PARAMETERIZATION, UNCERTAINTY – AND “REAL WORLD” MODELING Examples of Uncertainty from Everyday Models See www.sspa.com for more details
Model Applications See www.sspa.com for more details SIMULATION PREDICTION PARAMETERS CALIBRATION PRIOR INFORMATION Saline Intrusion Maximum Salinity Boundaries, flow, transport Heads, flows, salinities Pump tests Contaminant transport Contaminant source location Heads, flows, concentrations, mass flux Pump tests, slug tests Maximum influent concentration Pump tests, slug tests, domestic water use (recharge) Water supply Baseflow depletion, wetland levels Boundaries, flow Heads, flows, depletions Recharge (DPIRSIM/Regression), Pump tests Local flow Surface water seepage Heads, flows, seepage measurements See www.sspa.com for more details
Examples PROJECT UNCERTAINTY LI-1 LI-2 MI Uniondale, Long Island MTBE plume, IRM design Geologic setting - results presentation Hampton Bays, Long Island Contaminant distribution - mass Michigan Water supply study Surface water impacts Flows - problem formulation and simulation LI-1 LI-2 MI See www.sspa.com for more details
Salient Points Build the model with parameterization in mind right from the beginning Construct calibration targets in the beginning and revise throughout to obtain details Build calibration tools/programs right from the start Construct graphical post-processing tools from beginning and throughout – graphs, maps, summary statistics See www.sspa.com for more details
PEST as Pre-Post Processor Early steps in model construction – code selection, domain, spatial discretization – are reasonably unlikely to change Hence - can use PEST from this point forward to control runs, vary parameters, etc Build open-ended programs – modular, error handling Calibration/uncertainty tool remains current as the model develops Model testing takes place as model built See www.sspa.com for more details