By Cheryl Porter (University of Florida), Sander Janssen (Wageningen University), and Ioannis N. Athanasiadis (Democritus University of Thrace)
The Agricultural Model Intercomparison Project (AgMIP) is a distributed climate- scenario simulation exercise for historical model intercomparison and future climate change conditions with participation of multiple crop and world agricultural trade mod- eling groups around the world. The goals of AgMIP are to improve substantially the characterization of risk of hunger and world food security due to climate change and to enhance adaptation capacity in both developing and developed countries.
Advertisement
6 comments
Comments feed for this article
May 14, 2011 at 12:29 am
Spencer Rugaber
The paper presents an ambitious vision of what the software infrastructure might look like in support of intercomparison of agricultural models. Its mentions a wide variety of software opportunities and challenges. One implicit assumption is that the effort should/will be top-down, even though requirements elicitation is mentioned. An alternative is to approach the problem in a consortial fashion in which existing efforts are pieced together driven by community interest. There are numerous examples within the climate community of efforts undertaken in this fashion leading to emergent standards. In fact, it may be the building of the community itself that is the chief outcome. In any case, this is certainly a worthwhile goal and would be of interest to workshop participants.
May 17, 2011 at 7:38 pm
Ioannis N Athanasiadis
Spencer, you are actually right in your argument that this should be a bottom up approach (and actually will be such). The AgMIP community is a global project with very little funding for its activities: There will be funding for local projects, and no funding for central developments. So, a top-down model doesn’t not work. What we want to achieve though, is to coordinate in a global scale the local projects’ software development, and possibly have a common infrastructure for data and software sharing. Getting the various pieces developed independently, working together will be a real challenge. Certainly, relevant experience from the climate community will be a great asset for guiding our future steps.
May 19, 2011 at 5:41 pm
Elena Pérez-Miñana
the project is certainly an ambitious one. Having familiarised myself over the last couple of years with repositories that store data from the UK agriculture sector, I can certainly agree this is an interesting research activity although I’m not sure the top down approach suggested by the description will provide the best set of results. It certainly is a proposal worth discussing.
May 19, 2011 at 5:53 pm
Adrian Schroeter
when I first read that paper I had a similar impression that it might be worthwhile to start bottom up, simply because people need the local information asap. I think the community at large should go a similar way as the OSLC project (http://www.open-services.net/html/Home.html) and define a standard for data exchange between the different models. Overall I think it is a worthwhile project, with a goal of model integration set as primary goal, but I have the feeling that the approach is a bit “last century” in the sense that it proposes a centralized system.
June 4, 2011 at 6:43 pm
Ioannis N Athanasiadis
Thanks to Adrian and Elena for their comments. Looking forward to discuss further our ideas in Lancaster!
June 7, 2011 at 1:43 pm
Steffen Zschaler
Certainly a very interesting project to hear about and discuss at the workshop.
I have to admit that, as I am not familiar with the agricultural sector at all, I was confused by quite a bit of the terminology and presentation. You may wish to consider revising your paper to make it more easily accessible to the “uninitiated” software engineer like myself in order to get more and better feedback. For example, it may be useful to clarify what form these models generally take, or be a bit more specific about the comparison challenges faced.
It may also be useful to separate the technological challenges of the domain (e.g., the complexity of models and the substantive differences between them) from the more organisational difficulties (e.g., as reported in your comment above, the unavailability of funding for a centralised control, or the varying skill sets of the stakeholders).