Guest post by Steve Goddard
Archimedes had his eureka moment while sitting in the bathtub. Newton made a great discovery sitting under an apple tree. Szilárd discovered nuclear fission while sitting at a red light.
There was a time when observation was considered an important part of science. Climate science has gone the opposite direction, with key players rejecting observation when reality disagrees with computer models and statistics. Well known examples include making the MWP disappear, and claiming that temperatures continue to rise according to IPCC projections – in spite of all evidence to the contrary.
Here is a simple exercise to demonstrate how absurd this has become. Suppose you are in a geography class and are asked to measure the height of one of the hills in the Appalachian Plateau Cross Section below.
Image from Dr. Robert Whisonant, Department of Geology, Radford University
How would you go about doing it? You would visually identify the lowest point in the adjacent valley, the highest point on the hill, and subtract the difference. Dividing that by the horizontal distance between those two points would give you the average slope. However, some in the climate science community would argue that is “cherry picking” the data.
They might argue that the average slope across the plateau is zero, therefore there are no hills.
Or they might argue that the average slope across the entire graph is negative, so the cross section represents only a downwards slope. Both interpretations are ridiculous. One could just as easily say that there are no mountains on earth, because the average slope of the earth’s surface is flat.
Now lets apply the same logic to the graph of Northern Hemisphere snow cover.
It is abundantly clear that there are “peaks” on the left and right side of the graph, and that there is a “valley” in the middle. It is abundantly clear that there is a “hill” from 1989-2010. Can we infer that snow cover will continue to increase? Of course not. But it is ridiculous to claim that snow extent has not risen since 1989, based on the logic that the linear trend from 1967-2010 is neutral. It is an abuse of statistics, defies the scientific method, and is a perversion of what science is supposed to be.
Tamino objects to the graph below because it has “less than 90% confidence” using his self-concocted “cherry picking” analysis.
So what is wrong with his analysis? Firstly, 85% would be a pretty good number for betting. A good gambler would bet on 55%. Secondly, the confidence number is used for predicting future trends. There is 100% confidence that the trend from 1989-2010 is upwards. He is simply attempting to obfuscate the obvious fact that the climate models were wrong.
Science is for everyone, not just the elite who collect government grant money. I’m tired of my children’s science education being controlled by people with a political agenda.
Kevin Kilty (15:41:46) :
So, for example, the PDO cycle length looks to be 60 years to me, so why wouldn’t we try to enclose it so we don’t chase our tails running after climate change that is actually cyclic?
When the 30-yr period became established there was a lot of talk about the Bruckner period, which was 35 years, so that may have played a role in the choice. As you say, one could keep discovering longer and longer periods, 88 yrs, 200 yrs, 1500 yrs, 2300 yrs ,…, and where do you stop? 30 years has turned out to work quite well, so is still with us after 140 years.
Tamino calculated 99% confidence for the trend, before he did his undocumented “cherry picking” analysis. Looks to me like nature disagrees.
http://wattsupwiththat.files.wordpress.com/2010/02/dec-feb_snow_ext.png?w=510&h=291&h=291
I would agree with Leif. A trend it might be but is it significant? Depends on your error bars and the period used to determine average. I see nothing unusual about snow in Alabama, though my Aunt in Louisiana is claiming that the world has come to an end and it is the day after tomorrow. This year’s snow line, last year’s snow line, and the year before that is anecdotal till we see frozen pond scum in summer in the bayou (and yes I know that is an exaggeration…used only to make a point and amuse, not to be scientifically correct).
Steve Goddard (10:01:23) :
don’t care about scientists playing games with statistics
I took one of the models you were talking about. It is here:
http://www.leif.org/research/Snow-Cover-1850-2100-Overlay.png
and overlaid the snow cover data [the heavy orange graph].
As you can see the model does a good job at predicting the run of snow cover. This is obvious without statistics by your argument, so your assertion that the models do poorly seems to be unfounded.
Leif,
You must be correct.
The near record high winter snow cover is right in line with model predictions of rapidly declining winter snow cover.
Not
Steve Goddard (18:27:58) :
The near record high winter snow cover is right in line with model predictions.
Yes, indeed, as the direct comparison shows:
http://www.leif.org/research/Snow-Cover-1850-2100-Overlay.png
So we’re just ignoring my response to the “Ordovician Ice Age isn’t caused by CO2” post?
ginckgo (18:46:43) :
So we’re just ignoring my response to the “Ordovician Ice Age isn’t caused by CO2″ post
No, we are not. It is recognized that unusual circumstances triggered the ice age. The Milankovic cycles are always there, so something unusual must have made the glaciations possible, and made them disappear. For the argument here, it is immaterial really what the exact cause was.
ginckgo,
I used to be a geologist and understand that a very active imagination is necessary. That explanation for the Ordovician Ice Age sounded pretty imaginative!
Leif,
Looks like the trend line and predictions are at right angles in your graph. Difficult to imagine any worse correspondence.
Steve Goddard (19:15:15) :
Looks like the trend line and predictions are at right angles in your graph. Difficult to imagine any worse correspondence.
Look at the data instead of the phony trend lines. The arrow shows what I would call a Steve-trend, from the lowest to the highest point [“do you deny that snow cover now is the highest in 20 years?”]. The heavy wiggly ‘trend’ lines come from the models and are 9-year running averages, so on your graph would include data for 1985-2014. Perhaps your trend would look like that in four years time. Anyway, look at the individual years of the model prediction. Do you deny that the arrow shows points from the lowest to the highest value so showing a definite steve-trend predicted by the model?
Steve Goddard writes, “Tamino calculated 99% confidence for the trend, before he did his undocumented “cherry picking” analysis.”
You are talking about where Tamino explains, “this is the number you get if you use the wrong analysis.” Once again, if you use a simple linear regression and a test for significance of the slope, you are not allowed to select a subset of the data to analyze based upon visual inspection of the data. If you do this, your p value will be wrong, period. (On the other hand, it would have been perfectly legitimate to randomly select a starting point for the analysis without having looked at the data. If you don’t understand why picking the starting point that looks right to you invalidates the analysis, while picking a starting point completely at random does not, you are missing a very fundamental, very important point about statistics).
What Tamino then did that was really lovely was to use a standard statistical method, “monte carlo” analysis, to determine just how much selecting the starting point throws off the standard calculation of p-value. It’s not really accurate to call it “undocumented,” considering that he provided a very cogent explanation of what he did and why.
Steve Goddard (19:15:15) :
Difficult to imagine any worse correspondence.
since you seem to have difficulties with your eyesight, here is blown-up version:
Yes, indeed, as the direct comparison shows:
http://www.leif.org/research/Snow-Cover-1850-2100-Overlay2.png
Leif Svalgaard (19:59:12) :
Steve Goddard (19:15:15) :
Difficult to imagine any worse correspondence.
That the model predictions match is of course pure nonsense and coincidence and cherry picking [as others do not], but illustrates that the variability is just too large and the time span just too short and the power of cherry picking just too strong to make any ‘statistically’ or physically significant conclusions either way. Looking out the window just ain’t science.
Leif,
Three (2003, 2008 and 2010) out of four of the highest winter snow extents in the forty-four year record have been in the last decade. Your claim that winter snow is declining is unsupportable and absurd.
http://climate.rutgers.edu/snowcover/chart_seasonal.php?ui_set=nhland&ui_season=1
Looking out the window is absolutely one of the most important methods of observing nature. Most people are aware of changes in their environment, and their cumulative knowledge is infinitely more valuable than climate model predictions.
Steve Goddard, you’ve got to be joking. I present a detailed scientific study based on lots of data showing that your conclusion based on eye-balling highly generalised graphs is wrong, and you call it “imagination”?
And Leif, it’s not immaterial to the debate what the causes were – that is what the whole debate is about. The Ordovician constantly gets thrown up as a shining example against GHG driven climate change, and yet here’s a demonstrated possibility that it’s cause was dominated by large swings in CO2. The postulated causes are indeed unusual, as are the massive amounts of fossil carbon entering the system at the moment.
Steve Goddard (20:36:27) :
Three (2003, 2008 and 2010) out of four of the highest winter snow extents in the forty-four year record have been in the last decade. Your claim that winter snow is declining is unsupportable and absurd.
don’t you at least look at my graphs?
http://www.leif.org/research/Snow-Cover-1850-2100-Overlay2.png
Look at the arrow. It that pointing downwards? Look how closely its upwards striving show that the model predicts the matching observations. I cannot magnify the graph anymore to compensate for your poor eyesight, so make an effort, please.
Steve Goddard (20:36:27) :
Three (2003, 2008 and 2010) out of four of the highest
Here are the top ten [only two (not three) out of the four have been in the last decade]:
1978 48.981
2010 47.74
2008 47.459
1985 47.046
2003 47.017
1979 46.941
1986 46.609
1967 46.479
1972 46.461
1969 46.453
Here are the bottom ten:
1968 44.652
1990 44.652
1992 44.333
1976 44.282
1995 44.137
2007 43.873
1980 43.872
1999 43.844
1989 43.436
1975 42.979
1981 40.696
Note 2007.
The model I showed you has lots of high values in the current decade and in the next [for that matter]. Nice prediction, I would say.
Lets try and be rational about this controversy:
It is true that the data on snow cover have very small errors , so one can establish a hierarchy of “peaks”. That is a fact, as with the mountains, it cannot be disputed.
The meaning attributed can though.
It is also true that if one wants to establish a trend with statistical significance (i.e. peaks are growing in the y direction) one needs the normal statistics of number of peaks to be able to give a probability number. As long as “looking out of the window” means “something may be happening”, that is fine by me, the same as with watching for sunspots : I say “aha, they are coming thicker”, maybe we are out of the trough, but statistically we need more to make sure.
Now as for the models, the plots are not worth the paper they are printed on ( or the bits that display them). Talk about cherry picking.
If the true errors on those curves were displayed, they would inundate the graph making it meaningless. It is the reason error propagation is not done.
I explain: each spaghetti line on a GCM output is picked by the intuition of the modeler, i.e. that the CO2 feedback loop is heating up the earth fast and temperatures are rising. If temperatures are rising the snow line will be retreating and so we have the plots. The wiggles can have no meaning when their true errors are imposed, which must be thousands of kilometers in this case ( by analogy to the large spread in errors of future temperature, I have seen estimates of -1C to +2.5C on a 0.2C anomaly).
So yes, we cannot talk of a significant trend now, but we can speculate and have fun. Speculation is not science, but it is the seed that might bring fruit in the future.
gingcko,
That paper was a very determined attempt to force fit the geologic record into CAGW theory.
Leif,
This decade and particularly this year has seen a lot of snow at lower latitudes. Deal with it.
anna,
Thank you for thinking like a scientist!
Leif,
There is only “good” agreement between the model and observations during the hindcast portion of the model run (up to Jan 1 2001). After that, it is clear, especially from your blown-up graph, that the model “forecasts” a downward trend in January snow cover and the observations show a distinct uptrend. You don’t need a statistical test to see that. Steve is still right – the model missed the mark.
Leif, you produced a list of top ten which included 2003, 2008 and 2010 – and then declared that there were “only two years in the last decade.” It is really pointless discussing mathematics with you.
B.D. (05:57:15) :
There is only “good” agreement between the model and observations during the hindcast portion of the model run (up to Jan 1 2001). After that, it is clear, especially from your blown-up graph, that the model “forecasts” a downward trend in January snow cover and the observations show a distinct uptrend.
I don’t know why this is so hard. Here is the relevant part of the blown-up version:
http://www.leif.org/research/Snow-Cover-1850-2100-Overlay2.png
Both the model prediction and the Steve-trend are up. What I’m saying is that for both the model and for the observations, the variation is too large and the time-span too short to draw any conclusions at all [and statistical analysis shows that clear enough for the observations]. Especially the silly claim that since 2010 was higher than 1989 there is a significant ‘trend’. The model has a similar ‘trend’ as shown by the arrow. In both cases, this means nothing.
B. D.
One would think that it is apparent that you shouldn’t be on a long term downwards trend and simultaneously be near a record maximum. Yet we have people here arguing that point. They argue that this is not a “statistically significant” upwards trend, and at the same time argue that the trend is downwards.
http://wattsupwiththat.files.wordpress.com/2010/02/north_american_winter_dec-feb_snow_extent_1989-2010.png
It is truly Alice in Wonderland