By Justyna Zander, Harvard University, Cambridge, MA, and Fraunhofer Institute FOKUS, Berlin, Germany
Excerpt: Recently, the attention around climate change and global warming has been raised in the context of the models capabilities, their predictive power, and the reliability of the results. In the proceedings, a few examples of failures in climate models and their execution are pointed out, main reasons of the problems are discussed, and a novel perspective on analysis of those issues is proposed. The sources of information are extensively referenced to the multimedia publications elsewhere.
Advertisement
9 comments
Comments feed for this article
April 19, 2010 at 7:25 am
Adrian Schroeter
Review:
Problem:
Inaccuracy in prediction models in Climate Research.
Approach:
Use a crowed source approach that taps into behavior of flora, fauna and humans alike.
Comments:
I appreciate that the author provided empirical evidence that we face a problem with our prediction models. The paper however would have benefited a more detail description of the proposed methods accompanied with an example. I didn’t understand how the “Computation of Things” increases awareness among humans, but the idea that this approach helps to advise preventive measure is encouraging. Further more I appreciated that the author pointed out the value of preventive measures with respective to fatalities.
Side Note:
After reading this paper I was reminded of a talk I went to recently. It was a data mining talk explaining the problems arising with concept shifts within predictions. One problem with data is that over time the underlying distributions changes rendering older data useless and often even harmful to prediction models. These changes or shifts don’t happen gradually but sudden through certain data influencing events, which got me thinking how robust climate models are towards those shifts, which could have been introduced by the sudden climate change.
April 19, 2010 at 9:37 pm
Justyna Zander
Thanks Adrian.
Should you have any references to the talk please let me know.
April 19, 2010 at 11:01 am
anonymous
Summary: Uses “Computation of Things” (COTH) as a means to improve the quality of climate simulations.
Writing clarity: the paper could benefit from a good editing. Moreover, there were several places in the text where the choice of phrasing lead to ambiguities (“weather” and “climate”; “measurement data” and historical data”; “models” and “simulation”; “assumptions and abstraction level … belong”). Also, COTH should be described more fully so the reader can appreciate the author’s point.
Relevance: I suggest that the author have a look at the IPCC report executive summary, which talks extensively about how its conclusions are validated. Comparison with respect to this would strengthen the author’s argument.
Overall evaluation: Not recommended as currently written.
April 19, 2010 at 9:36 pm
Justyna Zander
Sorry for the writing style.
Thanks for the reference!
April 20, 2010 at 12:56 am
Lin Liu
Prediction is an interesting and difficult problem, and computation of things is a nice and emerging concept in response to the optimization problem. Although there are still great gaps in between, I see that the combination can inspire much discussion in the workshop.
April 22, 2010 at 2:51 pm
Ian Sommerville
I found this a rather difficult paper to understand, perhaps because of the problems pointed out above. I think the paper needs to be more focused to bring out the key message.
Is this a basis for discussion – perhaps, if the author can make clear to others what the research is about.
April 23, 2010 at 5:04 pm
Jorge Aranda
I had similar difficulties understanding the concepts in this paper. It is not clear to me how a “Computation of Things” can help improve climate predictions, especially in the short term (and let’s remember this is an urgent problem).
Furthermore, climate models are far more robust than this paper (and the infamous Watts Up With That blog that it cites) makes them out to be. They are not meant to predict particular snowstorms, but long-term climate trends. As Anonymous suggests, the executive summary of the IPCC report provides a good summary of the validity of climate models.
April 26, 2010 at 4:30 pm
Justyna Zander
Thank you for the comment, Jorge.
Indeed, some of the examples given in the paper are provocative.
To clarify a little bit, Computation of Things (CoTh) is meant to be an alternative complementary improvement of climate predictions. It is not meant as a replacement method.
To give an example, in case of a coming tsunami, a farmer located on his farm would be able to report about the non-usual behavior of his animals using the CoTh interface. If this kind of practice then grows to a crowd-sourced information set it could become a validated indication of something dangerous happening.
Moreover, and perhaps more interesting in this case, CoTh would be a means to predict the implications of climate change (diseases vulnerability, migration, etc) reported by an individual and meant to protect a single human being.
A nice summary of some of those thoughts is given here: http://www.youtube.com/watch?v=7XxrZAgIgMM&feature=player_embedded
Thank you again for all the comments!
April 28, 2010 at 5:54 pm
Ideas for the workshop program « Software Research and Climate Change
[...] a wider participation in climate related decision-making? (inspired by several papers, including Justyna Zander’s paper on Computation of Things, Lin Liu and colleagues on carbon footprint calculators and climate [...]