I think the distinction between design model and evaluation model is important; the latter serves as the proxy for the real world, and verifying that a strategy designed in one (or more) models is robust when implemented in a “blind test” on some other model is essential.
In that light, one multi-model framework that would be valuable would be to test a control strategy in not one, but multiple models, and evaluate how well it works (or whether it works). That isn’t quite Andrew’s suggestion, which I take to be seeing which control algorithm works best, but evaluating multiple algorithms could also be useful eventually. doug From: [email protected] [mailto:[email protected]] On Behalf Of Kravitz, Ben Sent: Thursday, February 05, 2015 2:10 PM To: Andrew Lockley; Ken Caldeira Cc: geoengineering; Doug MacMartin Subject: Re: [geo] Re: Assessing the controllability of Arctic sea ice extent by sulfate aerosol geoengineering Hi Andrew - I'm really not sure what you're getting at with this suggestion. But more important to me is that I have a real problem with using the metaphor of a TV show. When I discuss the idea of controlling the entire planet's climate, I think it's very important that I not appear flippant or whimsical. When I write about a technology that could potentially affect billions, people read it, and not all of the readers are people who know me. More than that, GeoMIP is not about finding which is the best participant – it's about collectively using the combined knowledge and effort of many models to understand robust model response. The models don't compete. I don't see why the control algorithms should compete either. Perhaps I've misunderstood what you're trying to say. Best, Ben ______________________________________________________ Ben Kravitz Postdoctoral Research Associate Atmospheric Sciences and Global Change Division Pacific Northwest National Laboratory P.O. Box 999, MSIN K9-24 Richland, WA 99352 Tel: (509) 372-6846 Fax: (509) 375-6448 [email protected] From: Andrew Lockley <[email protected]> Date: Thursday, February 5, 2015 at 10:48 AM To: Ken Caldeira <[email protected]> Cc: geoengineering <[email protected]>, Ben Kravitz <[email protected]>, Doug MacMartin <[email protected]> Subject: Re: [geo] Re: Assessing the controllability of Arctic sea ice extent by sulfate aerosol geoengineering Is there an opportunity to create a GeoMIP for control systems? It could work a bit like the TV show robot wars, with algorithms fighting each other to control the climate most effectively. A On 5 Feb 2015 17:13, "Ken Caldeira" <[email protected]> wrote: Just to be disgustingly self-promoting, I will also point out a couple of papers that look at optimization outside of the control context: As far as I know, the Ban-Weiss and Caldeira (2010) paper was the first to treat solar geoengineering as an optimization problem: http://iopscience.iop.org/1748-9326/5/3/034009 George A Ban-Weiss and Ken Caldeira 2010 Environ. Res. Lett. 5 034009 <http://dx.doi.org/10.1088/1748-9326/5/3/034009> doi:10.1088/1748-9326/5/3/034009 Geoengineering as an optimization problem Also, see: http://www.nature.com/nclimate/journal/v3/n4/full/nclimate1722.html Nature Climate Change 3, 365–368 (2013) doi:10.1038/nclimate1722 Received 09 May 2012 Accepted 18 September 2012 Published online 21 October 2012 Management of trade-offs in geoengineering through optimal choice of non-uniform radiative forcing · <http://www.nature.com/nclimate/journal/v3/n4/full/nclimate1722.html#auth-1> Douglas G. MacMartin, · <http://www.nature.com/nclimate/journal/v3/n4/full/nclimate1722.html#auth-2> David W. Keith, · <http://www.nature.com/nclimate/journal/v3/n4/full/nclimate1722.html#auth-3> Ben Kravitz · & <http://www.nature.com/nclimate/journal/v3/n4/full/nclimate1722.html#auth-4> Ken Caldeira _______________ Ken Caldeira Carnegie Institution for Science Dept of Global Ecology 260 Panama Street, Stanford, CA 94305 USA +1 650 704 7212 <tel:%2B1%20650%20704%207212> [email protected] http://dge.stanford.edu/labs/caldeiralab https://twitter.com/KenCaldeira My assistant is Dawn Ross <[email protected]>, with access to incoming emails. On Thu, Feb 5, 2015 at 9:01 AM, Doug MacMartin <[email protected]> wrote: And a follow up on the follow up to briefly note that we used a very simple model predictive control (i.e. relying on a simple dynamic model to make an initial estimate of the needed radiative forcing that can subsequently be corrected with feedback) in the paper Ken mentioned yesterday. (The focus of that paper was to introduce the idea that the goal of a geoengineering deployment need not be to keep things constant, but potentially just to slow the rate of change… to do so we used feedback and MPC to track a desired trajectory, though we didn’t use the MPC terminology.) I think we’re gradually making small progress as a community on treating this as a design problem rather than turning down the sun and noting what happens, that is, we get to choose how much (solar) geoengineering to do and how to distribute that in space, in order to focus on particular objectives such as maintaining Arctic sea ice. doug From:[email protected] [mailto:[email protected]] On Behalf Of Ben Kravitz Sent: Thursday, February 05, 2015 11:42 AM To: [email protected] Subject: [geo] Re: Assessing the controllability of Arctic sea ice extent by sulfate aerosol geoengineering Hi everyone - Just to follow up on Ken's reply, there has actually been quite a bit of work on this topic (papers attached): D. G. MacMartin, B. Kravitz, D. W. Keith, and A. Jarvis (2014), Dynamics of the coupled human-climate system resulting from closed-loop control of solar geoengineering, Climate Dynamics, 43, 243-258, doi:10.1007/s00382-013-1822-9. B. Kravitz, D. G. MacMartin, D. T. Leedal, P. J. Rasch, and A. J. Jarvis (2014), Explicit feedback and the management of uncertainty in meeting climate objectives with solar geoengineering, Environmental Research Letters, 9, 044006, doi:10.1088/1748-9326/9/4/044006. S. R. Weller and B. P. Schulz, Geoengineering via solar radiation management as a feedback control problem: Controller design for disturbance rejection. (I think this is a white paper, but if someone finds a better citation, please correct me.) I think there is a lot of potential for using control theory in geoengineering research, or in basic climate research for that matter. I would consider this a wide open field right now. Putting controllers in climate models can be applied to many different variables to achieve many different climate goals. In the attached MacMartin et al. and Kravitz et al., we applied simple proportional-integral control (explained in more detail in the two papers), where we used information from past years to calibrate the response of geoengineering. In the Weller and Schulz paper, they used a different method of control to do something similar. What I really like about Jackson et al. is that they used model predictive control - instead of just using information from past years, they also ran a simple predictive model after every year and used that information to guide their controller as well. As Andrew said, a very clever approach. The important thing to note in these papers is the use of two separate models: a control design model and a real world proxy. This mimics how control on SRM might actually work in the real world (assuming such a thing were actually doable in the real world and that SRM would work in a similar way to how the models say it would). One would effectively get as many climate model simulations as one wanted, but there is only one Earth, so it's very important to "get it right" in the real world proxy the first time. Best, Ben -- You received this message because you are subscribed to the Google Groups "geoengineering" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/geoengineering. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "geoengineering" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/geoengineering. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "geoengineering" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/geoengineering. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "geoengineering" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/geoengineering. For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "geoengineering" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/geoengineering. For more options, visit https://groups.google.com/d/optout.
