EPA530-D-03-001d
July 2003
SAB Review Draft
Multimedia, Multipathway, and
Multireceptor Risk Assessment
(3MRA) Modeling System
Volume IV: Evaluating Uncertainty
and Sensitivity
-------
EPA530-D-03-001d
July 2003
SAB Review Draft
Multimedia, Multipathway, and
Multireceptort Risk Assessment
(3MRA) Modeling System
Volume IV: Evaluating Uncertainty
and Sensitivity
prepared by
U.S. Environmental Protection Agency
Office of Research and Development Office of Solid Waste
Justin E. Babendreier Washington, B.C.
National Exposure Research Laboratory
Athens, GA
-------
This document is the fourth volume of a five-volume set. This Volume IV describes the
methodology used to evaluate sensitivity of model parameters and characterize different types of
uncertainty in the 3MRA modeling system. Volume I describes the conceptual design, scientific
rationale, and supporting data that are the foundation for the 3MRA modeling system. Volume II
describes the data developed and used to run the 3MRA modeling system. Volume III describes
the approach to quality assurance, including verification and validation activities ranging from
extensive peer reviews to multimedia model comparisons. Volume V describes the 3MRA
Technology design and users guide.
-------
Table of Contents
Table of Contents
Section Page
List of Figures xii
List of Tables xv
1.0 Introduction 1-1
1.1 Report Objectives 1-2
1.2 3MRA Risk Assessment Problem Statement 1-2
1.2.1 National Risk Assessment Problem Statement Formulation 1-4
1.2.2 Validity of the National Risk Assessment Implementation 1-5
1.3 Overview of Uncertainty and Sensitivity Analyses Concepts 1-6
1.3.1 Notation and Terminology of Uncertainty and Sensitivity Analyses 1-7
1.4 Quality Assurance in Modeling 1-8
1.4.1 Qualitative and Quantitative Model Evaluation Approaches 1-9
2.0 Taxonomies for Probabilistic Assessment 2-1
2.1 Probabilistic Analysis 2-1
2.1.1 Context for 3MRA Populations, Subpopulations, and Individuals 2-3
Population Protection Criteria 2-4
2.1.2 Frequency Vs. Probability for Populations and Individuals 2-5
2.2 Taxonomy of Model Input Quantities 2-5
2.3 Model Evaluation 2-6
2.3.1 The Model Validation Paradox 2-7
2.3.2 Validation and Uncertainty in Prediction 2-8
2.3.3 Model Synthesis and Analysis 2-9
2.3.4 Pedigree of Information 2-9
2.3.5 Components of Model Evaluation 2-10
-------
Table of Contents
Table of Contents (continued)
Section Page
2.3.6 Model Use and Desired Outcome 2-12
2.3.7 Validating Predictive Exposure and Risk Assessments 2-13
2.3.8 Summary of Model Evaluation Concepts 2-15
2.4 Taxonomy of Uncertainty 2-15
2.4.1 Variability 2-16
2.4.2 Input Uncertainty 2-16
2.4.3 Model Uncertainty 2-17
2.4.4 Total Uncertainty 2-18
2.5 Monte Carlo Simulation 2-19
2.5.1 Historical Perspective of the Use of Monte Carlo Simulation 2-19
2.5.2 MCS Input Sampling Approach and Use of the Uniform Distribution.. 2-20
2.5.3 Assessing Simulation-Induced Output Sampling Error (OSE) 2-21
Confidence Interval Width for Normally Distributed Outputs 2-21
Bounded Interval For A Specific Fractilefor Any Distribution Type.... 2-22
2.5.4 Latin Hypercube Sampling 2-24
LHS versus Random Monte Carlo Sampling 2-25
2.5.5 Dealing With Correlations Among Random Inputs 2-27
2.6 Distinguishing Variability From Uncertainty 2-27
2.6.1 Subjective Versus Objective Uncertainty 2-28
2.6.2 Sources of Variability and Uncertainty in 3MRA 2-29
2.6.3 3MRA Model Input and Output for a Single National Realization 2-30
Interpretation of3MRA Site-Based National Realizations 2-32
2.6.4 Four Cases of Uncertainty 2-33
Case 1: Constant and Certain 2-33
Case 2: Constant and Uncertain 2-34
Case 3: Variable and Certain 2-34
Case 4: Variable and Uncertain 2-35
2.6.5 Four Cases of Uncertainty for 3MRA Input Descriptions 2-36
Sample Measurement Error (SME) 2-36
Input Sampling Error (ISE) 2-36
Assignment of Cases for Current 3MRA Static Distributions 2-3 7
2.6.6 Two-Dimensional Analysis of Uncertainty 2-38
General Form of 2nd-order Analysis Algorithm for Risk Analysis 2-3 8
Pseudo 2nd-order Analysis Algorithm for Risk Analysis 2-40
Output Sampling Error (OSE) 2-40
-------
Table of Contents
Table of Contents (continued)
Section Page
2.6.7 Classification of 3MRA Inputs and Interpretation of Outputs 2-41
Hybrid Dimension of Total Uncertainty 2-41
On Separating the Hybrid Dimension of Total Uncertainty 2-42
Population Vs. Individual Risk 2-42
2.7 Taxonomy of Sensitivity Analysis 2-43
2.7.1 Classification of Sensitivity Analysis Methods 2-44
2.7.2 Overview of Screening, Local, and Global Approaches 2-44
2.7.3 Screening Methods for Sensitivity Analysis 2-46
Morris' OAT Design 2-47
Cotter's OAT Design 2-47
IFFD Method of Andres 2-48
2.7.4 Regression/Correlation Variance-Based Global Methods for SA 2-49
Scatter Plots 2-49
Regression Analysis 2-50
Partial Correlation Measures 2-50
Step-Wise Regression Analysis 2-51
Rank Transformations 2-52
2.7.5 Regional and Tree-Structured Global Methods for SA 2-52
Regionalized Sensitivity Analysis (RSA) 2-53
Kolmogorov-Smirnov Statistic 2-53
Behavior Definitions 2-54
Tree-Structured Density Estimation (TSDE) 2-54
Uniform Covering by Probabilistic Reject!on (UCPR) 2-54
2.7.6 Other Variance-Based Global Methods for Sensitivity Analysis 2-55
2.8 Sensitivity-Based Performance Validation Using RSA-TSDE 2-56
2.8.1 Hornberger-Spear-Young (HSY) Algorithm 2-57
2.8.2 Summary Excerpts From Chen and Beck (1999) 2-57
Overview 2-57
Example Model Description 2-58
Output Uncertainty as a Function of Different Site Characteristics 2-58
Key and Redundant Inputs in Predicting a Percentile Concentration... 2-5 9
2.9 EPA Guidance for Use of Sampling Based Approaches in
Risk Analysis 2-60
in
-------
Table of Contents
Table of Contents (continued)
Section Page
3.0 Overview of the 3MRA Science Methodology,
Data Sets, Uncertainty, and Implementation 3-1
3.1 Historical Perspective of Hazardous Waste Management
andHWIR 3-1
3.2 3MRA Science Plan Objectives 3-3
National-Scale Risk Assessment Approach 3-3
Fate and Transport 3-4
Exposure Pathways and Receptors 3-5
3.3 Science Based Modeling Approach 3-5
3.3.1 Science-Based Models 3-6
3.3.2 Runtime System Level Processors 3-6
3.3.3 Exit Level Calculation Steps 3-7
3.4 Sources of Data and Uncertainties for the National
Assessment Strategy 3-8
3.4.1 Sources of Data Actually Collected By Science Module 3-9
3.4.2 Waste Management Unit 3-10
3.4.3 Waste Properties 3-10
3.4.4 Meteorological Database 3-10
3.4.5 Surface Water And Watershed Layout and Characteristics 3-11
3.4.6 Soil (VadoseZone) Properties 3-12
3.4.7 Aquifer (Saturated Zone) Properties 3-12
3.4.8 Food Chain Or Food Web Characteristics 3-13
3.4.9 Human Receptor Data 3-14
3.4.10 Human Exposure Factors 3-15
3.4.11 Ecological Receptor Data 3-15
3.4.12 Ecological Exposure Factors 3-16
3.4.13 Chemical Properties Database 3-17
3.5 Monte Carlo Implementation for the National Assessment 3-18
3.5.1 3MRA Site-Based Approach Monte Carlo 3-19
3.5.2 Limitations in Implementation of the Monte Carlo Approach 3-20
3.5.3 Summary Points on 3MRA Version 1.0 Implementation 3-21
IV
-------
Table of Contents
Table of Contents (continued)
Section Page
4.0 Overview of 3MRA Version 1.0 Technology 4-1
4.1 Models and Modeling Systems Within a Technological
Framework 4-2
4.2 3MRA Version 1.0 Modeling System 4-2
4.2.1 Multimedia Multipathway Simulation Processor 4-3
4.3 Modeling System Execution Flow in 3MRA 4-4
4.3.1 Site Definition Files For a Given Scenario 4-4
4.3.2 Model Output and System Level Error and Warning Management 4-5
4.3.3 Data Management For Uncertainty and Sensitivity Analysis 4-5
Aggregated Versus Disaggregated Exit Level Processing 4-6
4.3.4 3 MR A Pseudo-Random Number Generator 4-6
SDP Seed Algorithm for Random Sampling 4-7
4.4 Use of 3MRA In National Risk Assessment Strategies 4-8
4.4.1 Cross-Sections of Data Analysis Possible 4-9
4.4.2 National Site Sample Selection 4-10
4.4.3 Representation of the Site-Based Sample of Subtitle D Facilities 4-10
4.4.4 Representation of Meteorological Data Assignment 4-11
4.4.5 National Risk Assessment Problem Statement Formulation 4-11
4.4.6 Waste Stream Concentration Exit Level 4-12
4.5 Science Module Overview - 3MRA Version 1.0 4-13
4.5.1 Model Temporal Scale 4-13
4.5.2 Source Modules 4-14
Landfill (LF) 4-14
Wastepile (WP) 4-14
Land Application Unit (LA or LAU) 4-14
Surface Impoundment (SI) 4-15
Aerated Tank (AT) 4-15
4.5.3 Transport, Fate Processes, and Intermedia Contaminant Fluxes 4-15
Air Module (ar) 4-15
Water shed Module (ws) 4-16
Groundwater (Vadose and Aquifer) Modules (aq, vz) 4-16
Surface Water Module 4-16
-------
Table of Contents
Table of Contents (continued)
Section Page
4.5.4 Food Chain/Food Web Components 4-17
Farm Food Chain Module (ff) 4-17
Terrestrial Food Web Module (tf) 4-18
Aquatic Food Web Module (of) 4-18
4.5.5 Human Receptors and Exposure Pathways 4-18
Human Receptors 4-18
Human Exposure Pathways 4-19
Human Exposure Module (he) 4-19
Human Risk Module (hr) 4-20
4.5.6 Ecological Receptors and Exposure Pathways 4-20
Ecological Receptors 4-20
Ecological Exposure Pathways 4-21
Ecological Exposure Module (ee) 4-21
Ecological Risk Module (er) 4-22
4.5.7 Risk Metrics 4-23
Human Health Risk Summary Output 4-24
Ecological Risk Summary Output 4-24
4.5.8 Aggregation Versus Disaggregati on of Risk Metrics 4-25
4.6 Calculation Flow and Outputs 4-25
4.6.1 Model Execution Steps for Stand-Alone PC Applications 4-26
Inner and Outer Loops 4-27
Calculation of Exit Levels 4-28
4.6.2 3MRA Output Vector Dimensionality 4-29
4.6.3 Protection Criteria 4-30
Cancer Risk Level 4-30
Human Hazard Quotient (HQ) 4-30
Ecological Hazard Quotient (HQ) 4-30
Population Percentile 4-31
Probability (Frequency) of Protection 4-31
5.0 SuperMUSE Parallel Computing Hardware 5-1
5.1 3MRA Model Evaluation Computational Needs 5-2
5.1.1 Massively Parallel Vs. Embarrassingly Parallel Modeling Problems 5-2
5.1.2 Historic Responses to Embarrassingly Parallel Modeling Problems 5-3
5.1.3 PC-Based Parallel Computing Clusters 5-4
VI
-------
Table of Contents
Table of Contents (continued)
Section Page
5.2 ORD's Supercomputer for Model Uncertainty and
Sensitivity Evaluation 5-5
5.2.1 SuperMUSE Initial Design, Cost, and Effort 5-6
Electrical Service and Heat Dissipation 5-6
KVM (Keyboard, Video, Mouse) Switches 5-7
5.2.2 Runtime Management Issues 5-8
Virus Protection and PC-Client Hard-Disk Image Ghosting 5-8
5.2.3 SuperMUSE Expansion and Life-Cycle 5-9
5.2.4 MiniMUSE - Staging Software Enhancements for SuperMUSE 5-10
5.3 Identification of Software Toolset Needed for SuperMUSE 5-10
5.4 Benefits of Integrated SuperMUSE Hardware and Software 5-11
6.0 3MRA Version 1.x - UA/SA Enhancements e-i
6.1 Overview of SuperMUSE's Supporting Software Toolset 6-1
6.1.1 Supercomputing Supporting Software System Needs 6-2
6.1.2 Tools Facilitating the Distribution of Workloads Among PCs 6-3
6.1.3 Tools for Managing Files and Data Across PCs 6-4
6.1.4 Tools Facilitating 3MRA Post-Processing Data Analysis Tasks 6-5
Sensitivity Analysis for Exit Level Model Outputs 6-5
OSE Uncertainty Analysis for 3MRA Exit Level Outputs 6-6
6.2 Distributed Execution Program Management Tool Set 6-6
6.2.1 CPU Allocator 6-7
6.2.2 Model Tasker 6-8
6.2.3 Tasker Client 6-8
Example Run. bat File 6-9
6.2.4 Client Monitor 6-10
6.3 File and Data Management Tools 6-11
6.3.1 Update Client 6-11
6.3.2 Command Tasker 6-12
6.3.3 Process Messages 6-12
6.3.4 Aggregated ELP1 Client Collector 6-13
6.3.5 Disaggregated ELP1 Client Collector 6-14
vn
-------
Table of Contents
Table of Contents (continued)
Section Page
6.4 3MRA Modeling System Data Analysis Tool Set 6-14
6.4.1 Site Visualization 6-15
6.4.2 Site Summary 6-15
Statistical-Based Summaries 6-16
6.4.3 Aggregated ELP1 for MySQL Databases 6-17
6.4.4 Aggregated ELP2 for MySQL Databases 6-17
ELP 2 Vis for MySQL Databases 6-18
ELP2 for MySQL Databases 6-18
6.4.5 Disaggregated Exit Level Processor Set 6-19
Exit Level Processor User Interface 6-20
SuperELPl 6-20
SuperELP2 6-21
6.5 UA/SA Sampling Tools 6-21
6.5.1 Distribution Statistics Processor 6-21
6.5.2 Enhanced3MRAHwirMC.dllandStat.dll 6-22
6.5.3 Latin Hypercube Sampling (LHS) 6-22
6.5.4 Other Model Input Sampling Tools 6-23
Uniform Coverage By Probabilistic Rejection (UCPR) 6-23
One at a Time (OAT) Sampling Routine 6-23
Factorial Sampling Routine 6-23
LPT and Winding Stairs Sampling Routines 6-23
Fourier Amplitude Sensitivity Test (FAST) Sampling Routines 6-24
Monte Carlo Markov Chain Routine 6-24
6.6 Initial 3MRA Time Trial Analysis 6-24
6.6.1 Initial 3MRA Benchmarking 6-24
6.6.2 Num-Type Site Layout Parameter Regression Analysis 6-25
6.6.3 Chemical Specific Runtimes 6-26
6.6.4 MMSP Runtimes by Source 6-27
6.6.5 MMSP Runtimes By Science Module 6-27
6.6.6 Solubility Errors Encountered During Runtime 6-27
6.7 System Level Quality Assurance Testing Via SuperMUSE 6-28
6.7.1 Problems Corrected as a Result of System-Level Verification Tasking 6-28
Error Detectable Only Through Random Parallel Simulation 6-29
6.8 FRAMES 3MRA Version 2.0 6-29
Vlll
-------
Table of Contents
Table of Contents (continued)
Section Page
7.0 Example 3MRA Uncertainty Analysis for Benzene 7-1
7.1 Benzene Disposal Simulation Design 7-1
National Risk Assessment Problem Statement 7-2
7.1.1 Experiment A: Aggregated lst-Order Total Uncertainty Analysis 7-3
7.1.2 Experiment B: Pseudo 2nd-Order Analysis of Variability and
Empirical Uncertainty 7-3
7.1.3 Interpretation of Variability and Uncertainty in the Experiment 7-3
Output Sampling Error (OSE) 7-4
7.1.4 Summary of Experiment A andB 7-4
7.2 Benzene Disposal Uncertainty Analysis 7-5
7.2.1 Pseudo 2nd-Order Total Uncertainty Analysis Based on Confidence
Intervals 7-6
7.2.2 Pseudo 2nd-Order Total Uncertainty Analysis Based on CDFs of 95%
Probability 7-6
7.2.3 Analysis of All Source Types 7-7
7.2.4 Dominant Exposure Pathway Analysis 7-7
7.3 Efficacy of the Integrated SuperMUSE Approach for 3MRA 7-8
8.0 3MRA Inputs, Outputs, and Dimensionality s-i
8.1 Dimensionality of 3MRA Inputs and Outputs 8-1
Number of Model Runs for the National Assessment 8-2
Number of Exit Levels Possible in the National Assessment 8-2
8.1.1 Dimensionality of Individual Elements Within Input and Output
Vectors 8-3
8.1.2 Convention for Representing Input and Output Vectors in 3MRA 8-3
8.2 System-Level Summary of Model Inputs and Outputs 8-3
8.3 Module-Level Summary of Model Inputs and Outputs 8-9
Dictionary Summary Key 8-9
Fixed-Index Field 8-10
Internal Variables 8-11
Summary of Input Variant and Invariant Model Inputs 8-11
IX
-------
Table of Contents
Table of Contents (continued)
Section Page
9.0 3MRA Version 1.0 Uncertainty Analysis and Sensitivity
Analysis Plan 9-1
Overview of 3MRA Modeling System Documentation 9-1
9.1 Review of Model Evaluation Tasking Concepts 9-2
9.1.1 Composition 9-2
9.1.2 Performance 9-3
9.2 3MRA Model Evaluation Tasking Completed To Date 9-3
9.3 UA/SA 3MRA Model Evaluation Tasking To Be Completed 9-4
9.3.1 Five Components of UA/SA 9-4
Summary of the UA/SA Plan 9-5
Performance uncertainty analysis (UAP) 9-5
System-level sensitivity analysis (SA) 9-6
Sensitivity analysis based performance validation (SAP) 9-6
Representative Chemicals To Be Evaluated 9-6
9.3.2 Overview of Greatest Questions of Concern 9-7
Example Questions Concerning Population and Subpopulation Risk 9-7
9.3.3 Pseudo2nd-Order Uncertainty Analysis (UAP) 9-8
Estimated UAp Runtime 9-8
Execution Scheme and Data Management 9-9
Qualitative Analysis of Dominant Media, Pathways, and Receptors
Driving Exit Levels 9-9
9.3.4 Sensitivity Analysis (SA) 9-9
Regression/Correlation Approaches 9-10
Univariate RSA andMultivariate TSDE Global SA Techniques 9-10
Experimental Design and Level of Effort 9-10
9.3.5 Sensitivity-based Performance Validation (SAP) 9-11
9.3.6 ISE-Based 2nd-Order Uncertainty Analysis 9-11
Tiered Evaluation Approach for Investigating Higher Dimensional
Analyses 9-12
Practical Limitations in Investigating Higher Dimensional Analyses ...9-12
9.4 Long-Term Research Supporting Future Model Evaluation
Capabilities 9-13
X
-------
Table of Contents
Table of Contents (continued)
Section Page
10.0 References 10-1
Appendices
(A) A Framework For Finite-Source Multimedia, Multipathway,
and Multireceptor Risk Assessment 3MRA;
(Marine^/., 1999)
(B) Electronic Reprint of Quality Assurance of Multi-Media Model
For Predictive Screening Tasks; (Chen and Beck, 1999)
(C) Draft Documentation for the FRAMES-3MRA Technology
Software System, Parallel Processor Modifications; (Castleton
et al., 2003)
(D) Section 3 of Volume 8: Specifications. 1998. PNNL-11914,
Vol. 8, Pacific Northwest National Laboratory, Richland,
Washington; Specifications for the Distribution Statistics
Processor (DSP).
(E) Agenda: International Workshop on Uncertainty, Sensitivity,
and Parameter Estimation for Multimedia Environmental
Modeling, August 19-21,2003. Organized by the Workgroup
on Uncertainty and Parameter Estimation; Interagency
Steering Committee on Multimedia Environmental Modeling
XI
-------
Table of Contents
List of Figures
Figure Page
1-1 FRAMES 3MRA Software Versions - Naming Conventions 1-3
2-1 Probability Curves (PDF, CDF, and ICDF) for Random Variable X = N(0, 1) 2-67
2-2 Example of Measured Sampled Empirical Input or Estimated Model Output 2-68
2-3 Model Evaluation Paradigm 2-69
2-4 Model Quantities Subject to Probabilistic Uncertainty Analysis 2-70
2-5 Typical Monte Carlo Flowchart for Stochastic Model Solutions 2-71
2-6 Estimating Sample Size: Error in Estimating the Normal Mean 2-72
2-7 Degrees of Variability and Empirical Uncertainty in Model Inputs 2-73
2-8 Two-Dimensional Monte Carlo Analysis for 3MRA 2-74
2-9 Two-Dimensional Monte Carlo Analysis Flowchart for 3MRA 2-75
2-10 Pseudo 2nd-Order Monte Carlo Analysis Flowchart for 3MRA 2-76
2-11 Conceptualization of Sensitivity Analysis Techniques 2-77
3-1 Conceptual Implementation of 3MRA Model 3-26
4-1 FRAMES Framework Technology, 3MRA Modeling System, and Example
Application 4-35
4-2 Conceptual Relationship Between Framework Technology, Models and Modeling
System, and Applications 4-36
4-3 FRAMES 3MRA Version 1.0 System Design: Stand-Alone Workstation Application 4-37
4-4 Details of the Multimedia Multipathway Simulation Processor 4-3 8
4-5 Interactions Within the Multimedia Multipathway Simulation Processor 4-39
4-6 Sample Population of Facilities Across the Contiguous United States 4-40
4-7 Area of Interest and Concentric Distance Rings 4-41
4-8 Example Cumulative Population Risk Histogram: Inhalation Pathway from
Ambient Air, Single Time Period, Single Receptors 4-41
5-la SuperMUSE Parallel Computing Cluster at ORD/NERL/ERD 5-13
5-lb SuperMUSE Parallel Computing Cluster at ORD/NERL/ERD 5-14
5-2 Conceptual Layout for PC-Based SuperMUSE Parallel Computing Cluster 5-15
5-3 SuperMUSE Network and KVM Switching 5-16
5-4 Example SuperMUSE Cabling 5-17
5-5 MiniMUSE Parallel Computing Cluster at ORD/NERL/ERD 5-18
5-6 Benefits of the SuperMUSE Hardware and Parallel Management Software Scheme... 5-19
6-la SuperMUSE Parallel Management Software Toolset - Conceptual Layout for 3MRA 6-31
6-lb SUITasker Failed Tasks Panel 6-32
6-lc SUITasker Failed Machine Panel 6-32
6-2 Server-Side Client Monitor Tool User Interface 6-33
6-3 Server-Side Update Client Tool User Interface 6-34
xn
-------
Table of Contents
List of Figures (continued)
Figure Page
6-4 Server-Side Command Tasker Tool User Interface 6-35
6-5 Example "Error" Table Created by the Process Messages Tool 6-36
6-6a Command Tasker Implementation of Aggregated ELP1 Client Collector Tool 6-37
6-6b CPU Allocator With Active Command Tasker and SUITasker 6-38
6-6c Example Client Collector Task Implementation on Tasker Client Side 6-3 8
6-7a Example "Description" Table Created by the Site Summary Tool 6-39
6-7b Example "Result" Table Created by the Site Summary Tool 6-39
6-8a Example Table Summary For Aggregated ELP1 MySQL Database 6-40
6-8b Example RTemplate Table in Aggregated ELP1 MySQL Database
(Human Cancer Risk Summary) 6-41
6-8c Example HTemplate Table in Aggregated ELP1 MySQL Database
(Human Hazard Risk Summary) 6-42
6-8d Example ETemplate Table in Aggregated ELP1 MySQL Database
(Ecological Hazard Risk Summary) 6-43
6-8e Example Sitelterlndex Table in Aggregated ELP1 MySQL Database 6-44
6-8f Example BinDes Table in Aggregated ELP1 MySQL Database
(Cancer/Hazard Risk Bins) 6-45
6-8g Example ChemDes Table in Aggregated ELP1 MySQL Database
(Chemical Descriptions) 6-45
6-8h Example CohDes Table in Aggregated ELP1 MySQL Database
(Cohort Descriptions) 6-45
6-8i Example CritDes Table in Aggregated ELP1 MySQL Database
(Critical Time Period Descriptions) 6-46
6-8j Example DistDes Table in Aggregated ELP1 MySQL Database
(Ring Distance Descriptions for Humans) 6-46
6-8k Example EDistDes Table in Aggregated ELP1 MySQL Database
(Ring Distance Descriptions for Ecology) 6-46
6-81 Example EHabGDes Table in Aggregated ELP 1 MySQL Database
(Ecological Habitat Group Descriptions) 6-46
6-8m Example EHabTDes Table in Aggregated ELP1 MySQL Database
(Ecological Habitat Type Descriptions) 6-47
6-8n Example ERecGDes Table in Aggregated ELP1 MySQL Database
(Ecological Receptor Group Descriptions) 6-47
6-80 Example ETrophicDes Table in Aggregated ELP1 MySQL Database
(Ecological Trophic Level Descriptions) 6-48
6-8p Example ExpDes Table in Aggregated ELP1 MySQL Database
(Exposure Pathway Descriptions) 6-48
6-8q Example PerDes Table in Aggregated ELP1 MySQL Database
(Population Percentile Descriptions) 6-49
6-8r Example RecDes Table in Aggregated ELP1 MySQL Database
(Human Receptor Group Descriptions) 6-49
Xlll
-------
Table of Contents
List of Figures (continued)
Figure Page
6-9 Benzene-Landfill Example ELP2Vis Output 6-50
6-10 Benzene-Landfill Example ELP2Vis Output (Less Thumbnail Graphics) 6-52
6-11 Example Scenario "Selections" Table in Aggregated ELP1 MySQL Database 6-53
6-12a ELP2Vis Output- Graphic Detail for Human Risk Summary of Ingestion 6-53
6-12b ELP2Vis Output - Graphic Detail for Human Risk Summary of Inhalation 6-54
6-12c ELP2Vis Output - Graphic Detail for Human Risk Summary of Inhalation Plus
Ingestion 6-55
6-12d ELP2Vis Output - Graphic Detail for Summary of Ecological Risk 6-56
6-12e ELP2Vis Output- Graphic Detail for Human Risk by Exposure Pathway 6-57
6-12f ELP2Vis Output - Graphic Detail for Human Risk By Cohort Group 6-58
6-12g ELP2Vis Output - Graphic Detail for Human Risk By Receptor Group 6-59
6-12h ELP2Vis Output - Graphic Detail for Human Risk By Distance 6-60
6-12i ELP2Vis Output - Graphic Detail for Ecological Risk By Distance 6-61
6-12J ELP2Vis Output - Graphic Detail for Ecological Risk By Receptor Group 6-62
6-13a Enhanced SUI panel for Disaggregated ELP Processing (Human) 6-63
6-13b Enhanced SUI panel for Disaggregated ELP Processing (Ecological) 6-64
6-13c Enhanced SUI panel for Disaggregated ELP Processing (Data Structure) 6-65
6-14a Example RTemplate Table in Disaggregated ELP1 MySQL Database
(Human Cancer Risk Summary) 6-66
6-14b Example Sitelterlndex Table in Disaggregated ELP 1 MySQL Database 6-66
6-15 Comparison of Average 3MRA Runtimes: SuperMUSE Parallel Vs.
Stand-Alone Execution Modes 6-67
6-16 Model Runtimes: Initial 3MRA Time Trial Analysis for Selected Chemicals 6-70
6-17 Model Runtime CVs: Initial 3MRA Time Trial Analysis for Selected Chemicals 6-71
6-18 3MRA Version 1.0 Model Runtimes By Chemical 6-72
6-19 3MRA Version 1.0 Model Runtimes By Source Type 6-73
6-20 Breakdown of Average 3MRA MMSP Runtime by Science Module 6-74
6-21 Ordering of 3MRA Science Module Runtimes for 43 Chemicals 6-75
6-22 Allocation of SuperMUSE Computing Resources To Date 6-76
7-1 Benzene Disposal in Landfills: Uncertainty for Sum of All Ingestion and Inhalation
Pathways - Pseudo 2nd Order Analysis for 50% Uncertainty (H) and 98% Associated
Confidence (I) Due to OSE 7-9
7-2 Benzene Disposal in Landfills: Uncertainty for Sum of All Ingestion and Inhalation
Pathways - Pseudo 2nd Order Analysis for 50% and 95% Uncertainty (H) and 98%
Associated Confidence (I) in 50% Uncertainty (H), Due to OSE 7-10
7-3 Benzene Disposal in Landfills: Exposure Pathway Analysis for Increased
Human Cancer Risk 7-11
xiv
-------
Table of Contents
List of Tables
Table Page
2-1 Probabilistic Risk Assessment Terminology 2-61
2-2 3MRA Population Examples 2-62
2-3 Types of Quantities Used in Models 2-62
2-4 Components of Model Evaluation 2-63
2-5 Sources of Uncertainty in Empirical Quantities 2-63
2-6 Types of Uncertainty 2-64
2-7 Terms Used for Describing Variability and Empirical Uncertainty 2-64
2-8 Sources of Uncertainty in Empirical Quantities 2-65
2-9 EPA Policy: "Conditions for Acceptance" of Probabilistic Risk Assessments 2-66
3-1 Data Collection Approach by Data Type 3-24
3-2 Chemical Properties Defined by Stochastic Distributions 3-25
4-1 Conceptual Model Elements Considered in 3MRA Version 1.0 4-32
4-2 3MRA Version 1.0 Module Input and Output Dictionary Files 4-33
4-3 3MRA Version 1.0 Waste Management Unit Type Distribution 4-34
6-1 Spearman-Rank Correlations for 3MRA Version 1.0 "Num" Type Parameters 6-68
6-2 R2 Values for Best Multiple Regressions Among "Num'Type Parameters 6-68
6-3 3MRA Version 1.0 Database: Organic Chemicals and Metals 6-69
7-1 Benzene Disposal: Uncertainty Analysis for Summation of All Ingestion and
Inhalation Pathways 7-12
8-1 Summary of 3MRA Version 1.0 Model Input Dictionaries 8-5
8-2 Summary of 3MRA Version 1.0 Model Output Dictionaries 8-5
8-3 Summary of 3MRA Version 1.0 Model Input and Output Data Types 8-6
8-4 Summary of 3MRA Version 1.0 Model Input and Output Dimensionality 8-6
8-5 Summary of National, Regional, and Site-based 3MRA Model Inputs 8-7
8-6 Summary of 3MRA Model Inputs Derived From Multiple Database Sources 8-7
8-7 Summary of 3MRA Model Input Distribution Types By Module 8-8
8-8 Summary of 3MRA Model Inputs With Multiple Distribution Types 8-9
8-9a SUI Header Input (SSF) Dictionary Summary 8-12
8-9b Site Lay out Input (SSF) Dictionary Summary 8-14
8-9c Chemical Properties Input (SSF) Dictionary Summary 8-19
8-9d Aerated Tank (AT) Input (SSF) Dictionary Summary 8-24
8-9e Land Application Unit (LAU) Input (SSF) Dictionary Summary 8-25
8-9f Landfill (LF) Input (SSF) Dictionary Summary 8-26
8-9g Surface Impoundment (SI) Input (SSF) Dictionary Summary 8-27
8-9h Waste Pile (WP) Input (SSF) Dictionary Summary 8-28
8-9i Aquifer (Saturated Zone) Input (SSF) Dictionary Summary 8-29
xv
-------
Table of Contents
List of Tables (continued)
Table Page
8-9j VadoseZone Input (SSF) Dictionary Summary 8-29
8-9k Air Input (SSF) Dictionary Summary 8-30
8-91 Watershed Input (SSF) Dictionary Summary 8-31
8-9m Surface Water Input (SSF) Dictionary Summary 8-32
8-9n Aquatic Foodweb Input (SSF) Dictionary Summary 8-33
8-9o Farm Foodchain Input (SSF) Dictionary Summary 8-34
8-9p Terrestrial Foodweb Input (SSF) Dictionary Summary 8-35
8-9q Human Exposure Input (SSF) Dictionary Summary 8-36
8-9r Human Risk Input (SSF) Dictionary Summary 8-38
8-9s Ecological Exposure Input (SSF) Dictionary Summary 8-38
8-9t Ecological Risk Input (SSF) Dictionary Summary 8-39
8-10a Site Layout Output (GRF) Dictionary Summary 8-39
8-1 Ob Source Output (GRF) Dictionary Summary 8-40
8-10c Aquifer (Saturated Zone) Output (GRF) Dictionary Summary 8-41
8-10d Vadose Zone Output (GRF) Dictionary Summary 8-41
8-10e Air Output (GRF) Dictionary Summary 8-42
8-10f Watershed Output (GRF) Dictionary Summary 8-43
8-10g Surface Water Output (GRF) Dictionary Summary 8-43
8-1 Oh Aquatic Foodweb Output (GRF) Dictionary Summary 8-44
8-10i Farm Foodchain Output (GRF) Dictionary Summary 8-45
8-lOj Terrestrial Foodweb Output (GRF) Dictionary Summary 8-46
8-10k Human Exposure Output (GRF) Dictionary Summary 8-48
8-101 Human Risk Output (GRF) Dictionary Summary 8-51
8-10m Ecological Exposure Output (GRF) Dictionary Summary 8-52
8-1 On Ecological Risk Output (GRF) Dictionary Summary 8-53
8-11 Partial List of Internal Variable Descriptions in 3MRA Version 1.0 8-54
xvi
------- |