Lying with Statistics: The National Climate Assessment Falsely Hypes Ice Loss in Greenland and Antarctica

Thus my maxim:
“Anyone who is giving you numerators without denominators is trying to deceive you.”

Since the total mass of ice at each location is not known very precisely, how can one measure the loss with any degree of accuracy? In other words, it seems to me the gain/loss of ice at both locations is “in the noise” — too small to know whether it is growing or shrinking, let alone to measure the rate of growth/shrinkage.

Reblogged this on Centinel2012 and commented:
We still have a few sane scientists!

How can Antarctica be losing ice if it just recently reached an all time greatest extent?

Well, quite possibly because the Antarctic extent measures the ice floating on the sea around Antarctica.
However, why is there ice loss on Antarctica? Are we perhaps seeing the downswing of a cyclic process?

Sublimation

Nice take-down. Thanks.

If all the ice were to melt from the polar region would slow the Earth 600ms. We’ve seen a 2 ms slowdown since 1970. Like most every Earth measurement though, there are numerous conflating factors.

and ~400km³ per year spread over 360 million km² is roughly 1mm of sea level rise per year.

Straight out of How to Lie with Statistics, one of my favorite books. Present a cherrypicked slice of the data on a non-absolute scale. In this case we can throw in presenting an irrelevantly short interval — one whole decade. Really? For a climate trend?
A third problem is that — once again — the data is presented without error bars! We do not know — note well — how much ice there is on Greenland or Antarctica to within one percent, and the entire range of data in either graph is less than a tenth of that, but since we do not get to see either the raw data points underlying the curves above or the error bars per point, we are left with the impression that the lines are accurate to within the width of the line and that the measurement process is not subject to any sort of systematic error that might produce the observed bias without it reflecting any sort of actual trend in the ice volume. For example, even if the curve is based on GRACE data, GRACE can very likely not accurately resolve the full volume distribution of the underlying mass — it effectively projects the mass distribution onto some presumed spherical shell.
But what shell? If Antarctica and Greenland are continuing their post-ice-age subsidence, then dense magma is flowing out from underneath them, their land surface is falling, and of course their gravitational surface field is reducing. But GRACE cannot resolve this from field reduction due to ice loss without unprovable assumptions being made, and this sort of thing has to at the very least contribute to the probable error. But if the data is presented without that error being shown, the entire “trend” in the data above could be nothing but irrelevant noise.
It is equally instructive to do precisely the same thing — plot the absolute temperature instead of the supposed temperature “anomaly” — over the last 165 years:
http://www.woodfortrees.org/plot/hadcrut4gl/from:1850/to:2015/offset:288/plot/hadcrut4gl/from:1850/to:2015/scale:0.0001
Not quite a hockey stick, eh? It would be even better (again) to add error bars to this curve — say plus or minus 1 C for the 288, another plus or minus 1 C (or more) for the bulk of HADCRUT4 gradually shrinking down to the 0.15 C they acknowledge over the last decade or so (which is IMO optimistic by at least a factor of 2, given the divergence between HADCRUT4 and e.g. GISS and the fact that HADCRUT4 doesn’t correct at all for UHI, where UHI alone is probably twice this).
One day it will be very instructive to look back at the horrendous abuses of the theory of statistics in the general field of climate science, especially in its presentation to both policy makers and the general public. Somewhere in between the actual research papers (which often do treat uncertainty at a level that would pass most referees) and the Summary for Policy Makers and CNN, error estimates and uncertainty either disappear altogether or are replaced by indefensible weasel-words like “high confidence”.
High confidence based on what, exactly? Not the central limit theorem. Not the data. Not on the excellent agreement between the models and the data. Confidence in statistics has a precise meaning, and one has to be able to back up any assignment of confidence with a computation based on the laws and theorems of axiomatic statistics. The use of the term “confidence” in the Summary for Policy Makers, on the other hand, appears to be much more akin to the use of the term “confidence” as it is used in “confidence racket”, or “con(fidence) artist”. “Trust me, I’m a Climate Scientist” seems to be a free pass to making indefensible assertions using the language of statistics but without any of the numerical backing one would require of research in any other heavily statistics-based field.
This is one of the reasons I’m active on this list. I actually understand statistics pretty well, and know what one can and cannot expect out of predictive models. One cannot squeeze blood from a stone. Nor can one solve the Navier-Stokes equations at a spatiotemporal resolution that — for cells 100 km x 100 km x 1 km, and timesteps determined by the speed of sound divided into the longest (instead of the shortest) cell dimension — is roughly 10^20 times what one expects to have to use given a Kolmogorov scale for the nonlinear dynamics of around (less than) 1 cm. Climate models are based on nothing more than the fond hope that none of the dynamics neglected or approximated on shorter spatiotemporal scales is relevant, in spite of the direct evidence (from the models themselves) that this is not so!
rgb