Modelling for policy advice under a global change contextInternal

Expert workshop on measuring and improving the fit of spatial-temporal aquatic ecosystem models

Activity Overview

Type: Foresight Workshop
Start Date: June 3, 2019
End Date: June 5, 2019
Host: CSIC
Venue: CSIC-ICM (Institute of Marine Sciences), Passeig Marítim, 47, 08003 Barcelona
Contact: Marta Coll Monton
E-mail: mcoll [at] icm.csic.es
Funding Call: EuroMarine 2018 call for FWS and WG proposals
Total Budget: €17,800
Funds Granted: €7,500

Manager(s): Marta Coll Monton
Co-organiser(s): Jeroen Steenbeek

Ecosystem modelling tools are increasingly applied in the contexts of management advice, policy exploration, and environmental impact analysis under climate change scenarios. The interdisciplinary modelling required integrates a wide range of disciplines - environmental change, biochemistry and hydrology, food web dynamics, and human activities, at local and global scales, and at various time scales. An Achilles heel of this type of interdisciplinary and holistic modelling is input data uncertainty, which propagates and proliferates through the various stages in a model, and may overwhelm significant trends in model results. Because of several compounding factors such as the complex parameterization, their long run times, limited computing facilities, etc., full uncertainty assessments are rarely undertaken, casting doubt on the usefulness of modelling exercises.

Distributed / cloud computing provides the ideal framework for implementing both types of features. This foresight workshop, with scientific leaders in the world of spatial-temporal ecosystem modelling, will conceptualize how these challenges should be included into existing and future modelling complexes.

KEY OBJECTIVES

The workshop will discuss practical cases around the Ecopath with Ecosim (EwE) food web modelling approach, a freely available and open source spatial-temporal ecosystem modelling software package, but overarching recommendations will be made for the field of ecosystem modelling in general, considering the expertise of participants with other modelling approaches (e.g., Atlantis, Osmose, Bayesian species distribution models, etc.).

EXPECTED OUTCOMES

The outcomes of this workshop will be:

  1. A position paper with an overview of how existing ecosystem models incorporate uncertainty, and recommendations for ecological models to encapsulate these capabilities in general
  2. A best practices paper for spatial-temporal modelling using the EwE approach
  3. A step-wise implementation plan to implement the uncertainty assessment and spatial fitting tools into the EwE approach. The step-wise approach allows EwE to gradually attain desired functionality in incremental steps, each funded through and providing meaningful content to funding projects.

The foresight workshop is the essential starting point for these efforts, as it provides a unique opportunity to bring together interdisciplinary experts in ecological modelling and Bayesian belief network statistics.

EXPECTED IMPACTS

Participants of the foresight workshop will use the discussed products to seek funding for constructing uncertainty assessment and spatial fitting tools for the EwE, where material developed during and after the workshop will provide direct input to funding proposals. These tools will be made available at no cost to the EwE user community, which consists of an estimated 8000 users in 170 countries worldwide, and will provide a roadmap for addressing spatial-temporal model uncertainty in other ecological models.

*This workshop has set aside 4-5 spaces for interested EuroMarine Members. If you are a member of EuroMarine and would like to inquire about attending, please contact the workshop manager or co-organiser (click on their names for contact details).

This foresight workshop attracted 20 participants from 18 organisations from around Europe, North America and Australia. Held in Barcelona, the workshop was held under the premise that in order for holistic spatial‐temporal models to move beyond academic exercises and into real‐world applications, the field of ecosystem modelling requires two capabilities that to date have not really been developed due to lack of computing power and facilities, but that are within reach with present‐day technological advances:

  • Measuring the impact of input parameter uncertainty onto the model results of spatial-temporally explicit ecosystem models
  • Validating spatial-temporally explicit ecosystem models by fitting to observational data, taking uncertainty in input data into account

This workshop thus set out to conceptualize how to include these facilities into existing and future modelling complexes. 

The participants concluded the workshop having developed the skeleton for a position paper with an overview of how existing ecosystem models incorporate uncertainty, and recommendations for ecological models to encapsulate these capabilities (i.e. measuring the fit, model fitting) in general. The attendees also began work on a draft skeleton of an Ecospace best practices paper. Finally, the workshop discussed how to implement, in an incremental and step-wise manner, a spatial-temporal model fitting routine in Ecospace. This discussion led to the development of a roadmap for both the immediate steps and ideal eventual features of such a model fitting routine.

Over the coming months the workshop members intend to continue to develop the position paper and best practices paper, as well as the roadmap representing the technical and scientific framework for their proposed solutions. The goal of the researchers is to present all three of these outputs and their key findings at the "Ecopath 35 years" conference in St. Petersburg, Florida, USA in December 2019. A poster of the roadmap can be viewed via the link below.