banner


February 12, 2013 | Kellogg Center, Riverside Room | Michigan State University

The Environmental Science and Policy Program (ESPP) and the Sustainable Michigan Endowment Project (SMEP) hosted a one-day symposium on Environmental Systems Modeling. This need was identified through a survey of MSU faculty engaged in simulation modeling of environmental systems. Presentations and round-table discussions will engage faculty and students in a forum to develop approaches for strengthening our modeling capacity and brainstorm next steps.

smeptable

smep logoespp logo

Invited panelists addressed environmental modeling approaches and shared thoughts about existing as well as needed capacities for leadership in this area. Facilitators provided for breakout groups helped participants develop recommendations for how to build interdisciplinary modeling collaborations. The focus of the symposium was on open discussions among all participants in the breakout and wrap-up discussions.

To view the white paper that lead to this symposium, click here: modeling whitepaper


Presenters

Speaker Department
Bruno Basso Geological Sciences
Stephen Gasteyer Sociology
Daniel Hayes Fisheries & Wildlife
Richard Horan Agricultural, Food & Resource Economics
Mike Jones Fisheries & Wildlife
Arika Ligmann-Zielinska Geography
Phanikumar Mantha Civil & Environmental Engineering
Sandra Marquart-Pyatt Sociology
Laura Schmitt-Olabisi Community, Ag, Recreation & Res Studies
Alex Adam Smith Plant, Soil and Microbial Sciences

Schedule

Time Description
8:00 - 8:15 a.m. Check in and light breakfast
8:15 - 9:00 a.m. First Session - White Paper: Modeling environmental systems at MSU Drs. Laura Schmitt-Olabis and Adam Smith
9:00 - 9:10 a.m. Break
9:10 - 10:40 a.m. Second Session - Linked Environmental Models Drs.Dan Hayes, Bruno Basso, Richard Horan and Phanikumar Mantha
10:40 - 10:50 a.m. Break
10:50 - 12:20 p.m. Third Session: Participatory Modeling Drs. Stephen Gasteyer, Mike Jones, Laura Schmitt-Olabisis
12:20 - 1:30 p.m. Lunch
1:30 - 3:00 p.m. Fourth Session: Teaching Environmental Modeling at MSU Drs. Arika Ligmann-Zielinska, Sandra Marquart-Pyatt and Laura Schmitt-Olabisi
3:00 - 3:10 p.m. Break
3:10 - 4:00 p.m. Next Steps and Wrap Up

Review of the Symposium and What Comes Next

Models are used to answer the question "what if?". But without improved communication, the ideal answers will continue to evade modelers. While Michigan State University is home to a wide diversity of modeling skills, and most are open to collaboration and have plenty of mutual respect, several barriers remain. The university needs to remove social/institutional barriers, and increase the sharing of information if its environmental modelers are going to thrive.

These are only a few of the suggestions that flowed from an Environmental Systems Modeling Symposium, hosted by the Environmental Science & Policy Program, Community, Agricultural, Recreation & Resources Studies (CARRS) and the Sustainable Michigan Endowment Project on February 12.

The symposium is part of an ongoing effort to provide leadership and interdisciplinary cooperation. It came out of the research of Drs. Alex Adam Smith and Laura Schmitt-Olabisi who determined that more training, education and collaboration will strengthen MSU's ability to both create and use models but also to train future modelers.

In response to the research, the symposium provided a forum for faculty and students to exchange ideas on approaches for strengthening MSU modeling capacity and brainstorm next steps. Panelists from MSU faculty and students led discussions during four main sessions on the nature and methodology of modeling approaches, MSU's strengths and gaps in modeling, and recommendations for how to build interdisciplinary modeling collaborations in the environmental arena. Panelists included Bruno Basso, Stephen Gasteyer, Dan Hayes, Richard Horan, Mike Jones, Arika Ligmann-Zielinska, Phanikumar Mantha, Sandra Marquart-Pyatt, Laura Schmitt-Olabisi, and Alex Adam Smith. In the SMEP (Sustainable Michigan Endowed Project) tradition, this symposium will allow us to build toward a modeling academy during the next academic year.

The first session focused on Linked Environmental Models. This type of simulation model can provide powerful insights into the environmental challenges of managing social ecological systems. Rather than asking what can linked models do?, it would be more appropriate to ask, what can they NOT do?.

Examples of linked environmental models include landscape hydrology linked with Integrated Landscape Hydrology Models and the Systems Approach to Land Use Sustainability extending modeling of vegetation. Other examples include linked human disease and wildlife, and water sustainability in megacities. Likely funding sources in other cases, such as for an oil spill, would be the industry/private sector.

Proposals for funding to address climate change in the Great Lakes region should link modeling research to specific study sites. Water could be an environmental case to attract funding. Linking models must be done at appropriate scales and take account of the complexity of model boundaries and exogenous and indigenous variables. Scales of interest must be defined which include grid and sub-grid processes and the models for each.

However, in addition to process and model uncertainty, there is the deep uncertainty due to unknown unknowns. There is a need for expertise for how to communicate this uncertainty to public/stakeholders. Also, feedback among several linked models creates complexities, especially when the model approaches are fundamentally different, such as deterministic and probabilistic.

In terms of support, more lead time is needed for MSU research teams to both anticipate and respond to funding cycles.  Suggested approaches to this at a university level would be to strengthen support across departments; e.g., reward for collaboration among faculty, recognition of research activities, who is lead on proposals, published papers etc.

Support for administration of research projects could include grant writing, sustained funding for maintaining models and systems, model integrity and evolution. Also, university support for a course in Systems Science and Modeling would be useful across departments. Among faculty, academic leadership in modeling is needed. Examples could be found by consulting the history at MSU where leadership has been demonstrated and recognized. Further, it is important to break down barriers between departments and colleges and support interdisciplinary collaboration. The university needs to make a long-term commitment to these disciplines and interdisciplinary work in order to break down silos across colleges.

One way to work toward those goals is a two day workshop targeted on a specific technique or topic that could do the following: identify missing components of a particular environmental model; and, identify "go to" institutions or persons in terms of who is doing what and where. Approaches might include inviting top modelers to MSU for a couple of days (through faculty nominations) or sabbatical or faculty improvement support at another MSU department. Also, the workshop could identify linkages to qualitative models.

In addition, ongoing support could be provided through an ESPP website. This resource could help researchers search for types (techniques) of environmental models and provide links to MSU modelers and their representative papers/publications. It could provide links to modelers outside of MSU.  Suggestions for facilitating collaboration included a web site functionality, such as Google group or LinkedIn, whereby researchers could self-identify their interests and establish mutual research interests.


Participatory Modeling

In the case of Participatory Modeling (P-modeling), best practices for linking simulation models involve working with modelers who provide input to your model and/or who use your outputs such as the Global Trade Analysis Project (GTAP) at Purdue University.  It is also important to have the trust of other modelers. While there exist some best practices, universally accepted rules are not available. Suggestions for P-modeling best practices offered by symposium participants include:

  • Develop metrics for evaluation processes which relate to goals
  • Employ stratification of stakeholder participants when they are diverse
  • Limit participation to 30-40 people in a process problem and rely on self-organization to occur within the group (even if not perfect)
  • Be pragmatic about the approach taken
  • Consider what interests are represented in the P-model
  • Involve authorized entities in P-modeling whose scale of authority is key to effective problem solving
  • Learn about facilitation which is key to being effective with P-modeling
  • Use an iterative analytical deliberative process for building a P-model
  • Maintain transparent processes which involve stakeholders is key to building their trust and their acceptance of the large amount of uncertainty typical with P-models
  • Resist the temptation to describe P-model results as more certain or with more confidence than is really justified
  • Carefully assess what the boundaries are on the P-modeling approach being used and, if necessary, break it down again and again.

Because participatory modeling can be both challenging and costly relative to other more exclusive approaches, there needs to be a good rationale for its use. P-modeling might appear to be a desirable approach if the funder's Request For Proposal (RFP) requires stakeholder engagement. But use P-modeling with caution as engagement device for a research project deploying linked models.

P-modeling is desirable when "people" are involved, which is common with complex and/or diverse issues, and for which answers can't be obtained through secondary data.

For example, MSU Extension could be a venue for access to stakeholders, but appropriate boundaries would need to be set

Good research centers for P-modeling include:  University of British Columbia (UBC), Portland State, SUNY Albany, Maryland, Arizona State. In addition, European work in water using P-modeling has appeared in top journals.

There are currently some courses at MSU which provide training in P-modeling such as those in CARRS. Courses at MSU need to open up to quantitative modelers. Other recommendations for training students include:

  • Teach role playing and public speaking
  • Participate in field-testing ideas for environmental models
  • Provide courses in experimental economics

As far as communicating the modeling courses, the ESPP website could provide a list of MSU environmental modeling courses and various channels can be used to advertise the MSU-ESPP environmental modeling certificate.  This could provide detailed information down to the department level. We suggest exploring support for this through funding from a NSF IGERT proposal. Initial efforts for a certificate could focus on Sustainability modeling, perhaps with an emphasis on Water.

The benefits to students depend in part on their skills. Students interested in a modeling certificate or curriculum may lack quantitative skills. If so, this needs to addressed in academic advising and curriculum development. Also, systems thinking should be a fundamental component of the curriculum.

Overall, modeling is an important part of research, academics and real-life problem solvings. By using an interdisciplinary approach and sharing the vast knowledge of modeling across MSU, we can increase both the understanding of environmental systems as well as the application of modeling in students and researchers of the future.


Photographs from the symposium

smepzhao
ESPP Director Jinhua Zhao welcomes presenters and attendees to the first Environmental Systems Modeling Symposium at MSU's Kellogg Center

smepolobisi
Dr. Laura Schmitt-Olabisi opened the symposium by discussing current modeling efforts at Michigan State University

smepover
Symposium overview

smepsmith
Alex Adam Smith, primary author of the white pages that lead to this symposium, reviewed his research.

smithhoran
Dr. Richard Horan explains his use of linked environmental models

smephayes
Dr. Daniel Hayes on linked environmental models

smepbasso
Dr. Bruno Basso

smepswinton
Dr. Scott Swinton

smepbreak
Breakout session I

smepbreak2
Breakout session II

smeppyatt
Dr. Sandra Marquart-Pyatt discusses teaching modeling systems