160 likes | 282 Views
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
E N D
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
IMSF Approach • Link physical process models to system dynamics models • Common GUI • Common data store • Two-way communication • Automatic calibration
SD Model T A B U Spatially Indexed Database PP Model Integrated Modular Simulation Framework Dynamic Data Manager Impervious Cover GUI Stream Buffers Pipe Leakage Min. Spring Flow Compare Results Pumping Limits Add Method Drought Triggers Optimization
Barton Creek Williamson Creek Interstream recharge Slaughter Creek Bear Creek Onion Creek No recharge Example: Barton SpringsGroundwater Availability Model (GAM) • 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 4 5 Effective parameters are extracted from the MODFLOW model. Powersim model is calibrated using a TABU search. 1 2 6 7 9 8 10 11
Calibration Zone 8 • Flow b/t Zones • Average Heads • Spring Flow 8 to 7 8 to 2 8 to 6 8 to 9 8 to 10 8 to 11 Powersim Powersim MODFLOW MODFLOW Powersim MODFLOW Powersim MODFLOW Powersim MODFLOW Powersim MODFLOW 3 4 5 1 2 6 7 9 8 10 11
Calibration 3 1 • Flow b/t Zones • Average Heads • Spring Flow 2 7 11 10 8 3 4 4 6 9 5 2 1 7 6 5 9 8 10 11
Calibration Barton Springs Cold Springs • Flow b/t Zones • Average Heads • Spring Flow Powersim MODFLOW Powersim MODFLOW 3 4 5 2 1 7 6 9 8 10 11
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