Integration of Spatially Aggregated Physical Process Models with

Integration of Spatially Aggregated Physical Process Models with Systems Dynamics Models to Assist the Decision Support Process Sandia National Laboratories University of Texas – Austin Geological Society of America 2005 Annual Meeting

Introduction • Motivation • Approach – Integrated Modular Simulation Framework (IMSF) – Rapid Dispute Prevention (RDP) • Example – Barton Springs segment of the Edwards Aquifer, Austin, TX

‘High-Level’ Motivation • Incorporate scientific analysis into the decision making process • Allow stakeholders to guide the scientific process • Employ advanced policy and decision making techniques • Maximize economic, environmental, and demographic sustainability

Barton Springs • Assess impacts of development (e. g. impervious cover) on water quantity and quality issues – Spring flow – Drought triggers – Economic impacts • Groundwater flow models • Stakeholder involvement Areal extent of Austin from 1885 to 1985

‘Core’ Motivation • Need to incorporate spatially detailed modeling capabilities • Need to analyze systems level responses • Need this as one tool that can be implemented by non-modelers

Approach • Link physical process models to system dynamics models – Common GUI – Common data store – Two-way communication – Automatic calibration IMSF

Integrated Modular Simulation Framework Dynamic Data Manager SD Model Impervious Cover GUI Stream Buffers Pipe Leakage Min. Spring Flow Compare Results Add Method Optimization Spatially Indexed Database T A B U Pumping Limits Drought Triggers PP Model

Example: Barton Springs Groundwater Availability Model (GAM) Barton Creek Williamson Creek Interstream recharge Slaughter Creek Bear Creek Onion Creek No recharge • 120 x 120 cells • 1000 m x 500 m cell size • Steady State and Transient Versions • Recharge, well pumping, drains (Barton and Cold Springs)

Change of Resolution Convert PP model to coarse-resolution SD model through zonation 3 Effective parameters are extracted from the MODFLOW model. Powersim model is calibrated using a TABU search. 1 7 2 6 9 8 11 4 10 5

Calibration Zone 8 8 to 7 3 1 4 2 6 7 9 8 10 11 5 8 to 11 MODFLOW Powersim 8 to 10 Powersim MODFLOW Powersim MODFLOW • Flow b/t Zones • Average Heads • Spring Flow 8 to 9 MODFLOW 8 to 6 MODFLOW 8 to 2

Calibration 3 1 2 7 • Flow b/t Zones • Average Heads • Spring Flow 3 4 2 1 7 11 5 10 8 9 4 6 6 9 8 10 11 5

Calibration 4 2 1 7 6 9 8 10 11 5 Powersim MODFLOW Powersim • Flow b/t Zones • Average Heads • Spring Flow 3 Cold Springs MODFLOW Barton Springs

Benefits and Summary • SD model executes much faster than the PP model – Scenario testing – Stakeholder education • Allows for connecting important physical processes to other systems • Provides a single user interface that works with both models • Provides on-the-fly calibration between each model • Modular approach allows for application to different types of problems

Acknowledgements • Sandia National Laboratories – Thomas S. Lowry – Vincent C. Tidwell • University of Texas – Suzanne Pierce – John M. Sharp – Marcel Dulay – David Eaton – Michael Ciarleglio – Aliza Gold – Roy Jenevein – A host of others…. . • William Cain

Thank You