Dear all, It is fairly evident to me, and I think most glaciologists (though I cannot speak for others), that various scaling concepts and parameterizations are better than having no data on by far most glaciers in the world, or having some data on lots of glaciers in terms of length but being unable to translate a one-dimensional measurement (or measured change in that variable) to three dimensions (plus time in case of multi-temporal measurements). Obviously there are hard limits in what can be done using scaling, and scaling is no substitute for direct measurements of things like mass balance and volume; but it is a complement to such direct measures, and this complement has the effect of making more truly global and comprehensive our monitoring of the world's glaciers-- with the necessary underpinning that we must continue/expand real monitoring by intensive ground-based methods. GLIMS and the direct measurements we produce would have fairly little use in the absence of some sort of scaling or the capacity later to apply scaling. So.... there is no need for as a fratricidal argument, but there is a need (as there has been all along for decades) for more in-situ measurements, improved approaches to scaling, wider application of scaling to existing data, and more intensive remote observations (and analysis of existing images) that can be scaled. And there is a need for responses to Will's questions, below. I encourage at least people in the GLIMS community (others, too, of course) to formulate thoughtful responses to all the points that have been made by the discussants, assuming that the argument aspect is a non-issue (thus not in need of further comment) and all the substance matters are issues. My apologies to those who are on both mailing lists used, but I want to be sure that the substance of the discussion is relayed fully to the GLIMS subsector of the glaciology community. Perhaps GLIMS respondents can mail directly to Will Harrison, Mark Meir, David Bahr, and Wilfried Haeberli, so that there is not rabbit-like reproduction of comments, which I hope will be numerous but not double-mailed (to some people) like my message! Then perhaps those people can sort through the comments and decide on something with consultation with IGS officers. Just a thought. Cheers, Jeff Kargel -----Original Message-----
From: "William D. Harrison" <Harrison@gi.alaska.edu> Sent: Aug 20, 2007 10:40 AM To: Mark Meier <Mark.Meier@Colorado.EDU>, Martin Sharp <martin.sharp@ualberta.ca>, Waleed Abdalati <waleed@icesat2.gsfc.nasa.gov>, MMM <magnus@igsoc.org>, kargel@hwr.arizona.edu, Andreas Kääb <andreas.kaab@geo.uio.no>, Martin Truffer <truffer@gi.alaska.edu>, Regine.hock@gi.alaska.edu, Mark Dyurgerov <mark.dyurgerov@natgeo.su.se>, Jo@av-in2.su.se, Jacka@av-in2.su.se, jglac@bigpond.com Cc: Wilfried Haeberli <haeberli@geo.unizh.ch>, "Bahr, David" <dbahr@regis.edu> Subject: Re: volume/area/thickness scaling of glaciers
Dear List,
I think there is plenty to be done on area-volume scaling yet, both concerning the data upon which it is based, and its theoretical underpinnings. The only question in my mind is whether it should be done through normal channels, or whether it needs a more visible format as Wilfried suggests. I am inclined to favour the latter. Does he suggest a meeting with Annals coverage?
Regards to you all, Will
At 12:35 PM 8/16/2007, Mark Meier wrote:
It has come to our attention that Wilfried Haeberli has suggested to you that the Journal of Glaciology host a "publicly accessible discussion about scientifically sound and realistically practicable concepts for estimating thicknesses and volume of unmeasured glaciers." We do need, of course, continued research on this issue, and we know that several investigators are working on it.
However, we do not see the need for public discussion (or argument?) about this. Dr Haeberli seems to be trying to make scaling into a debatable issue, calling it "problematical", "violate[s] the most basic principle of correlation statistics", "give a misleading impression", "may even lead to delicate questions about honesty." These accusations are inflammatory and are based in part on a serious misunderstanding of the method - we know of no one who, when using this scaling, conflates the independent variable with itself, as Haeberli suggests.
Scaling has had a long and productive history in hydrology and in geophysics. In glaciology, area/thickness or area/volume estimates using power-law scaling in a statistical sense began with Chen and Ohmura and several Russian authors in the 1960s ; Bahr in the late 1990s showed that the power-law formulation has its basis in glacier physics. Recent work, such as the paper by Radic et al (Ann. Glac. 46) further establishes the usefulness of this approach.
Obviously, the scaling works best with aggregates because individual glaciers have different characteristics, and the variance from an averaged power-law is higher when regressing thickness vs area than with volume vs area. In our most recent paper, by Meier, Dyurgerov, Rick, O'Neel, Pfeffer, Anderson, Anderson, and Glazovsky, Science, Aug. 25, 2007, we used scaling and "estimated that the error in calculating thicknesses and thus volumes from area values is of the order of 25% for global aggregates but far greater, on the order of 50%, for individual ice masses."
Refining these estimates and incorporating them in other methodologies using new quantitative information will always be useful. But this scaling tool is well established and there is no longer any need to have a public discussion of its basic validity.
Mark F. Meier, University of Colorado David B. Bahr, Regis University