heat from the Earth’s interior does not control climate
Figure 6 diagram is repeated everywhere, in fact I think it’s the only one I’ve seen. It glosses over the fact that the final output results from the difference between 2 large numbers, both of which are known only approximately, being zero when Earth is neither heating nor cooling, and approximately balances otherwise. In these circumstances, compared to a mean of zero, the 47 TW released by Earth, and perhaps even the 3.75 TW of tidal energy, is no longer “puny”.
-
Tom Curtis at 14:50 PM on 20 March 2016CO2 measurements are suspect
h4x354xOr @82, the 'eerily rigid' trajectory is effectively explained by this graph:
The ratio between cumulative emissions and cumulative atmospheric increase in CO2 is explained by the equilibrium of partial pressure of CO2 in the oceans relative to the atmosphere, ie, Henry's Law (with a couple or nuances relating to the biosphere and the deep ocean).
The variation in actual atmospheric concentration from that predicted by that curve from year to year is almost entirely explained by variation in GMST and to a lesser extent precipitation, ie, by the temperature dependance in Henry's Law and by increase or decrease of global vegetation driven by changes in precipitation.
That is the bare bones explanation. Scientists specializing in the CO2 cycle have modelled the various exchanges in detail and are able to reproduce the increase in CO2 given anthropogenic emissions and GMST changes with surprising accuracy.
-
h4x354x0r at 14:34 PM on 20 March 2016CO2 measurements are suspect
Regarding the issue of accuracy or agreement of various sampling methods or ice cores: Dendrochronology (tree rings) provide another surprisingly rich, near continuous data source that includes decipherable clues about CO2 levels. Data from dendrochronology agrees, very strongly, with data from the ice cores.
-
h4x354x0r at 14:19 PM on 20 March 2016CO2 measurements are suspect
The Keeling Curve seems almost uniquely consistent to me, compared to almost every other type of measurement in the natural world, which tends to be highly variable, regardless of long-term trends. The Keeling Curve, by comparison, is almost eerily rigid in trajectory, especially compared to variability of total carbon releases over time. It's like it's got some kind of governor that simply keeps the increase and accelleration from going any faster, no matter how much more carbon we pump into the atmosphere.
Can anyone help me understand how the Keeling Curve is so "tight", while so many other measurements have so much higher variability? Is it just because the other sinks - ocean, land, and plants - have so much more capacity that they can absorb whatever else the atmosphere can't? Even if that's a reasonable simple explanation, I still feel like I'm missing something from the 'limiter' aspect of the phenomenon.
I mean don't get me wrong, I'm figuratively poop-my-pants scared about climate change, particularly that "unprecidented rate" part. I've just recently been struck by the relatively tight natural regulation of the Keeling Curve, compared to a much greater variability I observe in almost every other aspect of nature.
Appreciate anything that might help a mostly layman level understanding of things dude wrap his head around that. Thanks! -
Tom Curtis at 12:57 PM on 20 March 2016Global Warming Basics: What Has Changed?
Just for Billev's (@1 and @3) benefit, here are the NASA GISTEMP annual values detrended to make the trend from 1880-1945 zero, and with the 1974-2002 trend line of the adjusted data shown:
First, you can see quite plainly that while 7 out of 13 years after 2002 do indeed fall below the 1974-2002 trendline, that is not because of any evident reduction in the trend. There is no more basis for assuming a reduction in the trend then there would have been in the 13 years after 1984. Indeed, with an odd number of years in the interval, of necessity the number of years below the trend cannot have equalled those above the trend. If we included 2002 so that we had an even number, we would have equal number under and over the trend. Further, with the monthly value for January and February of 2016, it is almost certain that the 2016 temperature will be at least near that of 2015, and hence over the trend, again leading to equal numbers over and below.
Further, the 1974-2002 trend is exagerated by the near record breaking La Nina of 1974/5, and the near record breaking El Nino of 1997/98. In contrast the period after 2002 has (until now) not experienced a strong El Nino and has experienced near record breaking La Nina's in 2008 and 2012/13. The slight deviation from the 1974-2002 trend in nearly fully accounted for by short term ENSO fluctuation. It is fully accounted for when changes in forcing are included. In contrast, Billev's nascent theory would predict the data, having been adjusted for the trend over a full cycle, should show a strongly negative trend after 2002.
Second, it should be noted that the data, having been adjusted for the trend over a full cycle, does not show significant similarity between phases of the purported cycle. Neither the slope nor temperature pattern from 1880-1910 matches that from 1945-1974. Nor do slope or pattern match between 1910 to 1945 and 1974-2002. Nor to the periods of the half cycles match , with respective half cycle lengths of 30, 35, 29, and 28 years. If you look at longer temperature records you also find the purported cycle disappears prior to 1880, or at best halves its period.
In all, Billev's ad hoc theory has nothing to support it. It is based on an interpretation of only 1 and a half cycles, which is a very statistically tenuous projection; is falsified in its first actually predictive interval; does not have a clear cycle and has no physical basis. Even the half periods are ad hoc, as can be seen by the fact that the 1974-2002 trend fits the data quite will for several years before 1974, which is (as previously mentioned) a near record La Nina year.
-
Digby Scorgie at 11:08 AM on 20 March 2016Global Warming Basics: What Has Changed?
Trends don't have to be linear. When I look at the graphs for shortish periods, the trend seems linear. When I look at the graphs for really long periods, the trend looks to me to be following a curve with gradually increasing slope.
-
billev at 10:22 AM on 20 March 2016Global Warming Basics: What Has Changed?
In the second chart labled NASA a line is drawn from the 1970's to the present. If the line were drawn from the temperature at 1880 to the present the line would show a warming trend through the period it covered. It would also show that there was not constant warming throughout that same perod. If you alter the line shown in the second NASA graph to pass through the temperature for 2002 (what I believe was the end of the warming period that started around 1974) you will observe that most of the temperatures after 2002 fall below the altered line.
-
Leto at 07:37 AM on 20 March 2016Lots of global warming since 1998
Bob@45,
That's great. I hadn't seen them, though I have seen people chases mirages based on variations of the third and fourth graphs - I once was silly enough to start a PhD with some scientists (bad ones, obviously) who had not looked at their raw data in a scatterplot. In the end, they were basing a major project on something like the 4th graph, and the whole thing was spurious.
-
Bob Loblaw at 06:41 AM on 20 March 2016Lots of global warming since 1998
Whenever discussion of linear regression and related statistics comes up, I always like to refer to Anscombe's Quartet. Although the four datasets result in very similar statitical results (of the ones listed on the wiki page - follow the link), it is abundantly obvious from the graphs that the four data sets are not at all similar.
I won't embed the graphs (spoiler), so look at the wiki page.
-
Hank11198 at 23:45 PM on 19 March 2016Tracking the 2°C Limit - February 2016
Ubrew12 I completely agree. Are you a structural engineer? I started engineering long before computers when everything was done with hand calculations using basic assumptions. I have trained several engineers and with the latter ones coming right out of college I’ve become very concerned that they so completely depend on the computer. I’ve had them bring me results that I could see in 2 seconds was not even close. It’s like the proverbial pulling eye teeth to get them to do a rough hand calculation.
Unfortunately some of my models are complicated plate and solid models which can be difficult to do classical hand calculations but I try to at least do exactly as you have said, run different meshes to see if they converge on a number.
-
barry1487 at 22:57 PM on 19 March 2016Lots of global warming since 1998
Kevin C,
Nice tester!
I've read the answers - I would have picked Green as having the most uncertainty, though it actually has the least.
Couple of queries/clatifications:
Blue: 0.2 +/- 0.0
Green: 0.2 +/- 0.07
Red: 0.2 +/- 0.15
Cyan: 0.2 +/- 0.12...We can see the trend in the red line very clearly, and yet it isn't significant.
1. Red is (statistically) significant isn't it? (I'm red green colour blind, so I assume the order of the values matches the order of the charts).
2. Anomalising the seasonal data would decrease the uncertainty, right?
BTW, 2nd chart looks like satellite temp anomalies, 3rd looks like Antarctic, 4th looks like global sea ice mean (+trend), though they are artificial.
...testing the statistical significance of a trend may be helpful, but it can't settle an argument.
Technically yes, though I think it can if the cofidence interval is mutually agreed upon as a robust degree of uncertainty. Ie, a centennial temp trend of 2.3C +/-0.04, 95% confidence interval, would settle that there is a warming trend for all but the most reclacitrant (eg Lubos Motl).
-
Paul D at 21:08 PM on 19 March 2016Global Warming Basics: What Has Changed?
Sigh...
Climate isn't the stock market and economics (both only constructs in the human mind), climate science is real billev. -
barry1487 at 19:18 PM on 19 March 2016Lots of global warming since 1998
Tom,
Had Dana said something more nuanced regarding 1998 sat trends I almost certainly would not have commented. As it is the statement is unambiguous, whatever the intention.
Why, in fact, in a popular discourse, are you discussing a purely academic question.
SkS straddles the line between the academic and the popular. Maybe I set too much store in what the words in the blog title actually mean (no sarc intended).
Stephen Schneider's famous quote is apropos:
"Each of us has to decide what the right balance is between being effective and being honest. I hope that means being both."
I put in my 2 cents with that idea in mind. The ongoing has been a bit more expensive!
[I read up on Carter's article about 4 years ago (written in 2006, not 2005, I believe). There's plenty of dross just like it around the skeptiverse]
-
SirCharles at 17:26 PM on 19 March 2016Tracking the 2°C Limit - February 2016
Re: Climate models
-
Kevin C at 16:59 PM on 19 March 2016Lots of global warming since 1998
The fourth curve is actually chaotic - it's a cubic spline fit through a logistic equation.
There's an error in my analysis though, because I only checked for short range autocorrelation. If I'd corrected for longer autocorrelations (specifically 12 months), then the autocorrelation correction would have corrected for the annual cycle, and given a somewhat lower trend uncertainty for the red curve in line with out expectations.
I should have used a pseudo-periodic function where the period varies over time. The trend would still be plainly apparent to the human eye, but would still give a large uncertainty even with long range autocorrelation correction.
I've got another example where the trend uncertainty tells us the wrong thing which doesn't depend on uncertainty - I'll try and work up some pictures later.
-
ubrew12 at 12:45 PM on 19 March 2016Tracking the 2°C Limit - February 2016
Hank@20 said "from a structural engineering perspective... [the models] better at least be close or bad things happen" I typically make 3 models of anything I'm tasked with understanding: a full 20k-2000k finite-element or difference model using appropriate software, a 100 node model performed with a spreadsheet, and a 1-5 node model calculated 'back-of-envelope' by hand. Always tie your calculations back to something you did by hand. It confirms the most critical assumptions in your calculation and prevents embarrassment. A 2000k-node model can be right in its particulars, and yet spectacularly wrong in its generals.
"anything can be fit with a Fourier series" John von Neumann: "with four parameters I can fit an elephant, and with five I can make him wiggle his trunk." Earth does not exist outside Physics. If your friend cannot tie his math modelling back to the Physical fundamentals, thats a 'red flag'.
-
billev at 09:51 AM on 19 March 2016Global Warming Basics: What Has Changed?
The trend may be up but global temperature has not risen steadily since 1880. There were pauses in temperature rise from about 1880 until about 1910 and from about 1945 until about 1974 and now since about 2002. If this pattern continues then there will only be a steady rise in warming from about 2030 until 2060 and then again from 2090 until 2099 in the current century. The warming trend will continue, however, until about 2350 if the climatologists are correct in saying the Earth began to experience a 500 year warming period around 1850.
-
Tom Curtis at 07:18 AM on 19 March 2016Lots of global warming since 1998
Kevin, am I correct in assuming the fourth curve involves the superpostion of 3 or 4 sine curves of different amplitudes plus linear trend? Also, is the noise model for the second curve white noise or red noise?
I am surprised at the ordering of error margins (obviously), but not surprised that I guessed wrong given that I was trying to eyeball the 95th percentile range of variation from the trend. Although 2 and 4 have the greater peak amplitude, over much of their range the amplitude of the residual is much less than in 3.
-
Kevin C at 07:04 AM on 19 March 2016Lots of global warming since 1998
Tom: I adjusted for autocorrelation in every case, but I thought that was too much detail!
-
Tom Curtis at 07:03 AM on 19 March 2016Lots of global warming since 1998
Kevin C @36, I would guess that all have the same trend by construction, although the second (green) looks to have had its trend reduced slightly by random variability. Clearly the first is statistically significant, while the following three are not, which look to have approximately the same standard deviation by construction, with the linear trend being approximately 0.75% of the error margins. Not sure if they would have the same error margin useing an autocorrelative model though. If not, order as per rkrolph.
-
Kevin C at 06:59 AM on 19 March 2016Lots of global warming since 1998
Well done for giving it a go! As you may have guessed, the examples were set up to expose a misconception, and so the answers are rather counterintuitive. Here are the actual trends and uncertainties:
Blue: 0.2 +/- 0.0
Green: 0.2 +/- 0.07
Red: 0.2 +/- 0.15
Cyan: 0.2 +/- 0.12So something interesting is going on here. All the trends are the same. You correctly identified the straight line as being most significant. But you identified the trend+seasonal cycle (red) as next most significant when it was least significant. And the trend+noise as least significant, when it was second.
We can see the trend in the red line very clearly, and yet it isn't significant. It's very hard to spot the trend in the green line, and yet it is. What's going on?
It's easy, on the basis of common discussions on the web, to imagine that linear regression is some magical tool which can spit out objective answers concerning the existance or non-existance of a trend. But it isn't.
Linear regression is a model which we can use, and which may help our understanding. However if it is used blindly, then it can also lead to completely invalid conclusions. Most importantly are the answers it can give, which can be characterized roughly as:
- Yes, there appears to be a trend, assuming that the model is correct. Or...
- No trend could be detected, either because there is no trend, or because the data are inadaquate, or because the model is inaqequate.
In the latter case we don't know whether there is no trend, or whether there is a problem in the model or the data. Even in the former case, we may detect a non-existant trend if our model is wrong. So the results are in no way objective, and are contingent on a number of other factors.
In other words, testing the statistical significance of a trend may be helpful, but it can't settle an argument.
So let's go back to the examples and try and understand them some more. Linear regression involves an implicit model: that the data consist of a constant, a linear trend, and noise. The green data fit that model exactly. And the noise is not too large, so the calculation correctly identifies the presence of a trend
In the other two cases, the model isn't really representative of the data. In the case of the seasonal cycle, the deviations from linearity aren't noise-like. If we used a better model which included a seasonal cycle, the trend would have come out clearly. Linear regression tells us it can't detect the linear trend because the data aren't linear, even when the trend is totally clear to us.
Of course the application to real data arises because we know the real data also contains contributions which are not remotely linear - in particular very strong El Ninos at either end of the period (and indeed significant La Ninas just in from them). Since this non-linearity is not accounted for in the linear model, it inflates the uncertainty such that the underlying trend is not detected. (That's not a stupid - if we didn't happen to have El Ninos at both ends of the study period, they would also distort the trend as well as inflating the uncertainty.)
-
Hank11198 at 06:44 AM on 19 March 2016Tracking the 2°C Limit - February 2016
Ubrew12:
Thanks for the links. There were helpful and interesting.
No my denier friend has never given any projections by skeptics. I’ve even pointed out that they have been predicting a downturn in temperature for 20 years and it hasn’t happened yet. Those emails are ignored. Although he has sent me some curve fitting that was used to show ‘natural change’ (never any definition of what ‘natural’ is) and then subtracted from the actual to show that climate sensitivity is very low. I’m sure you have seen those. A few natural cycle sine curves with the amplitude and frequency adjusted and anything can be fit with a Fourier series.My friend is very intelligent. I’ve even asked him how he picks between which science fields he accepts and which he rejects since he rejects very few but just get a convoluted answer about it has to be based on the evidence of all things! He accepts the temperature is rising and CO2 has some effect but not enough that even if we did something it would matter much or that it’s going to get that bad. I’m pretty sure the problem is his conservative views about the government getting involved in this with taxes. You know the story. I don’t think I will ever convince him in this lifetime but I actually enjoy learning something new and our debates require me to study the subject in order to give factual and rational responses. So that’s the benefit I get from our debates. I’m a long way from knowing a lot about the subject but I can follow the math and science part enough to know I need to be worried about my children and grandchildren.
I will respectively disagree with you on comparing projections with models. I do know that models are incorrect as I use finite element almost every day in my work. However at least from a structural engineering perspective they better at least be close or bad things happen. On occasion I have the opportunity to verify my designs by observations from the field. It happened just a few weeks ago when a structure had to be loaded for testing and we took deflection readings while it was loaded. The deflections were within 10% of what my model predicted so I was ecstatic. It verified my model was created correctly and was also a step towards validating the entire method of the software finite element program I was using. I think projections reasonable close to matching the models is a powerful argument that the science is correct and we need to do something about global warming.
-
Rob Honeycutt at 06:30 AM on 19 March 2016Tracking the 2°C Limit - February 2016
Here's an interesting comparison I just found...
-
Rob Honeycutt at 06:23 AM on 19 March 2016Tracking the 2°C Limit - February 2016
There was a lot of heat coming out of the ocean in the mid latitudes as well...
Whereas I believe the 1998 El Nino was primarily limited to the equitorial Pacific.
-
DennisMyers at 06:12 AM on 19 March 2016How Exxon Overstates the Uncertainty in Climate Science
Maybe these chart makers need to standardize their assumptions and only make charts based on current laws, or current pledges. Showing the difference between what has been pledged, and what has been implemented only.
-
rkrolph at 03:53 AM on 19 March 2016Lots of global warming since 1998
Ok Kevin, I will give my visual assessment. I would say that all datasets show a trend. I would guess that if you put these datasets into an excel spreadsheet and applied linear trend lines to each, the trendlines would all match closely to the slope of the top blue line. In order of most to least statistical significance I would say blue, red, cyan, and green.
-
wili at 03:42 AM on 19 March 2016Tracking the 2°C Limit - February 2016
I keep reading reports about how climate scientists are responding to this February peak with words like ‘shocking,’ ‘stunning,’ and just ‘wow.’
I agree, and I know they expected _some_ increases in an El Nino year. But now I want to know what the science didn’t anticipate that has made this monthly peak so high.
Is it just the super strong El Nino on top of GW? Or aret there carbon, or albedo or other feedbacks kicking in? Are we losing our 'aerosol umbrella'? Perhaps calculations of climate sensitivity have been too low? Do we have any clue at all why the reading for February is so much higher than anticipated?
-
wili at 03:30 AM on 19 March 2016Worst Mediterranean drought in 900 years has human fingerprints all over it
Would it be correct to see this as part of the predicted shift of the Sahara and Arabian Deserts north, just as the tropical zones are shifting poleward?
-
Kevin C at 01:52 AM on 19 March 2016Lots of global warming since 1998
I think there is possibly a more fundamental issue here than the semantics, which relates to the meaning of 'statistical significance'. Statistical significance is not an indicator of whether there is a trend or not, it is an indication of whether a trend can be detected by a particular approach under a given set of assumptions.
So what is being identified as a 'lack of statistically significant trend' in some datasets does not mean that there is no trend. It means that under a particular set of assumptions using a particular method of diagnosing a trend, the diagnosed trend could not be confidently distinguished from zero. However, the assumptions affect the answer. These include features such as the nature of the noise, the noise model, the variable used, the presence of a residual annual cycle and other factors. And the assumptions in the method have no bearing on whether or not there is a trend in the data, only on whether we diagnose one.
To illustrate the problem, would anyone care to speculate on which of the following datasets contain a trend? Is the trend significant in each case?
-
Rob Honeycutt at 01:08 AM on 19 March 2016Tracking the 2°C Limit - February 2016
Tom Curtis... Mann also tweeted several comments related to this anomaly being 2°C over preindustrial.
-
ranyl at 21:29 PM on 18 March 2016Tracking the 2°C Limit - February 2016
Basically Feb has peaked at ~2C above pre-industrial, a level considerred so dangerous to geopoltical stability that even the politicians have made honest pledges that they don't want to go above it.
What carbon budget??????
Like we can seriously put more GHG into the atmosphere safely, what a load of nonsesne.
2015 was also the highest yearly CO2 rise at ~3ppm.
Wars are raging and Wars cost lots CO2 (manufacturing all those nice weapons) and cause mass environmental damage.
And the highly elite driven system humankind is under isn't intending to change anything any time soon and when pressure falls on the basic necessities and the weather ravages homes and crops, more wars and migrants are inevitable.
The pope has already called the plight of the Syrian refugees who are average kind people escaping tyranny and terror an invasion of a holy nature into Europe.
Antartica is melting quickly, Greenland too, the Arctic sea ice is vanishing rapidly and the permafrost is melting, forest fires raging and peaks lands burning all relaeasing more GHG than expected in all the policy makers models.
???
Isn't time to transform everything we do?
Impossible???
Or does anyone know a place of sanctuary for those who care to escape to?
-
Tom Curtis at 19:29 PM on 18 March 2016Lots of global warming since 1998
barry @29:
"Once again you are presenting a different argument to the one I made."
No! I am presenting my own argument in relationship to the claim that you made. If you were not so busy trying to call strawman or agreement, you might have learnt something - possibly realized that what you already knew but weren't thinking about made your original claims misleading and obfusticatory.
Very briefly, once more, had Dana said:
"The line which has the least sum of the squared differences for the satellite data has a positive slope"
and you objected:
"It does not. Not to statistical significance."
Your claim would be simply false, representing a category error. You would have interpreted a claim about the data as a claim about reality.
Conversely, had Dana said:
"That the line which has the least sum of the squared differences for the satellite data has a positive slope shows temperatures in fact rose over the interval"
and you had objected:
"Not it does not, for the error margin on the slope of the line includes zero"
you would have been correct.
Dana, of course, said nothing so precise as either of those statements. He wrote:
"The satellites show warming since 1998[.]"
The question is then, which of the more formal statements above most closely conveys Dana's meaning. It think it is the first, and clearly the first. That is (firstly), because it was made to a popular audience, and popular audiences do not trouble themselves with error margins. Therefore, it is invalid to interpret the claim in a way that requires understanding of error margins to be understood unless dictated by necessity. It is also because Dana would, if challenged as to whether or not the satellite record by itself shows the warming to be real (ie, statistically significant), be genuinely puzzled as to why you would ask that question to the exclusion of the majority of the data, and to the exclusion of physics. Why, in fact, in a popular discourse, are you discussing a purely academic question.
Perhaps you are confused on this point because the 'no warming since 1998' has gone through several incarnations. It started in 2005 with Bob Carter pointing out that the annual temperature of no year to 2005 was higher than 2005 (totally ignoring trends). Overtime, it started bringing in discussion of statistical significance, but over the last year the argument has reverted to Carter's fraudulent basis. In particular, the deniers have countered the fact that 2014 and 2015 were record setting years in the surface record by pointing out that they were not in the satellite record. And, that simplistic argument no longer works. Indeed, given that Spencer and Christy know about the lag in satellite temperature response to ENSO as well as anybody else, they knew it was only a matter of time before that argument would stop working when they made it. That is, they knew the argument to be deceptive as they spoke.
-
Tom Curtis at 18:53 PM on 18 March 2016Tracking the 2°C Limit - February 2016
It turns out that Michael Mann's claim was made in an email, the rest of which has not been quoted. As such, I can make no informed comment on what he meant. Supposing TonyW @13 is correct, however, then using GISTEMP, the February 2016 anomaly relative to 1951-1980 was 1.9 C, making it 2.18 C greater than the 1880-1909 average.
-
bozzza at 18:09 PM on 18 March 2016Lots of global warming since 1998
Let's not get rid of the term 'statistical signicance' because it will mean the deniers keep trying to undermine the science by simply saying uncertainty exists rather than providing their own predictions!Contrast draws the eye and the wolves in sheets clothing will remain easier to spot be keeping the term 'statistically significant' by my rationale... That's what you call a cunning strategy: "..shhhhhhhhh!" -
Glenn Tamblyn at 17:40 PM on 18 March 2016Volcanoes emit more CO2 than humans
Ybnvs
Just a reaction to several things you have said:
"My interest in this topic came from hearing environmentalists voice the absurd notion that mankind is responsible for climate change."
Why absurd? This sounds like an 'Argument from Incredulity'. Surely whether humans are responsible for climate change depends on two factors, both ultimately quantitaive.- How big does some influence have to be to impact Climate?
- How big an influence does humanity have?
Both these questions can be explored through measurement, observation and quantification. Just using the label 'absurd' is a cop out.
"Decades ago I noticed that the hockey stick graph actually showed temperature increasing before CO2 increased, but it was only recently that I heard someone else point that out"
Then you need to check your sources. The 'temperature increased before CO2' argument applies to the ice cores that cover time scales 10 to 80 times longer than the 'hockey stick'. The 'hockey stick' does not reference CO2 levels at all. And the ice core ecord is more complex than that. Different ice cores show different raltionships.
'and I'm suspicious of demagogues whose solution to a perceived problem is monetary. '
So what does this have to do with the science? If the science says that we need to reduce CO2 emissions, that is not as as such a monetary question. Implementing it might be but it is a logical fallacy to say that 'Problem A requires a solution that looks like B and since I don't like B, A is not real'.
"I understand that my thoughts will be criticised as simple (Occums razor) but my assertion that we can't establish cause and effect is rock solid."
Sorry, this is illogical. If your assertion is rock solid then you are claiming that you have established an alternative cause and effect relationship. Also there is a fundamental distinction between saying we 'haven't' established a cause and effect relationship and saying we 'can't'. Finally, if you think a cause and effect relationship hasn't yet been established, you need to back that up. -
TonyW at 17:35 PM on 18 March 2016Tracking the 2°C Limit - February 2016
Regarding Mann's 2C call, I believe it is referring to the northern hemisphere only. As a whole, Mann calculated NH Feb as 1.95C above pre-industrial though I've seen references to a period during Feb when NH was above 2C. Sorry, I don't have the links to hand. -
One Planet Only Forever at 15:23 PM on 18 March 2016Lots of global warming since 1998
barry,
There is a difference between 'being particular in an effort to help others better understand something' and 'being particular in an attempt to misrepresent something'.
I am sure you understand that. Misleading political marketing point makers understand how to make 'their points' to drum up popular support for something they understand does not deserve to be popular.
-
Ian Forrester at 15:08 PM on 18 March 2016Lots of global warming since 1998
The problem is that people who have no knowledge of statistics believe that "not statistically significant" means "no warming". That was shown with the interview with Phil Jones a few years ago. Just use a p value, knowledegable peope know what it means and others can ask. Let's get rid of the term "statistical significance" since it allows deniers to mislead.
-
barry1487 at 13:22 PM on 18 March 2016Lots of global warming since 1998
As for being a stickler, blame Tamino for my insistence. He's my stats 'guru.' And comments above kind of make my point. We criticise skeptics in fine detail based on statistical analysis, including stat sig. It surprises me when we start suggesting that this is pedanticism.
-
barry1487 at 13:20 PM on 18 March 2016Lots of global warming since 1998
Ian Forrester,
However, I do get upset when people start quibbling about “statistical significance" when I'm not sure if they know what it means. Firstly, it does not mean that there is no trend if it does not meet statistical significance criteria.
I wrote above:
"RSS - No stat sig warming since 1992: 24 yrs
UAH - No stat sig warming since 1995: 21 yrs
This doesn't 'prove' no warming, of course. That's not how the null hypothesis works. But it does mean you cannot say there has definitely been warming. Not according to those data sets. A broader lok at the climate system tells a different story."
As others have pointed out, starting from 1998 is a cherry-pick. But that was the cherry-pick raised in the OP, and the claim that followed re satellite records was statistically illegitimate.
-
barry1487 at 13:13 PM on 18 March 2016Lots of global warming since 1998
Tom,
If you think there is a contradiction between the claim that "the likelihood that 2010 was warmer than 2014 is greater than 0" and the claim that "the probability that 2014 was the warmest year on record (as of 2014) was 1...
Once again you are presenting a different argument to the one I made.
What you can legitimately say is that, given the error margins, the satellite records from 1998 to 2015 do not conclusively show that it warmed in fact from 1998-2015.
That is exactly what I said. Could you not just have agreed in the first place?
Why do we need to restrict ourselves to just one or two minimally accurate data records?
We don't. A point I also made. You're shadow-boxing. My criticism was very specific. We are, in fact, not in disagreement about any of the above points. Perhaps you think I'm attacking the overall message?
-
ubrew12 at 12:35 PM on 18 March 2016Tracking the 2°C Limit - February 2016
Hank@1. All models are incorrect (as a structural engr I'm sure you already know that). Hence, comparing predictions against reality is bound to lose, the predictions of climate scientists properly should be compared against other predictions, like those of climate deniers. Has your denier friend every offered you such a prediction by a prominant denier? If not, why not? Who told him he is automatically 'off the hook' for such information? This discusses James Hansens 1981 prediction and This discusses Wallace Broecker's 1974 prediction, and how each compared with the subsequent reality. But its not the comparison with reality that matters, but the comparison with the predictions of competing theories. If your friend is unable to field such predictions, let him know that this is actually saying something very significant about the theories he prefers.
Prev 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 Next
