quote:

Originally posted by Mond:
As I said before, you really cant say anything from that graph. There is as much of a correlation pre 1980 as there is post 1980. Looking at it I would argue that there is no correlation.

Look again there is a clear inverse correlation pre - 1975 - imagine inverting one of the plots and this will become clear.

quote:

Originally posted by CobblyWorlds:
The problem is if a small trend like you show could account for the observed warming where is the response in global average temperature that correlates with solar cycles? I've read that it's detectable, but it's not as apparent as the post 1970s warming. And any effect should be quick as CRF should make cloud condensation nuclei immediately.

Hi CobblyWorlds,

I do not think this graph supports the solar theory, though I can 'see what they mean' pre-1980 - when cosmic ray flux is low temps are high and vice versa. It all goes to pot later tho'.

On the show they inverted one of the plots and scaled it perfectly to support their rubbish.

Not really. pre-1975 there is that high temp, high cosmic ray (around 1972-3). Then in 1960-1 comsic ray is increasing while temp is also increasing. Its hard to say what is happening in 1950 itself without seeing what went on before. However, you can see that there is high temperature with high cosmic ray flux (though the cosmic ray flux looks like it may be coming up from a lower temp). In fact, looing at it, the only time post-1975 where the graph loses the supposed inverse correlation is the mid 1990s, and maybe in the late 1980s. So both pre and post the where you clain the correlation is lost there is variation from the correlation. The correlation looks pretty good again since the mid 1990s, so the late 80's and the mid 90's may just be blips on the graph due to other reaons.

I meant cosmic ray flux coming up from a lower cosmic ray flux in the 1950's.

quote:

Originally posted by Mond:
Not really. pre-1975 there is that high temp, high cosmic ray (around 1972-3). Then in 1960-1 comsic ray is increasing while temp is also increasing. Its hard to say what is happening in 1950 itself without seeing what went on before. However, you can see that there is high temperature with high cosmic ray flux (though the cosmic ray flux looks like it may be coming up from a lower temp). In fact, looing at it, the only time post-1975 where the graph loses the supposed inverse correlation is the mid 1990s, and maybe in the late 1980s. So both pre and post the where you clain the correlation is lost there is variation from the correlation. The correlation looks pretty good again since the mid 1990s, so the late 80's and the mid 90's may just be blips on the graph due to other reaons.

One of the problems with this whole area of research is that people have been playing around with things until they find correlations, one of the graphs on the programme used sunspot length, but they showed the original uncorrected graph. In 2004 Damon & Laut showed that the post 1960(ish) correlation was due to 'oddities' in the calculations.

Basically without a physical mechanism any apparent correlation has to be considered very cautiously.

When you look at the ~0.2degC/decade ramping trend after the '70s it really demands a single cause, it's a bit of a coincidence for it to have a series of causes.

And for me it's stretching coincidence a bit too far when studies such as Meehl's additivity attribution attribute the post 1970s warming to the increase in CO2.

Svensmark has what I think is a reasonable theory, but to try to use it to explain the last 30 years is something he's not done in the studies I've read. And it's not a problem with my reading because people like Rasmus Benestad and Urs Neu agree he's not done it.

Rasmus B. at RealClimate: "‘Cosmoclimatology’ - tired old arguments in new clothes"
http://www.realclimate.org/index.php/archives/2007/03/c...ents-in-new-clothes/

OK, I've had a look at these URLs as posted above:
http://www.imagehosting.com/show.php/319439_CosmicTemp.jpg.html
- obviously the temperature is on an upward trend, the cosmic rays are not (they are cycling nicely). Any correlation (inverse or not) is going to be difficult with such a situation. Obviously something else is causing the temperature to go up.

A critique by Sir John Houghton (former co-chair IPCC Scientific Assessment working group 1988-2002) of Channel 4 "The Great Global Warming Swindle" is at http://www.jri.org.uk/index.php?option=com_content&task=view&id=137&Itemid=83 for example

quote:

Other possibilities such as cosmic rays affecting cloud formation have been very carefully considered by the IPCC (see the 3rd Assessment Report on www.ipcc.ch) and there is no evidence that they are significant compared with the much larger and well understood effects of increased greenhouse gases such as carbon dioxide.

Thats more accurate. While it is hard to see a trend, the fact that cosmic rays cycle within the same degree of magnitude while temperatures increase continually would indicate that cosmic rays are not responsible.

Yes, that graph from the program is no good, as it is impossible to see what is happening now. However the trend is more obvious from their graph. However, there are times when cosmic rays are on the decrease while temperature is on the increase. So if the correlation is true, then there is still variation within it. Having said that, there is also a million other reasons to explain something like that (esp over the time scale in question).

Anyway, just to make sure you know, I never said there was/wasnt a correlation (im not trying to argue for either, though benifit of the doubt would have to say that there is not a correlation). I just said that it is impossible to tell from that.

[quote]What is perhaps more significant is the magnitude of the cosmic ray flux does not change, while the temp is clearly going up. [quote]

I knew that I had said this already. Here is what I said from a previous post. While I doubt the correlation both pre and post 1975, from the graph that was provided, I had already pointed out the above (although I could have perhaps worded it, this does imply that the two are not related)

My I offer some numbers to clarify this debate. Using the data Bamber used, and very properly gave us all links to, I have calculated a Students Paired T-Test for a 20 year rolling window from 1953 to 2005. What this does is calculate the percentage likelihood that the 20 pairs of numbers are from the same distribution. I inverted the temperature anomaly, and rescaled it so that it had the same variance as the Cosmic Ray difference from the 53 year average.

Year Paired T-Test Raw Correlation
Percent Big estimation error here
53 to 72 1.1 -0.53
54 to 73 1.7 -0.46
55 to 74 1.3 -0.44
56 to 75 1.7 -0.39
57 to 76 3.0 -0.45
58 to 77 13.7 -0.35
59 to 78 29.0 -0.27
60 to 79 49.9 -0.31
61 to 80 74.9 -0.48
62 to 81 73.0 -0.56
63 to 82 50.1 -0.57
64 to 83 41.8 -0.62
65 to 84 39.2 -0.60
66 to 85 37.9 -0.55
67 to 86 61.3 -0.47
68 to 87 84.3 -0.33
69 to 88 47.4 -0.39
70 to 89 56.9 -0.44
71 to 90 54.9 -0.57
72 to 91 66.8 -0.61
73 to 92 80.1 -0.60
74 to 93 85.4 -0.61
75 to 94 56.3 -0.54
76 to 95 35.8 -0.35
77 to 96 22.3 -0.18
78 to 97 16.9 0.05
79 to 98 9.9 0.24
80 to 99 6.3 0.25
81 to 00 4.9 0.20
82 to 01 2.4 0.12
83 to 02 0.7 -0.03
84 to 03 0.4 -0.11
85 to 04 0.1 -0.13
86 to 05 0.1 -0.07


As you can see there are two periods where the 20 year test fails in a big way, at the start and the end. Even a 80% confidence is not that high. We would expect to get that once in every 5 independent data sets just by chance. (We have just over 2½ independent data sets here, (53/20))

I also calculated the 20 year rolling correlation (-1 = perfect negative fit, 0 = no relationship, 1=perfect fit). You can see that the last few 20 year periods the correlation fall toward zero. With only 20 data points the estimation error on correlation is likely to be very high, especially with a relatively weak correlation such as this, I would call anything inside -0.2 to 0.2 effectively zero. Even out side that, as the correlation is not stable one, one would need to be very careful.

In conclusion; IMHO there is at best a weak or transitory correlation between these two datasets.

But as we all know climate change is not about one driver but many. If we included, say the cooling effect of volcanoes we may see a significant improvement in the fit.

Does anybody know where a list of all the modelled effects in the latest IPCC temperature vs expected temperature graph could be found? This graph shows a good fit between modelled (expected if you like) temp and actual up to about 25 years ago. I can only assume it must include the neutron flux effect, but I would really like to know for sure.

Sorry, the three columns of data did not come out too well, they are

Date eg "86 to 05"
T-Test as percent eg "0.1"
Raw correlation eg "-0.07"

Nice work Richard.

Interesting numbers, Richard, but what is the mean cosmic ray number you used and what are the other variables in the model? Therefore, what degrees of freedom are there in the model? If you have just run temperature against cosmic rays, one would expect a high standard error because of model specification.

quote:

Originally posted by Richard Y:
My I offer some numbers to clarify this debate. Using the data Bamber used, and very properly gave us all links to, I have calculated a Students Paired T-Test for a 20 year rolling window from 1953 to 2005. What this does is calculate the percentage likelihood that the 20 pairs of numbers are from the same distribution. I inverted the temperature anomaly, and rescaled it so that it had the same variance as the Cosmic Ray difference from the 53 year average.

Year Paired T-Test Raw Correlation
Percent Big estimation error here
53 to 72 1.1 -0.53
54 to 73 1.7 -0.46
55 to 74 1.3 -0.44
56 to 75 1.7 -0.39
57 to 76 3.0 -0.45
58 to 77 13.7 -0.35
59 to 78 29.0 -0.27
60 to 79 49.9 -0.31
61 to 80 74.9 -0.48
62 to 81 73.0 -0.56
63 to 82 50.1 -0.57
64 to 83 41.8 -0.62
65 to 84 39.2 -0.60
66 to 85 37.9 -0.55
67 to 86 61.3 -0.47
68 to 87 84.3 -0.33
69 to 88 47.4 -0.39
70 to 89 56.9 -0.44
71 to 90 54.9 -0.57
72 to 91 66.8 -0.61
73 to 92 80.1 -0.60
74 to 93 85.4 -0.61
75 to 94 56.3 -0.54
76 to 95 35.8 -0.35
77 to 96 22.3 -0.18
78 to 97 16.9 0.05
79 to 98 9.9 0.24
80 to 99 6.3 0.25
81 to 00 4.9 0.20
82 to 01 2.4 0.12
83 to 02 0.7 -0.03
84 to 03 0.4 -0.11
85 to 04 0.1 -0.13
86 to 05 0.1 -0.07


As you can see there are two periods where the 20 year test fails in a big way, at the start and the end. Even a 80% confidence is not that high. We would expect to get that once in every 5 independent data sets just by chance. (We have just over 2½ independent data sets here, (53/20))

I also calculated the 20 year rolling correlation (-1 = perfect negative fit, 0 = no relationship, 1=perfect fit). You can see that the last few 20 year periods the correlation fall toward zero. With only 20 data points the estimation error on correlation is likely to be very high, especially with a relatively weak correlation such as this, I would call anything inside -0.2 to 0.2 effectively zero. Even out side that, as the correlation is not stable one, one would need to be very careful.

In conclusion; IMHO there is at best a weak or transitory correlation between these two datasets.

But as we all know climate change is not about one driver but many. If we included, say the cooling effect of volcanoes we may see a significant improvement in the fit.

Does anybody know where a list of all the modelled effects in the latest IPCC temperature vs expected temperature graph could be found? This graph shows a good fit between modelled (expected if you like) temp and actual up to about 25 years ago. I can only assume it must include the neutron flux effect, but I would really like to know for sure.

If I remember correctly, the graph did not plotcosmic rays against temperature but it did sunspot activity.

The earth's magnetic field deflects a large proportion of cosmic rays. However, the set up by cosmic rays on cloud formation might be subject to to an accumulating relationship. Thus, it may not only be magnitude but the build up. Convesely, a period of low cosmic radiation may not disperse cloud cover immmediately but start in the dissipation process. As such, models may require a lag factor in their estimation and statistical analyses.

quote:

Originally posted by Mond:
Thats more accurate. While it is hard to see a trend, the fact that cosmic rays cycle within the same degree of magnitude while temperatures increase continually would indicate that cosmic rays are not responsible.

Yes, that graph from the program is no good, as it is impossible to see what is happening now. However the trend is more obvious from their graph. However, there are times when cosmic rays are on the decrease while temperature is on the increase. So if the correlation is true, then there is still variation within it. Having said that, there is also a million other reasons to explain something like that (esp over the time scale in question).

Anyway, just to make sure you know, I never said there was/wasnt a correlation (im not trying to argue for either, though benifit of the doubt would have to say that there is not a correlation). I just said that it is impossible to tell from that.

For the T-test I used (SQRT((Var(NFt)/Var(TAt)) * (-TA)) vs (Ave(NFt) – NF)
For the correlation I used (TA) vs (Ave(NFt) – NF)
(both on a rolling 20 year window)


Where
NF is the Neutron Flux annual average, corrected for pressure
TA is the Temp Anomaly that the IPCC use
NTt, TAt is the whole 53 year series
Ave(NFt) is over the full 53 years available

Var = variance, the variance modification is just to scale things
Ave = average,
SQRT = square root.



Dev, I agree there are not really enough degrees of freedom here, or put another way, the data series are not long enough, hence my warnings of estimation error and certainty. Your point about this being a second order effect, driven by the solar wind (sun spots) is a good one, and is yet more reasons to treat these correlations with great caution.

Im not aware of any cumulative effects of cloud seeding over the year long time scales all this data is on, and its difficult to imagine one.

As for the time off set to check it, Ill do that after lunch and let you all know.

Much of the post here seems to have forgotten the effect on the ozone layer and the resultant "hole" that increased warmth. Surely any study would nered to factor in this, especially in the 80s to the mid 90s, perhaps incorporating a lag effect.

Perhaps we should also consider the the effects of CO2 and cosmic rays on cloud formation and how cloud formation can increase global warming. Furthermore, we sohould also consider lan and marine photosynthesis and the part played by CO2 here.

Clouds absorb and re-emit long-wave radiation from the Earth’s surface, hence the oven or greenhouse effect. The key here is watervapour and cloud formation. CO2 can form the condensation nucleus for clouds, so too sulpur dioxide, SO2. In the absenc of cloud condensation nucleii, water vapour can be supercooled below 0 °C (32 °F) before droplets spontaneously form.

Condesation can also occur on particles o dust soot and black carbon. I suppose that carbon dioxid which is 1.53x denser than could form a cloud condensation nucleus.

However, the combination of CO2, sunlight and water should encourage plant and phtyoplankton growth to produce oxygen. Therefore, it would lead to the postulation that CO2 will be transmutted into oxygen in the photosynthetic process. Takeaway sunlight, plants or phytoplankton and then you may have a condition where CO2 rises into the troposhere bouyed there by rising warm air to form could condensation nucleii than in turn would form clouds with the effect of a woollen blanket.

Therefore, there may be a greater effect in preserving and rejuvenating the rainforest rather than focusing on the reduction of carbon emmissions. However, carbon emmission should not be neglected as decaying undergrowth that occur more in secondary forests give of CO2 and methane, to which uncontrolled humn activity may exacerbate. Phyoplankton do release sulphates and methane sulphonic acid that would encourage could formation but the sulphates would, I think, first increase the salinity of the oceans (sodium sulphate = salt).

The cosmic ray/particle line is that they act as catalysts. I have not come across a study that examines the decay of cosmic rays, which are surmised from the detection of the by-product of cosmic ray collisions with the atmospheric particles decaying into muons. Therefore cosmic rays are detected by the presence of muons. I do not know how long it takes the kaons and pions produced from atmosheric collisions of cosmic rays to form muons, nor do I know how long a sun spot eruption would translate into cosmic rays reaching earth's atmosphere. It therefore would seem probble that there is a lag effect from sunspot activity to cloud formation.

The number of sunspots suitably lagged and examined against global warming do produce a correlation. which in the absence of investigation in conjunction with other variables must remain superficial.

In conclusion, there is a case for sunspot activity and cloud formation that woul act lik a blanket keeping in heat. There appears o be less of an argument for CO2 unless plant and phytoplankon activity are insufficient to transmute CO2 into oxygen. As such, the focus, while not throwing carbon emmission control with the slty bath water, should give greater emphasis to plants, forests and phytoplankton propagation. The phytoplankton will encourage the marine food chain.

The carbon emmission control will control atmopheric pollution that would affect the human and biological physionomy. The more efficient usage of energy sources is in it self a good perpetuating reason for improved and more efficient sources and use.

Just on a side note, would global warming and the melted waters allow a migrtion of human population to now uninhabited or less inhabited areas and will the flora and funa so react? Further, would man use technology to be able to live off the higher levels of water?

quote:

Originally posted by antidogma:
Is it reasonable to discuss any correlation in a timeframe of a couple of decades, for processes which are measured in centuries ?