Thank you Kenneth and Mark for continuing the discussion.

You are right. It differs from a field to the other. Let me organize my answer in a general way. As you have mentioned, it is a general question, so a general answer suits it.

For a successful uncertainty assessment, we need to address two issues as precise as we can:

1- The goal of the study; i.e. what are we going to assess its uncertainty, which parameter, to what accuracy, etc.

2- The sources of the uncertainty in the case study.

I did not find other general item to add! My field of study is limited mostly to petrophysics, and I like to develop a general uncertainty assessment algorithm in petrophysics. Maybe you can add some other important items from the viewpoint of your professions, e.g. geochemistry!

I accept all the uncertainties, accompanying geo-related studies. In fact, the earth is like an alive creature! Full of unpredictable phenomena. And we, as humans, with limited observations, try to model (i.e. simplify) the reality.

By the way, I just emphasize the importance of UNC-assessment, and hopeful to find some general ideas for having a tentative tool for using. E.g. "the scale" of study, not important time- or coordinate- scale, are controlling factors in defining the goal of study. In addition, economic risk in engineering projects is essential also.

I do not search for a satisfactory answer. I seek for a pleasurable discussion for getting some insights through the ideas and experiences of each other :)

Uncertainty assessments are well-developed where they need to be – that is, in the small, but important branches of the discipline where the goal is to make predictions about the future. So, for example, resource estimation, geohazard characterization, earthquake prediction all have uncertainty assessment as a critical focus…however I would leave it to the experts in these fields to offer their thoughts on whether a better job could be done, if we were to use a ‘naturally’ predictive science, such as physics or chemistry, as a yardstick.

I agree with Kenneth and Mark that statistical rigor, including uncertainty analysis, is a sine qua non, wherever the data and measurement techniques being used, allow. However it is worth pausing to reflect on what earth scientists frequently do with their precise, accurate, error- and uncertainty-characterized datasets, namely to engage in untestable speculations about the geological past whose unquantifiable uncertainties dwarf those of the underlying data.

That is not a criticism, it is simply the nature of how geological research works in practice. The goal of the vast proportion of the earth sciences is, of course, to make retrodictions (e.g. Kitts 1976), or predictions about the past, and since we can never (in a formal sense) be truly certain of either the initial or boundary conditions of any past geological process, we can never escape the indeterminacy that has prompted some (but not me) to claim that geology as a discipline, is unscientific. I leave it to others to comment on whether uniformitarianism offers geologists a philosophical fig-leaf.

Kitts' paper has a good discussion on how and why geological uncertainty arises and the ways in which it makes geological enquiry distinctive.

P.S. Note to Kenneth – If you weren't already aware of them, check out these other two references regarding Kelvin’s estimate of the age of the earth which draw slightly different lessons compared to the conventional anecdote you refer to.

Kitts, David B. "Certainty and uncertainty in geology." American Journal of Science 276.1 (1976): 29-46.

Richter, Frank M. "Kelvin and the age of the earth." The Journal of Geology (1986): 395-401.

England, Philip, Peter Molnar, and Frank Richter. "John Perry's neglected critique of Kelvin's age for the Earth: A missed opportunity in geodynamics."GSA TODAY 17.1 (2007): 4.

Best Regards

Graham

Merci Graham for the references and expressing your idea.

The references are far from my specialty (except the 1st one) but your text persuaded me to go through them tonight!