vxEPA
United States
Environmental Protection
Agency
Stochastic Human Exposure
and Dose Simulation Model for
Multimedia, Multipathway
Chemicals
SHEDS-Multimedia Model
Version 3
User Guide
-------
-------
EPA600/R-08/118
September 2008
www.epa.gov
Stochastic Human Exposure
and Dose Simulation Model for
Multimedia, Multipathway
Chemicals
SHEDS-Multimedia Model
Version 3
User Guide
Casson Stallings1
Valerie Zartarian2
Graham Glen1
1Alion Science and Technology, Inc.
2US Environmental Protection Agency
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
-------
TABLE OF CONTENTS
LIST OF FIGURES iv
LIST OF TABLES vi
Acknowledgments vii
Display Issues, Disclaimer, and Support viii
ACRONYMS AND ABBREVIATIONS ix
1 Considerations for New Users of SHEDS-Multimedia 1
1.1 Introduction 1
1.2 Entering Distributions 2
1.3 Configuring SHEDS to the Scenarios of Interest 3
2 Overview 5
2.1 Introduction 5
2.2 Using the Demonstration File 6
2.3 General Interface Hints 8
2.3.1 Display Issues 8
2.3.2 Grayed Out Buttons or Widgets 8
2.3.3 Entering Probability Vectors 8
2.3.4 Distributions Supported 11
2.3.5 Entering Distributions 11
2.3.5.1 Individual Distribution Widgets 11
3 Installation 13
3.1 Requirements 13
3.2 Installation under MS Windows 13
3.2.1 Starting With a CD 13
3.2.2 Starting with a Downloaded File 13
3.2.3 The Standard Installation Process 14
3.2.4 Starting the Model Interface 16
3.2.5 Removing SHEDS-Multimedia 16
4 The SAS User Interface 18
4.1 The SAS Screen 18
5 SHEDS-Multimedia: The Graphical User Interface 21
5.1 SHEDS-Multimedia Main Interface Screen 21
5.2 Specify Run Name and Files 22
5.2.1 Specify Run Name Dialog 22
5.2.1.1 Edit Selected Run 23
5.2.1.2 Copy Selected Run To New Run 23
5.2.1.3 Create a New Run 24
5.2.1.4 View Results of Selected Run 24
5.2.1.5 Delete Selected Run 24
5.2.2 New Run Name Dialog 24
5.2.3 Specify Files Dialog 25
-------
5.3 Specify Population and Sampling 25
5.4 Specify Simulation Information 27
5.4.1 Simulation Length 27
5.4.2 Source-To-Concentration Approach and Application Dates 27
5.4.3 Dermal Exposure Method 28
5.4.4 Soil Ingestion Approach 28
5.4.5 Exposure-To-Dose Method 28
5.4.6 Maximum Diary Event Length 28
5.4.7 Keep Intermediate Variables 28
5.4.8 Save the Log File 29
5.5 Specify Application Scenarios Simulated 31
5.5.1 Specify Exposure Scenarios 31
5.5.2 Specify Exposure Scenario Details 33
5.5.3 Specify Application Dates: Model Determined 34
5.5.4 Specify Application Dates: User Specified 36
5.5.5 Specify Co-Occurrence 36
Specify Application Times 38
5.5.6 Specify Re-entry Times 39
5.6 Specify Concentration-Related Inputs 39
5.6.1 Specify Decay and Dispersion Distributions 40
5.6.2 Specify Decay and Dispersion Distributions: Background 41
5.6.3 Specify Post-Application Distributions 42
5.6.4 Specify Post-Application Distributions: Background 43
5.6.5 Specify Time Series Inputs 44
5.6.5.1 Time Series File Format 44
5.7 Specify General Exposure and Dose Factors 45
5.7.1 Variability Distributions 46
5.7.2 Edit Variability Distributions 47
5.7.3 Correlating Input Variables 50
5.7.3.1 Select Correlated Variables 51
5.7.3.2 Specify Variable Correlations 52
5.7.3.3 Contact Probabilities 54
5.7.3.4 Edit Contact Probabilities 56
5.8 Run Simulation 57
5.9 View Results 59
5.9.1 Additional Outputs and Files 60
5.9.2 View Results for the Population 60
5.9.2.1 Summary Table 63
5.9.2.2 CDF 63
5.9.2.3 Box and Whiskers 65
5.9.2.4 Contribution by Pathway 66
5.9.3 View Results for an Individual 69
5.9.3.1 Time-Series 71
5.9.3.2 CDF 71
ii
-------
5.9.3.3 Box and Whiskers 71
5.9.3.4 Contribution by Pathway 72
5.9.3.5 Summary Table 72
5.9.3.6 Detailed Table 72
5.9.4 View Diary Pool Sizes 73
6 SHEDS-Multimedia: Batch Mode 74
6.1 Installing Batch Capability 74
6.2 Defining Inputs for a Batch Run 74
6.3 Submitting a Batch Run 75
6.4 Uncertainty Runs 76
6.5 Sensitivity Runs 77
Appendix A. Directories and Files A-l
A.I Directories A-l
A.2 Critical Input Files A-l
A.3 User Specified Output Files A-2
A.4 Other Output Files A-2
A.4.1 Diary Related Output Files A-2
A.5 Exporting SAS Datasets A-3
A.6 Moving Run Files A-3
A.6.1 Background on the Runlnfo Dataset A-3
A.6.2 Moving the Run Directory A-3
A.6.3 Integrating the Old and New Runlnfo Records A-4
A.6.4 Replacing the Runlnfo File (Dataset) A-4
A.6.4.1 Integrating Records into the New Runlnfo File A-4
A.6.5 Running the % Runlnfo AddDirs Macro A-4
A.6.6 Moving a Run: An Example A-4
Appendix B. Reference Tables B-l
B.I Variable Names B-l
B.2 Media Descriptions B-2
Appendix C. Probability Density Functions C-l
C.I Beta C-l
C.2 Exponential C-l
C.3 Gamma C-l
C.4 Lognormal C-2
C.5 Normal C-2
C.6 Point C-3
C.7 Triangular C-3
C.8 Uniform C-3
C.9 Weibull C-3
C.10 Discrete Probability Density Functions C-4
in
-------
LIST OF FIGURES
Figure 2-1. Overview of the SHEDS Multimedia Interface at Version 3 7
Figure 2-2. Sample of Two Buttons, One Normal and One Grayed Out (Bottom) 8
Figure 2-3. Example of List Box with a Blue Background Indicating There are More Groups to
Complete 8
Figure 2-4. Specify Re-entry Times Screen Containing Two Probability Vectors 9
Figure 2-5. Example Distribution Widget 12
Figure 3-1. Setup Screens: Initial Screen 14
Figure 3-2. Setup Screens: Welcome and Information Screen 14
Figure 3-3. Setup Screens: Installation. Directory Screen 15
Figure 3-4. Setup Screens: Shortcut Folder Name 15
Figure 3-5. Setup Screens: Confirmation Screen 16
Figure 3-6. Setup Screens: Completion Confirmation 16
Figure 4-1. The Main Model GUI Interface Screen in the SAS Window on Startup 19
Figure 4-2. The SAS Menu Bar and the View Menu 20
Figure 4-3. The SAS Results Window 20
Figure 4-4. The SAS Explorer Window 20
Figure 5-1. SHEDS Multimedia Main Screen 21
Figure 5-2. Specify Run Name Screen 23
Figure 5-3. Confirm Delete dialog 24
Figure 5-4. Enter New Run Name screen 24
Figure 5-5. Specify Files Screen 25
Figure 5-6. Specify Population and Sampling Screen 26
Figure 5-7. Specify Simulation Information Screen 30
Figure 5-8. Specify Exposure Scenarios Screen 32
Figure 5-9. Specify Exposure Scenario Details Screen 34
Figure 5-10. Specify Application Dates: Model Determined Screen 35
Figure 5-11. Specify Application Dates: User Specifed Screen 36
Figure 5-12. Specify Co-Occurrence Screen 37
Figure 5-13. Specify Application Times Screen 38
Figure 5-14. Specify Re-entry Times Screen 39
Figure 5-15. Specify Decay and Dispersion Distributions 41
Figure 5-16. Specify Decay and Dispersion Distributions: Background Screen 42
Figure 5-17. Specify Post-Application Distributions Screen 43
Figure 5-18. Specify Time Series Inputs Screen 45
Figure 5-19. Variability Distributions Screen 46
Figure 5-20. An example of the Edit Variability Distributions Screen 49
Figure 5-21. Variable with Two Conditions shown 50
Figure 5-22. Select Correlated Variables Screen 52
Figure 5-23. Error Displayed if Zero or One Variables are Selected for Correlation 52
Figure 5-24. Error Displayed if No Variable Pairs are Selected for Correlation 52
Figure 5-25. Specify Variable Correlations Screen 54
Figure 5-26. Contact Probabilities Screen 56
iv
-------
Figure 5-27. Run Simulation Screen 57
Figure 5-28. Log Screen after Running a Small Simulation 59
Figure 5-29. View Results Screen 60
Figure 5-30. View Results for the Population Screen 62
Figure 5-31. Example Summary Table for a Population 63
Figure 5-32. Example CDF for a Population 64
Figure 5-33. Example Population Box and Whiskers Plot 66
Figure 5-34. Example Pie Chart Showing the Contribution by Pathway for a Population 68
Figure 5-35. View Results for an Individual Screen 70
Figure 5-36. Example Time-Series for an Individual 71
Figure 5-37. Example Pie Chart Showing the contribution by pathway for an individual 72
Figure 5-38. Example Detailed Table for an Individual 73
Figure 5-39. Example Diary Pool Size Table 73
-------
LIST OF TABLES
Table 2-1. Probability Vector Errors Encountered 10
Table 2-2. Supported Variable Distributions 11
Table 5-1. Output Variable Group and Dose/Exposure Terminology Differences 62
Table B-l. Variable Names From Distributions File B-l
Table B-2. Media Descriptions And Symbols B-2
VI
-------
Acknowledgments
We would like to thank Dana Vogel, Steve Nako, David Hrdy, and Phillip Villanueva of the U.S.
Environmental Protection Agency, Office of Pollution Prevention and Toxic Substances, Office
of Pesticide Programs and Eric Hall of the U.S. Environmental Protection Agency, Office of
Research and Development, National Exposure Research Laboratory for providing comments on
the draft User Guide.
vn
-------
Display Issues, Disclaimer, and Support
Display Issues
It is strongly advised that the SHEDS user maximize windows so that as much as possible of the
full dialog is displayed. You may still need to scroll down to see all of some dialogs. Refer to the
images in this User Guide to ensure that all components of a dialog are displayed.
Disclaimer
This is SHEDS-Multimedia version 3.15, an aggregate residential version of the EPA's
Stochastic Human Exposure and Dose Simulation model for multimedia, multipathway
pollutants that includes the inhalation, dermal, hand-to-mouth, object-to-mouth pathways and a
graphical user interface. This version also includes a sample database with hypothetical inputs
for test and demonstration purposes. All input parameters used in the SHEDS-Multimedia model
must be determined independently by the researcher and substantiated by references so that the
model results can be evaluated based on the validity of the data sources.
The United States Environmental Protection Agency, through its Office of Research and
Development, developed and funded the aggregate residential SHEDS-Multimedia version 3.15
model with assistance from contractor Alion Science and Technology. This model has been
subjected to Agency and external peer review. Work is underway for development of SHEDS-
Multimedia version 4, which will extend the aggregate (single chemical) algorithms to
cumulative (multi-chemical), incorporate a dietary (food and drinking water pathway) model,
include a case study, and add help screens and other enhancements.
Support
Please contact one of the following individuals with any questions, comments, or specific
suggestions related to this beta version of the SHEDS model:
Dr. Valerie Zartarian, (617) 918-1541, zartarian.valerie@epa.gov
Dr. Jianping Xue, (919) 541-7962, xue.jianping@epa.gov
Vlll
-------
ACRONYMS AND ABBREVIATIONS
CDFs - cumulative distribution functions
CHAD - Consolidated Human Activity Database
EPA - United States Environmental Protection Agency
GI - gastrointestinal
GM - geometric mean
GSD - geometric standard deviation
GUI - graphical user interface
NERL - National Exposure Research Laboratory
ORD -Office of Research and Development
PBPK - Physically-based pharmacokinetic
PDFs - probability density functions
SHEDS - Stochastic Human Exposure and Dose Simulation
ug (in SAS printout or variable names) - microgram
IX
-------
1 Considerations for New Users of SHEDS-
Multimedia
1.1 Introduction
SHEDS-Multimedia version 3 (also referred to as 'SHEDS') is a sophisticated but user-friendly
aggregate human exposure model for chemicals contacted in a residential setting. It requires the
user to select appropriate inputs and to interpret the resulting outputs. Users should therefore be
prepared to invest time to configure this model to exposure scenario(s) of their choice. Before
attempting to run the model, it is recommended that new users review this User Guide and the
Technical Manual to familiarize themselves with the model and the types of information that will
be required.
The SHEDS installation package includes the User Guide, Technical Manual, annotated SAS
code for use via the graphical user interface (GUI), Consolidated Human Activity Database
(CHAD) activity diary and population files, a height/weight data set, and default input
distributions. The installation process is explained in this User Guide, which is intended to guide
the user through the GUI. The interface has three functions: specifying inputs, executing the
model, and storing and viewing the results. The Technical Manual contains a detailed
description of the model's structure and algorithms.
The SHEDS model runs on a personal computer and requires that SAS 9.1 or higher be installed
prior to the installation of SHEDS. It has been run successfully on various laptops and desktops
using Windows XP. It has not yet been tested with Windows Vista. Though hardware
requirements are modest, a full-scale model run may take a substantial amount of time. As a
rough guide, a variability-only (1-stage Monte Carlo) run takes about 1 hour for every 1000
persons being simulated and a typical run may be several thousand persons. Uncertainty (2-stage
Monte Carlo) runs are much slower and may take several days to complete.
SHEDS is a stochastic model in which each simulated individual is different. There are
approximately 100 variables that are sampled randomly for each person; many of these variables
require multiple samples per person as they change over time throughout the simulation period.
Increasing the number of simulated individuals provides a better characterization of the
population under study. SHEDS version 3 uses new random number seeds each time the model
is run. This has one main benefit and one main drawback. The benefit is that results may be
combined across runs with the same input settings: for example, two runs of 5000 persons each
(with the same input settings) are equivalent to a single run of 10,000 persons and the outputs
from the shorter runs could be directly combined. The drawback is that model runs are not
reproducible. (The option to reproduce an earlier run exactly is planned for version 4.)
SHEDS separately constructs time series for the environmental concentrations in each simulated
person's house, as well as an activity sequence through time based on the selection of human
activity diaries from EPA's Consolidated Human Activity Database (CHAD;
1
-------
http://www.epa.gov/chadnetl). These are combined, using exposure pathway-specific equations
and exposure factors sampled from user-specified distributions, to generate a time series of
exposure for each simulated individual. The time step is variable and depends on the event
duration from the activity diaries, ranging from one minute to one hour. Therefore, a 1-year
simulation will have a very large number of such events (generally 10,000 to 20,000) for each
simulated person. These results are automatically aggregated over time to produce (for example)
daily exposure totals, although for some purposes the finer time resolution may be useful. Due to
their size, the event-level exposure time series are not usually saved as permanent output; as a
rule only summary statistics on daily and longer periods are saved.
SHEDS does not attempt to model the exposures of particular individuals. Instead it randomly
creates a population of simulated individuals who collectively represent the target population.
Thus, an appropriate result of a model run might be the statement 'SHEDS indicates that 5% of
the target population receives an exposure in excess of ...', but it is not appropriate to say that a
specific real-world individual receives any particular exposure.
SHEDS has a graphical user interface (GUI) that guides the user through the various input
screens. The GUI comes with a demonstration scenario. It may be useful to copy this under a
new name and to make changes to this copy. This is described in more detail later in this
manual.
SHEDS version 3 allows only one chemical per model run, although the chemical can effectively
be changed by altering appropriate inputs. The user may select several product categories (each
containing the same chemical) to be analyzed together in the same run. The model permits the
use of co-occurrence factors that control the likelihood of the various products being used
together. The human exposure results are automatically aggregated across the products used in
the given model run.
1.2 Entering Distributions
Many SHEDS inputs are randomly sampled from user-specified distributions. SHEDS accepts a
wide variety of continuous distributions as detailed in Appendix C. Some inputs require
probability vectors (see the Appendix and section 4.1.1 of the Technical Manual). For
continuous distributions, the user selects the type from a pull-down menu and then specifies the
desired parameters. For discrete distributions (probability vectors) the user enters the probability
of each outcome.
The GUI will highlight invalid choices with a yellow background. This usually results from
numeric parameters being specified that are incompatible with the type of distribution. For
example, the lognormal distribution requires a geometric mean greater than zero and a geometric
standard deviation greater than one. A missing numeric parameter will also result in the yellow
error indicator. For probability vectors, the sum of the probabilities must be one or else an error
will be indicated.
-------
The user has the option of truncating distributions at one or both ends by specifying minimum
and/or maximum values. If the random number generation produces values outside these limits
then the values are automatically reset to the maximum or minimum, depending on which was
exceeded.
1.3 Configuring SHEDS to the Scenarios of Interest
The user has the ability and the responsibility to configure SHEDS to a particular scenario(s) of
interest. This includes specifying the target population, the simulation period, the chemical and
application method(s) of interest, and the distributions for many model parameters. As an
example, SHEDS supports three methods of determining the chemical concentrations: user-
supplied time series, a decay/dispersion model based on chemical application dates, or a post-
application decrease based on time intervals. The user must choose which one is most
appropriate for the problem at hand. These three methods are described in more detail in the
SHEDS Technical Manual.
The SHEDS installation package comes with example input distributions for demonstration
purposes, to help orient users in how to run the model. The user is strongly cautioned that these
values were not created with any specific scenario in mind and may not be appropriate for a
particular chemical or scenario of interest. These values can be used as a starting point for
becoming familiar with the GUI, or modified by the user as appropriate.
SHEDS is designed to estimate human exposure in a residential setting, but is not geared to any
specific chemical. The user customizes the model run for the chemical of interest by setting
appropriate input parameters, for example, product types and usage frequencies, initial
application amounts, and concentration decay rates. The model allows the user to select one or
more product application methods from a pre-determined list in the GUI (which can be modified
in the SAS code).
The SHEDS GUI allows the user to select the gender and age range of the target population. It is
worth re-emphasizing that SHEDS is a population-based model. Therefore it would be
appropriate for simulating exposure for school-age children in general, for example, but should
not be expected to be reliable for simulating a specific child or even a specific set of children.
Beyond age and gender, the user may target populations by selecting input distributions that
reflect characteristics of those groups. For example, pica children could be simulated by
specifying a large soil ingestion rate.
The 'simulation period' is the user-specified time over which the exposure is tracked for each
simulated individual. The model allows this to range from one day to several years, but the user
should be aware of certain considerations. For example, if the user supplies concentration time-
series data, then the period is limited to the extent of this data. Note that the model does not
change the age of individuals, and therefore simulation periods longer than one year may not be
advisable for children. Finally, longer simulations require longer run times.
-------
SHEDS calculates exposure from several different pathways. If desired, the user may restrict the
model to selected pathways by judicious specification of inputs. For example, ingestion of
chemical via object mouthing can be 'turned off simply by setting the distribution for object-to-
mouth contact frequency to a point value of zero. Similarly, other pathways such as inhalation,
hand mouthing, direct soil ingestion, dermal absorption, and GI tract absorption could be
eliminated.
There is no facility to supply differing inputs by region within a single run in SHEDS version 3.
To model regional variation the user would have to construct a separate model run for each
region with distinct input distributions. The distributions for each run would then reflect the
variation within the given region (but not across regions). For example, pesticide usage may
differ in warm and cold climates. Seasonal variation may be handled in a similar manner. By
bypassing the GUI, advanced SAS users have the option of incorporating seasonal variation in
distributions within a single model run.
-------
2 Overview
2.1 Introduction
Reliable models for assessing human exposures are important for understanding health risks from
chemicals. The Stochastic Human Exposure and Dose Simulation model for multimedia,
multiroute/pathway chemicals (SHEDS-Multimedia), developed by EPA's Office of Research
and Development (ORD), National Exposure Research Laboratory (NERL), is a state-of-science
computer model for improving estimates of aggregate (single-chemical, multi-route/pathway)
and cumulative (multi-chemical, multi-route/pathway) human exposure and dose. SHEDS-
Multimedia is the EPA/ORD's principal model for simulating human exposures to a variety of
multimedia, multipathway environmental chemicals such as pesticides, metals, and persistent
bioaccumulative toxins. Exposure is defined in SHEDS-Multimedia as the contact between a
chemical agent and a simulated human target at the skin, lung, and gastrointestinal tract exposure
surfaces. Dose is defined in SHEDS-Multimedia as the amount of chemical that enters the target
after crossing the exposure surfaces.
SHEDS-Multimedia is a physically-based, probabilistic model that predicts, for user-specified
population cohorts, exposures incurred via inhaling contaminated air, touching contaminated
surface residues, and ingesting residues from hand- or object- to-mouth activities. To do this, it
combines information on chemical usage, human activity data (e.g., from time/activity diary
surveys and videography studies), environmental residues and concentrations, and exposure
factors to generate time series of exposure for simulated individuals. One-stage or two-stage
Monte Carlo simulation is used to produce distributions of exposure for various population
cohorts (e.g., age/gender groups) that reflect the variability and/or uncertainty in the input
variables. While the core of SHEDS-Multimedia is the concentration-to-exposure module, there
are various options (built-in source-to-concentration module; user-entered time series from other
models or field study measurements) for obtaining concentration inputs, and SHEDS-Multimedia
exposure outputs can be used as inputs to physically-based pharmacokinetic (PBPK) models.
For a more detailed technical description of the SHEDS model, please refer to the SHEDS-
Multimedia version 3 Technical Manual. The purpose of this User Guide is to assist the SHEDS-
Multimedia user in navigating through the graphical user interface to apply SHEDS-Multimedia.
Figure 2-1 provides an overview of the SHEDS-Multimedia user interface. The major interface
screens are each represented with one box. Not all screens are represented in Figure 2-1. In
general, the user will navigate through the interface from left to right and top to bottom as shown
on the overview. The exact screens visited will vary depending on the type of run being defined
and data required for that run. Typically, the interface will prevent the user from moving ahead
unless all data for the current step have been entered.
The user interface guides the user through a set of screens that define the inputs needed for a
SHEDS run. When a new run is created, most of these screens contain default choices, or
choices from a previous model run, that the user may either edit or accept. The best way to run
SHEDS is to carefully prepare the inputs for an initial 'baseline' run; thereafter, one can start
5
-------
with the baseline run and make a small number of changes from run to run, greatly speeding up
the process.
Some sections will need to be revisited a number of times to enter all information. In particular
the sections defining application details and dates, and those defining media concentrations will
need to be defined for each application or scenario type being simulated. The screens used to
enter variability distributions and contact probabilities will need to be revisited for each group of
variables being defined.
Every effort has been made to reduce the amount of information entered by the user. For
instance, if dermal transfer efficiencies are being used, then the user will not be presented with
the opportunity to enter data on dermal transfer coefficients. Likewise, if the decay and
dispersion module is being used to model media concentrations, users will only enter information
relevant to that and not to time-series or post-application distributions.
2.2 Using the Demonstration File
When installed, SHEDS-Multimedia version 3 comes with a completed sample run. This sample
run is named "Demonstration File" and is meant to assist the user in becoming familiar with the
interface. The Demonstration File simulates up to four scenarios for each individual. These
include a crack and crevice application along with pet, lawn, and garden applications. The
simulations last a year for each individual, use decay and dispersion, direct ingestion, transfer
coefficients and assume a number of input variables are correlated. The run supplied contains
results for 100 people. The user may change this value and re-run the simulation if a different
number is desired. A typical production run will require substantially more individuals.
It is very important to note that the default input values defined in the Demonstration File are not
meant to represent any specific chemical or situation. They are supplied for demonstration
purposes and should not be used for production runs.
The results from the file are available on installation and can be viewed by selecting the View
Results of Selected Run button on the Specify Run Name dialog (described on page 22). While
the inputs in the Demonstration File are safe when you use the View Results button, changes
made after selecting the file with Edit Selected Run will overwrite the original values. It is
strongly recommended that the user not edit the original Demonstration File. To view and edit
inputs from the demonstration, go to the Specify Run Name dialog (described on page 22),
highlight "Demonstration File", and click on Copy Selected Run To New Run. You will be
prompted for a new run name. After entering the run name and visiting the Run Files screen you
will be free to make any desired changes without affecting the original Demonstration File.
Warning: The data supplied in version 3 of SHEDS-Multimedia is not meant to represent
any specific chemical or situation and should be used for demonstration purposes only.
-------
Main
Specify Run Name
Specify Population and
Sampling
Specify Simulation
Information
Specify Application
Scenarios
New or Copied Run
(if needed)
Specify Files
Specify Exposure See
Specify Details _T"
L
narios (repeat for each application scenario)
Application Dates:
User Specified
Application Dates:
Model Selected
h
Co-Occurrence
(if needed)
Application Times
(if needed)
Application Re-
entry Times
Specify Concentration-Related Inputs (repeat for
each application scenario)
Time Series Inputs
Post-Application
Distributions
Background
Decay and
Dispersion
Distributions
Background
Specify General Exposure and Dose Factors
Variability
Distributions
Edit Distributions
Run Simulation
View Results
Individual Results
Population
Variability
Define correlated input variables (optional)
Select Correlated
Variables
Specify Variable
Correlations
Edit Contact Probabilities
Contact
Probabilities
Edit Contact
Probabilities
Figure 2-1. Overview of the SHEDS Multimedia Interface at Version 3.
-------
2.3 General Interface Hints
2.3.1 Display Issues
Display issues can arise when SAS does not have room to display the entire dialog. When a
screen does not initially draw correctly either the top title text or the bottom left button
(Continue, or Return) do not display. If in doubt compare the display to the appropriate figure in
this manual. The easiest method of minimizing these issues is to maximize the main SAS
window. On smaller monitors some screens may still not display fully. The screen's scroll bars
should be used to view the bottom of the screen. Unfortunately, SAS does not always display the
scroll bars automatically. Resizing the SHEDS dialog will force SAS to redraw the dialog and
include scroll bars if necessary.
2.3.2 Grayed Out Buttons or Widgets
Occasionally buttons will be grayed
out (displayed with muted text, see
Figure 2-2). This may mean that a
function has not been implemented. In
version 3 the Help and About buttons
have not been implemented.
Figure 2-2. Sample of Two Buttons, One Normal
and One Grayed Out (Bottom).
Groups Completed
In many cases a grayed out button means that additional steps are required before the user is
permitted to enter certain data. This is
particularly true on the main screen where data entry must be
completed sequentially, by starting at the top button and working
down. There are also several screens where the user must complete
data entry or editing for each group of variables listed before
continuing. In these cases, the groups are listed on the left and all
completed groups are listed on the right. If the groups remain
uncompleted, the completed list is highlighted in a light blue
(Figure 2-3).
[D irect-iriqestion-related
'Other
Activity-related
Chemical properties
Baths
Transfer-related
Dose-related
Another reason a button may be grayed out (Figure 2-2) is due to a
current error on the screen. Typically, the user is not allowed to
continue or navigate away from a screen when there are data entry
errors on the screen. These errors will always be identified by
yellow or red highlighting and often accompanied with a more
specific error message.
Figure 2-3. Example
of List Box with a Blue
Background
Indicating There are
More Groups to
Complete.
2.3.3 Entering Probability Vectors
There are a number of places in the interface where probability vectors must be entered. These
are used to enter probabilities associated with a set of mutually exclusive outcomes. Each uses a
similar graphical interface as can be seen on the Specify Re-entry Times screen (Figure 2-4). This
8
-------
screen contains two probability vectors where the user enters information on how long after a
treatment the simulated individual stays out of the treated area. Each probability vector is
composed of:
A container box with a descriptive title. In the case of the top probability vector this title
is 'Re-entry Time to Treated Indoor Area (probabilities for hours)';
A series of labeled text entry boxes. In this case 9 of them. Each is labeled and contains a
probability;
A space below the text entry boxes where error messages may be displayed; and
An OK button used to validate the entry.
Each text box must contain a probability, thus a number between 0 and 1 inclusive. The total of
all probabilities must sum to one. The user can enter values in all but the last box (the one with a
bold outline). The last box
automatically displays the amount
Specify Application Scenarios
Specify Reentry Times
Reentry Time to Treated Indoor Area (probabilities for hours)
0 1 2 3-4 5-6 7-12 13-24 25-48
~^[ ir if if q[
Reentry Time to Treated Outdoor Area (probabilities for hours}
0 _ 1 _ 2 _ 3-4 5-6 7-12
13-24 25-48
49-96
necessary to sum all probabilities in
the vector to 1. When an
inappropriate value is entered by the
user, the background will turn yellow
and an error message will be
displayed. The probability vector as a
whole will be validated, and the sum
recalculated every time the user uses
the return key or when the OK button
is clicked. Examples of possible error
messages are shown in Table 2-1.
The error messages are not displayed
in some contexts. The text boxes are
always highlighted in yellow. If there
are errors with individual probabilities or with the sum of probabilities, the user will not be
allowed to continue to the next screen. The errors must be corrected and the vector validated, or
the Cancel button must be used. No values will be saved if the Cancel button is used.
Cancel
Return
Figure 2-4. Specify Re-entry Times Screen
Containing Two Probability Vectors.
-------
Table 2-1. Probability Vector Errors Encountered
Error
The value
entered was not
a number.
A negative
value was
entered.
A value outside
the range of 0 to
1 was entered.
The sum of all
values entered is
too large.
Sample
Display
Reentry Time to Treated Indoor Area (probabilities for hours)
0123-4 5-6 7-12 13-24 25-48 49-96
1
0
Invalid Number.
0
0 0 0 0 I 0|
OK
Reentry Time to Treated Indoor Area (probabilities for hours)
0123-4 5-6 7-12 13-24 25-48 49-96
1
0
Invalid Number.
0
-.1 0 0 0 0 I 0|
iOKi
Reentry Time to Treated Indoor Area (probabilities for hours)
0123-4 5-6 7-12 13-24 25-48 49-96
0
0
5 urn of probabilities
0
1.5 0 0 0 0 -0.5
lOKI
must be 1 .
Reentry Time to Treated Indoor Area (probabilities for hours)
0123-4 5-6 7-12 13-24 25-48 49-96
.8
.1
Sum of probabilities
.1
.30000 -0.3
iOK!
must be 1 .
10
-------
2.3.4 Distributions Supported
Since the SHEDS Multimedia model is a stochastic model, a large number of its input variables
are defined by distributions rather than constants. The model supports point values (constants)
and 8 distributions. The distributions, parameter names, and rules for the parameters are given in
Table 2-2. Detailed information on the distributions and their parameters can be found in the
SHEDS-Multimedia Technical Manual.
Table 2-2. Supported Variable Distributions
Distribution
Point
Uniform
Normal
Log Normal
Triangle
Exponential
Gamma
Beta
Weibull
Number of
Parameters
1
2
2
2
3
2
2
2
2
Parameter Names
Value
Minimum, Maximum
Mean, Standard
Deviation
Geometric Mean
(GM), Geometric
Standard
Deviation(GSD)
Minimum, Mode,
Maximum
Minimum, Mean
Shape, Scale
Shape 1, Shape2
Shape, Scale
Rules for Valid Distribution
minimum < maximum
standard deviation > 0
GM>0, GSD>1
minimum <= mode <= maximum,
minimum < maximum
minimum < mean
shape>0, scale>0
shape 1>0, shape2>0
shape>0, scale>0
2.3.5 Entering Distributions
2.3.5.1 Individual Distribution Widgets
Distribution widgets are an interface object used to enter parametric distributions. They may be
found on various interface screens (e.g., Figure 5-15, Figure 5-16, and Figure 5-17 on pages 41,
42, and 43, respectively) and on the Edit Variables screen which also includes visualizations of
the resulting distributions (Figure 5-20). The distribution widget (Figure 2-5) is composed of:
A container box with a descriptive title.
A pull-down combo box with a list of distributions.
A series of labeled text entry boxes. The number of text entry boxes varies with the
number of parameters required.
An OK button to force validation.
11
-------
Enter a distribution by clicking on the pull-down arrow and making a selection from the list that
is displayed. Then enter values for each parameter. Pressing the tab key will move from
parameter to parameter. However, you will need to either press the enter key or click the OK
button to force the values to be validated. If the one or more of the values are incorrect when
validated, the background of the container box will turn yellow and you will not be permitted to
go to the next screen. Truncation minimum or maximums3 may be left undefined. The text box
containing these values will indicate an error, but the distribution as a whole will still validate.
Initial Residue/Concentration of Vegetable Garden (ug/cm2)
Distribution
"NORMAL T]
Mean
1 0.52
Std. Dev.
F 0.8961
Minimum
[ If
Maximum
1.2)
[ OK |
Figure 2-5. Example Distribution Widget.
After entering corrections hit return or click on the OK button to force validation and clear errors.
In some contexts specific error messages will be displayed.
3 Truncation minimums and maximums are not part of the distribution definition. These include the minimum and
maximum on the normal distribution and the maximum on the Log Normal, Exponential, Weibull, and Gamma.
12
-------
3 Installation
3.1 Requirements
SHEDS Multimedia was developed under SAS version 9.1 on machines running MS Windows
2000 and XP Professional.
To install and use SHEDS-Multimedia, you will need a computer running SAS version 9.1 or
higher. SAS must be installed prior to installing SHEDS. Your computer hardware needs to be
adequate to run SAS and MS Windows. Additionally you should have:
a 600 MHz processor,
64 MB of RAM, and
100 MB of free disk space.
However, it is recommended that you have:
a faster processor,
128 MB or more RAM, and
more than 200 MB of free disk space.
The software should run on other systems where SAS is implemented, but this has not been
tested.
Each run will consume about 16MB + (0.28*N) MB where N is the number of people simulated
for 1 year. So a run of 200 people will take up 72 MB of disk space.
3.2 Installation under MS
.... , WARNING: If you wish to save the results
Windows
Installation has been designed so that
administrative privileges are not required.
from model runs, copy the affected results
files from the installation directories to
another location before uninstalling or
reinstalling SHEDS.
3.2.1 Starting With a CD
If you have a CD, do the following:
1. Insert the CD into your CD reader.
2. Use Windows Explorer to navigate to the top level files on the CD.
Double click on the SHEDS_Multimedia_Setup.exe file to initiate the installation. The
installation may have version numbers after the "Setup". For instance .. .Setup_3.14.exe.
3.2.2 Starting with a Downloaded File
You may obtain the setup file via FTP or another electronic means.
13
-------
1. Save the attachment to a local or network hard drive. The method you use to do this will
vary depending on the program you use obtain the file.
2. Use the Windows explorer to navigate to the saved file.
3. Double click on the saved file to start the installation.
3.2.3 The Standard Installation Process
Once the installation wizard is initiated, simply follow the instructions. This will install the
necessary program and data files, create a program group on the Start menu, and create an icon
on the desktop. The desktop icon will execute the interface within SAS. The program group will
contain an additional shortcut (menu item) to uninstall the model and program data.
The screens encountered during install, and an explanation of each, are shown in the following
figures. For a default install, users should simply continue to click the Next buttons until the final
screen.
The initial screen informs the user what version
will be installed. Click the Next button to continue.
Welcome to the SHEDS Multimedia
Setup Wizard
install SHEDS Multimedia 3.14 on your computer.
It is recommended that you close all other applications before
continuing.
Figure 3-1. Setup Screens: Initial Screen.
Infaimalion
Please read the following important information before continuing
When you are ready to continue with Setup, click Next
Welcome to SHEDS Multimedia 3
Status
The United States Environmental Protection Agency through its
Office of Research and Development developed and Bonded die
SHEDS-Multimedia version 3 model with assistance from
contractor Mon Science and Technology. The model has not yet
been subjected to Agency review, but it has been approved for
dissemination for testing and research purposes. This working
version of the code represents an interim model not intended for
|
-------
Select Destination Location
Where should SHEDS Multimedia be installed?
I Setup will install SHEDS Multimedia into the following folder.
To continue, click Next. If you would like to select a different folder, click Browse.
:\Docurnents and Settings\cstallings\My Documents\Mu!timedia3.14 Browse...
At least 67.5 MB of free disk space is required
Back Nexl> Cancel
Figure 3-3. Setup Screens: Installation.
Directory Screen.
Select Start Menu Folder
Where should Setup place the program's shortcuts?
jjfrjl Setup will create the progtam's shortcuts in the following Start Menu folder.
To continue, click Next. If you would like to select a different folder, click Browse.
f Browse..
The user may elect to install the files
somewhere besides the default location. The
default location is in the user's My Documents
directory. If the user desires, the Browse button
can be clicked and a dialog allowing the choice
of an alternate location will come up. The main
reason one might want to install somewhere
else is that the user generated simulation results
are large and by default are stored under the
install directory. Note that the user may
redirect the output to another location.
Figure 3-4 shows the final window before
install begins. Allows the user to set the name
of the program group in the Start Menu. The
user should not need to change this.
Figure 3-4. Setup Screens: Shortcut Folder
Name.
15
-------
Ready to Install
Setup is now ready to begin installing SHEDS Multimedia on your computer.
Click Install to continue with the installation, or click Back if you want to review or
change any settings.
Destination location:
C;\Doeuments and SettingsScstallings\My Documents\Multimedia3.14
Start Menu folder
SHEDS Multimedia
The user has the chance to review installation
specifications on this screen before the
installation begins.
Figure 3-5. Setup Screens: Confirmation
Screen.
Completing the SHEDS Multimedia
Setup Wizard
Setup has finished installing SHEDS Multimedia on your
computer. The application may be launched by selecting the
installed icons.
Click Finish to exit Setup.
After the files have been extracted and placed in
the specified install directory, and the desktop
icon has been placed, the final screen will
indicate that the installation is complete.
Figure 3-6. Setup Screens: Completion
Confirmation.
3.2.4 Starting the Model Interface
The installation will place an icon on your desktop. The icon should appear as
the standard SAS icon, with the label "SHEDS Multimedia 3". However, this
varies depending on the version of SAS the user is running. It will typically
include an inverted triangle as part of the icon. Double click on the icon (figure
at right) to start SAS and the main screen of the SHEDS Interface.
3.2.5 Removing SHEDS-Multimedia
If SHEDS was installed using the installation wizard, the user can uninstall it in
a manner similar to other Windows programs. Removal of the program will
16
Figure 3-7.
SHEDS
Multimedia
Desktop Icon.
-------
remove all of the user's simulation output files unless these were saved elsewhere. To start
the removal process, click on the "Uninstall SHEDS Multimedia" choice in the SHEDS
Multimedia group of the Programs menu.
17
-------
4 The SAS User Interface
Most of the SAS user interface provides detailed fine-tuning capabilities that are usually not
necessary for typical model use. The average user may still find a brief review of this section
useful. If the user is unfamiliar with SAS and wishes to explore raw datasets used by the model,
then it is a good idea to read this section more carefully.
4.1 The SAS Screen
Assuming one uses the link in the MS Windows Programs menu or the screen icon to start the
model, the main GUI screen will start inside the main SAS window (Figure 4-1). By default, the
SAS log window also starts.
The SAS window is split into a number of distinct areas. The main area for viewing documents,
forms, data sets, and graphs is in the middle. This area may have multiple windows active at one
time. The user may activate a particular window by clicking on the title of the window. The bar
immediately below this area contains one button for each window in the main area. The buttons
indicate which window is active, and allow one to activate a different window. In Figure 4-1, the
main SHEDS screen is active as can be seen by the colored title bar and the depressed appearing
button. Note that since the main SHEDS screen does not have a title on the title bar, its button is
unlabeled. As the SHEDS GUI and model are run, informational and error messages will be
displayed in the log window. The pull-down menus are on the top of the screen, just under the
title. The toolbar is below this. At the very bottom of the SAS window is a status bar.
Additional SAS windows can be opened using the SAS View menu (Figure 4-2). The Graph
window displays all graphical output generated. The interface uses a separate output window for
this purpose so the Graph window will generally not be needed. It should also be noted, that
since the model interface overwrites images of the same type, SAS's graph window may appear
unreliable.
The SAS output window is where any tabular output generated is written by default. Generally,
the model does not provide this type of output.
The Results window (Figure 4-3) provides a list of tabular and graphical results generated and
printed to the graph and output windows. It is generally locked on the left side of the SAS
window. If additional windows are locked in this area, they are accessed with tabs that appear
below. The Results window may be used to navigate among the different outputs. However, it
will not open the output or graph windows; this must be done manually.
The Explorer window (Figure 4-4) is used to navigate through SAS libraries and files, data sets,
forms, and programs. It can be used to access the raw files used as input or output to the model.
This screen may float in SAS's main area or it may be locked on the left side of the SAS window
and be accessed via a tab on the bottom of this area. This screen may be toggled between tree
mode (shown) and a single pane, similar to MS Windows Explorer.
18
-------
SAS Environment
Contents of 'Common'
y Ǥ a
ffl B ai * ©
«g SAS Environment
- ;( Libraries
+ _# Airqual
+ S$ Common
* ^ Deraulfc
+ Jin
+ _j> Msps
I j Component:
_J PestsS
J Testcompor
+ _tl Sashelp
+ ,^ Sasuser
i||l File Shortcuts
- _^ Favunte Folders
+ -G^ AirtoxDev
+ i'^ AirtoxDevPfg
+ _^ f lulbmedia
+ ^ AirToxData
+ i^ AirToxRuns
+ i^ CTEPPOHV3
+ jat My Documents
+ JJB| My Desktop
+ [J My Computer
mj Components
Jj Testcomponents
O* Results iQ.1 Explorer
Specify Run Name and Files
'
NOTE: SftS (r) 9.1 (TS1M3) A
Licensed to MflNTECH ENUlRONMEMTftL TECHNOLOGY INC, Site 0027827003.
NOTE: This session is executing on the XP PRO platform.
NOTE: SftS 9.1.3 Service Pack 4
DTE: SflS initializat
real time
cpu time
1.67 seconds
0.59 seconds
5 ;HED; Multimedia . . I'D Log - (Untitled)
'I \Document-; and 5etrings\cstaings\My Douments\Projects\Muli:imedia\pests3
Figure 4-1. The Main Model GUI Interface Screen in the SAS Window on Startup.
19
-------
File View Tools Solutions Window Help
23 Enhanced Edi.tor
(j#) Program Editor
IH Output
i% Graph
I"? Results
l£j Explorer
fey Contents Only
fo^ Mv_ Favorite Folders
Figure 4-2. The SAS Menu Bar and the View Menu.
jjj Gplot: Absorption: (mg/kg)
Uj) Gplot; Absorption: (mg/kg)
_(S) Boxplot: Comparison Of Values Across Pathway
,jSJ Gchart: Percent Absorption: (mg/kg) By Pathw;
|r5p Gplot: Daily Time Series for Exposure: (mg/kg)
j£p Gplot: Daily Time Series for Exposure: (mg/kjg)
r£j Gplot: Daily Time Series for Exposure: (mg/kg)
r£j Gplot: CDF for Daily Values of Exposure: (mg/kj
lr§) Boxplot: Comparison Of Daily Values
chart: Percent Exposure: (mg/kg) By Pathway
Results
©I Explorer
Bt. ?:
-Inlx
5AS Environment Active Libraries
1§5 SAS Environment
B-
-------
5 SHEDS-Multimedia: The Graphical User
Interface
5.1 SHEDS-Multimedia Main Interface Screen
When SHEDS Multimedia is started the main screen will be displayed (Figure 5-1). This is the
main interface window that you will be returned to after completing each main step. It contains
the following buttons.
Specify Run Name and Files: Choose or define a new run name, identify existing input files
Specify Population and Sampling: Define basic information for the run, number of people
sampled, variability or uncertainty (only variability is active in the GUI for version 3),
ages of interest
Specify Simulation Information: Define the length of simulations, source-to-concentration
approach, and a number of other global simulation variables
Specify Post-Application Exposure Scenarios: Define the dates and times of applications, re-
entry times, and relationship between application dates
Specify Concentration-Related Inputs:
Specify decay and dispersion inputs, or
post-application distributions, or time-
series data
Specify General Exposure and Dose
Factors: Specify all other inputs
including transfer variables, use groups,
contact probabilities, input variable
correlations, etc.
Run Simulation: Run the current simulation
View Results: View results from previous runs
Help: Bring up the contents for the help
screens (to be implemented)
View Results
Exit
Version 3 Released June 11,2007
Figure 5-1. SHEDS Multimedia Main
Screen.
About: Bring up the help screen describing
21
-------
this version of the model (to be implemented)
Exit: Close the SHEDS user interface
These buttons allow the user to move through the proper sequence of steps to conduct a new
model run, or to view results and manage files from previous SHEDS simulations. Initially only a
few of the buttons are enabled. Buttons that are not enabled are displayed with a slight gray tint
to the font. The initially enabled buttons are Specify Run Name and Files and Exit.
Before beginning to specify the inputs for a new simulation run, editing the inputs for an existing
simulation, or viewing the results of a previous run, the user will need to specify the run name.
The run name is simply a title for a simulation run that references all input files for and output
files from a run.
Clicking on the "Specify Run Name and Files" button will bring up a screen allowing the user to
enter the run name. After specifying the run name and if necessary, the files (see below), the user
will be returned to the Main Window. If setting up a new run the Specify Population and
Sampling button will now be enabled. If viewing the results of a previous run, then the View
Results button will be enabled. The appropriate buttons will be enabled as the user completes
each step, and returns to the main menu.
5.2 Specify Run Name and Files
The first step in SHEDS-Multimedia version 3 is to click on the "Specify Run Name and Files"
button. The run name is like a file folder that keeps together all the information specific to a
simulation. In technical terms, it relates to a directory where files are stored and a record in a
database that stores information about a run.
5.2.1 Specify Run Name Dialog
The Specify Run Name dialog (Figure 5-2) allows the user to do several things:
Edit Selected Run: Restore the contents of a previous run for editing.
Copy Selected Run To New Run: Restore the contents of a previous run into a new run name
and allow them to be edited.
Create a New Run: Enter a new run name and begin defining inputs
View Results of Selected Run: Select run so that the user can view its results (edits will not be
allowed)
Delete Selected Run: Delete an existing run.
22
-------
The "Specify Run Name" dialog is composed of three main parts: The first is a list box
containing the names of previously defined simulation runs. The currently selected run name is
highlighted in this box. Below the list box, the description of the currently selected run name is
displayed. This provides additional information on the run. The user enters it when the run is
created. On the right of the dialog are the action buttons. These perform an action, usually on the
selected run name. As in other dialogs the "Help" button (when implemented) will bring up a
help screen for this dialog.
The Cancel button closes the
Specify Run Name
Select or Define the Run Name
Select A Defined Run Edit Selected Run
My New Run File
Copy Selected Run To New Run
Create A New Run
View Results of Selected Run
Delete Selected Run
Run Description
A 1 year simulation with applications to the house, pet, yard, and garden. Application
co-occurrence is set for outdoor applications. Removal efficiencies and transfer
coefficients are correlated.
Cancel
screen without selecting a
run name.
To select an existing run
name simply click on that
name in the Select A Defined
Run list box. In most cases
the user will need to select a
run name before clicking on
the action button.
5.2.1.1 Edit Selected Run
This button allows the user to
continue editing an existing
run. The user will be taken
to the "Specify Files" screen
(Figure 5-5). After that the
user will be returned to the
main menu. This feature is
useful, for example, if one is interrupted in the process of creating a run. The inputs already
entered may be saved and SHEDS exited. To finish creating the run, one selects the incompletely
specified run, chooses this "Edit" button, and continues defining the run. This button is also
useful in changing inputs on a previously completed run, if the user is willing to overwrite the
previous inputs and the resulting output. When becoming familiar with the model using the
Demonstration File it is recommended that you use the Copy Selected Run to New Run since any
changes made using this button overwrite the original file.
5.2.1.2 Copy Selected Run To New Run
A run name should be selected before clicking on this button. The information for that run will
be copied. The user will be allowed to define a new run name using the "Enter New Run Name"
screen (Figure 5-4) and will then be taken to the Specify Files Dialog and allowed to edit
information there. As distinct from the "Edit Selected Run" button, this button is used to modify
Figure 5-2. Specify Run Name Screen.
23
-------
inputs from a previous run, save the changes into a new run name, and still maintain the original
run.
5.2.1.3 Create a New Run
To create a new simulation from scratch, click on "Create A New Run". The "Enter New Run
Name" screen (Figure 5-4) will appear allowing the user to enter the new run name and
description. After this is completed, the "Specify Files" screen will be displayed (Figure 5-5).
Initially, defaults for all values will be used. Note that the default input values provided in
SHEDS-Multimedia version 3 are not intended to represent any specific chemical or situation.
5.2.1.4 View Results of Selected Run
Clicking this button chooses the currently selected run name and returns the user to the main
menu where the "View Results" button will be enabled. This will allow the user to view output
from the selected run. The user will not be allowed to view or edit run inputs in this model.
Confirm Run Name Delete
5.2.1.5 Delete Selected Run
Clicking this button deletes the selected
run name from the database. After a
dialog confirming that the user wants to
delete the run information (Figure 5-3)
the user is returned to this dialog.
5.2.2 New Run Name Dialog
The dialog shown in Figure 5-4 allows
the user to enter a new run name and a
directory path in which to store the
results. This screen is accessed when the
user clicks on either "Create A New
Run" or "Copy Selected Run To New
Run" from the Run Name Dialog. There
are three cases for the directory path.
When copying from another run, this box
shows the location of the prior run and
the user can change this. When creating
a new run, this box displays the storage location of the last run. If there is no prior run
(immediately after the installation), this box displays the default run location.
Delete the run named 'DeleteMe' and all associated data from the run info directory?
No I
Yes
Figure 5-3. Confirm Delete dialog.
Specify Run Name and Files
Enter New Run Name
Run Directory Location _
It: \Docui
pocumet
iments and SetrJngs\cstallings\My
!n&s\Projects\Multimedia\pests3\runs\
rNew Run Name_
[Untitled!
Cancel
Continue
Figure 5-4. Enter New Run Name screen.
The default run name is "UntitledN" where N is the next integer after the last stored untitled run.
The user may enter any valid Windows file names. After entering the run name, press return and
then click on the Continue button. If the name is the same as a previously used name an error will
be displayed and the user will be allowed to enter a new name. Clicking on continue will in
general take the user to the "Specify Files" Screen.
24
-------
Specify Run Name and Files
Specify Files (Libraries and Data Sets)
5.2.3 Specify Files Dialog
When creating, copying, or editing a run, the user will reach this dialog (Figure 5-5). This dialog
allows the user to verify the main directories and specify alternate dietary inputs.
The very top of the dialog simply displays the current run name and run directory. The run
directory should be similar to the run
name, but altered so that it is a valid
directory name. The
input and output files from the
simulation run will be stored in this
directory. The default data directory
is where the basic model input data is
stored.
The run description will be displayed
and may be edited in the next box
down.
The "Cancel" button will leave the
dialog without saving any changes.
The user will return to the Specify
Run Name Dialog. The "Save" button
will save any changes, and return the
user to the main menu.
Demonstration File
Run Directory
C:\Documents and Settings cstallings.My
ttocu men ts'ProjectS'MuIti media 'pests 3\runs\De monstration
Default Data Directory
C:\Documents and Settings cstallings My
locu merits Projects Multi media 'pests 3 data default.
Run Description (editable)
Cancel
Save
Figure 5-5. Specify Files Screen.
When the Specify Files dialog is initialized, the SAS libraries are also set up for the simulation
run or simulation results. If setup was for some reason not carried out correctly upon installation,
various errors may occur at this point. As of this current version, very little can be done to correct
these errors. In this case, reinstallation is recommended.
5.3 Specify Population and Sampling
Clicking on this button of the Main Window displays the related screen (Figure 5-6). Currently
only variability runs can be made through the interface. See the Technical Manual and the Batch
Mode section of this User Guide for information on doing sensitivity and uncertainty runs using
batch mode.
In the "Population: Age Groups and Sample Size" box, the user selects gender(s) and age(s) of
interest in the simulation (all ages, or some subset of those ages are available for selection). The
"Population Size" option allows the user to specify the sample size for the 1-stage simulation.
25
-------
Clicking on either "Males Only" or "Females Only" will clear all of the other gender allowing the
user to specify a single gender simulation. Once cleared, cohorts can be added back by clicking
on them.
The time taken to carry out the simulation increases linearly with the Size of Population and
Number of Populations selected. There is a lesser increase in run time if longer time-periods are
chosen. Depending on the computer hardware, a run of 1000 persons will typically take about an
hour to complete.
Specify Population and Sampling
Sampling Method.
Population: Age Groups and Sample Size.
Females Males
1 to
-------
5.4 Specify Simulation Information
This screen (Figure 5-7) allows the user to select a number of options that determine overall
simulation variables; each is discussed shortly. Clicking the "Save" button commits any changes.
The "Cancel" button prevents changes from being saved.
5.4.1 Simulation Length
The box labeled "SIMULATION START AND LENGTH" allows the user to specify the
beginning date (month, day, and year) and the number of days the simulation will track each
individual. One common choice of simulation period is to start on January 1st and continue for 1
year (365 or 366 days). However, the model allows a great deal of flexibility. A simulation may
begin on any day of the year and can be as short as one day or as long as desired and may cross
calendar years. However, caution should be exercised when specifying very long periods as they
take longer to run and the model does not alter many personal variable settings (including age)
over the simulation period.
5.4.2 Source-To-Concentration Approach and Application Dates
There are three options to generate residues and concentrations for scenario-relevant media:
Decay/Dispersion Model: Specific applications leave a residue on surfaces, in the soil,
and in the air. These residues decay with time and are moved to untreated areas.
Post-Application Distributions: Specific applications leave residues on various media.
The concentrations on the media are determined by distributions that change with the
time since the application.
User-Specified Concentration Time Series: The user supplies a time series of
concentrations for each medium.
The Decay/Dispersion model is a built-in source-to-concentration model that requires the user (in
subsequent screens) to enter application and decay rates, and the ratio of residue concentrations
in untreated to treated media for indoor applications. Additionally, background distributions can
be defined for the various media when Decay/Dispersion is used.
The Post-Application Distributions selection will require the user (in subsequent screens) to enter
distributions of relevant media residues and concentrations for discrete post-application time
periods (<1 day, 1-7 days, 8-30 days, 31-365 days).
The User-Specified Concentration Time Series option will require the user (in subsequent
screens) to enter time series of residues and concentrations for each medium of interest (e.g.,
from a measurement study or outputs from an external source-to-concentration model).
27
-------
If either Decay/Dispersion or Post-Application Distributions are chosen, a box will appear (to the
right of the Source-To-Concentration radio box in Figure 5-7) that allows the user to specify
whether the application dates will be User-Specified or Model-Determined. Details will be input
on subsequent screens. For user specified dates, the user will enter specific day numbers for each
application type, where day 1 is the first day of the simulation period. Applications will always
occur on the dates given. For modeled dates, the user will specify probability vectors for month,
day of week, number of applications, and time of applications.
5.4.3 Dermal Exposure Method
For dermal exposure, the user can select the Transfer Coefficient approach or the Transfer
Efficiency approach. These two methods are described in the SHEDS-Multimedia version 3
Technical Manual. Changing this value on a previously defined run will result in the user having
to visit the Specify General Exposure and Dose Inputs screen to define related variables.
5.4.4 Soil Ingestion Approach
For soil ingestion, the user can select the Direct or Indirect approaches, both of which are
described in the SHEDS Technical Manual. Changing this value on a previously defined run will
result in the user having to visit the Specify General Exposure and Dose Inputs screen to define
related variables.
5.4.5 Exposure-To-Dose Method
SHEDS-Multimedia version 3 allows the user to either use a built-in pharmacokinetic model
(discussed in the SHEDS Technical Manual) to compute absorbed dose, or to export SHEDS
exposure time series so that the files can be read by an external PBPK model.
5.4.6 Maximum Diary Event Length
The Maximum Diary Event Length box allows the user to break CHAD diary events into smaller
intervals, if desired. Diary events cannot exceed 1 hour. Average event duration while awake is
about 30 minutes. While carrying out simulations many stochastic decisions are made for each
event (e.g., which surface the individual contacts). The decisions apply to the entire event. To
force these decisions to be made more often, the diary can be broken up into more pieces, by
setting the Maximum Diary Event Length to a value smaller than 60 minutes. Using a smaller
value will slightly increase the run time for the simulation (see Technical Manual).
5.4.7 Keep Intermediate Variables
If the user checks the "Keep Intermediate Variables" box, SHEDS will retain variables used in
intermediate calculations for simulated individuals. This is useful to check the internal workings
of the model.
28
-------
5.4.8 Save the Log File
To save the log that is generated during the simulation run, the user can click on "Write to Log
File" which saves the log to a permanent file outside SAS, thus ensuring the log is saved even if
SAS crashes during the run. Unfortunately, under this option the standard SAS log screen is not
used, and the user cannot track the progress of the run.
If the log file is not saved, a line of output is written to the log window before each individual is
simulated, and this is the best way to tell how the simulation is proceeding. If the log is written
to a permanent file, the user can open the file with a text editor from time to time to check the
status of the run. The name of the log file is fixed. It is stored in the output library.
29
-------
Specify Simulation Information
SIMULATION START AND LENGTH .
Year ~* MT^h P'w
j«J
11 Jll
I1 ±j
Simulation Length (days)
366
SOURCE-TO-CONCENTRATION APPROACH
(* Decay/Dispersion Model
(~ Post-Application Distributions
f" User-Specifed Concentration Time Series
Application Dates
<" User-Specified Date
f*" Model-Determined Dates
Dermal Exposure Method
Help
SoillngestionMethod u^i.,
irr^HiiiiBBBH '
[indirect
P I
Export File
EXPOSURE-TO-DOSE METHOD h^n., ,m»^ ^H
lllfjTj'&'jS^^^^&ttti^^^^M
pKport Exposure Time Series
B
S ettingsScstallingsSMy
.Maximum Diarw Event.Lenath f mini ,
(60 ^
f~ Keep Intermediate Variables
Log File
I~ Write Log To File ICADocuments and Settings\cst=
IlingsSMy DocumentsSPr
Cancel
Save
Figure 5-7. Specify Simulation Information Screen.
Note: To utilize the co-occurrence
option, the user must select
Model-Determined Dates on this
screen. The user must also select
Specify Parameters before clicking
the Specify Inputs button on the
Specify Exposure Scenario Details
screen (Figure 5-9) for each
scenario for which co-occurrence
inputs are to be specified.
Application co-occurrence is
explained in section 5.5.5.
30
-------
5.5 Specify Application Scenarios Simulated
Clicking on "Specify Application Scenarios" on the main screen begins a series of screens
through which the user must navigate in their entirety before returning to the main screen. The
exact screens vary depending on options selected. The following screens may be visited:
The first screen is used to identify the scenarios to be simulated.
Specify Exposure Scenarios (Figure 5-8)
The second screen is used to specify details for each scenario and is used to get to the
Application Dates and Co-Occurrence screens for each scenario.
Specify Exposure Scenario Details (Figure 5-9)
These screens are visited once for each scenario, after which the user is returned to the Specify
Exposure Scenario Details screen.
Specify Application Dates (Figure 5-10 and Figure 5-11. The version visited depends on
date types specified)
Specify Co-Occurrence (Figure 5-12. Only visited if the user chooses this option)
Clicking the Continue button on the Specify Exposure Details screen takes the user to these
screens. They are visited once for each simulation.
Specify Application Times (Figure 5-13. Only visited for model determined application
dates)
Specify Re-entry Times (Figure 5-14)
5.5.1 Specify Exposure Scenarios
The specific scenarios or applications to be simulated are specified on this screen if the user is
simulating chemical applications. In the later screens, additional information will need to be
entered for each scenario defining the probability, dates and co-occurrence of applications, and
re-entry times. The exact screens visited will depend on options chosen by the user. If the user is
becoming familiar with the interface, it is recommended that only one or two scenarios be
chosen.
This screen will not be shown (the button on the main screen will be grayed out) if the User-
Specified Time Series was chosen as the source-to-concentration method. If the user is
simulating a multimedia measurements study then specific applications are not specified.
31
-------
Each scenario chosen is assigned a priority. The priority becomes important when application co-
occurrence is of interest. That is, when one is interested in using the dates of one scenarios'
applications to influence the dates of applications in other scenarios. The dates of an application
can only be influenced by applications occurring in scenarios given a higher priority.
The left most list box, titled "Application Type" indicates the supported application types or
scenarios. Clicking on an application in this list box copies it to the list of "Selected Scenarios"
in the middle of the screen. The scenarios are prioritized from top (1st) to bottom in the Selected
Scenarios list. Click on a scenario description to select it. Use the up and down arrows on the
right to change the priority of the selected scenario. Clicking on the "Delete Selection" button
will delete the currently selected scenario from the Selected Scenarios list.
When all desired scenarios are selected and prioritized, clicking on the Continue button will take
the user to the remaining screens.
Specify Application Scenarios
Specify Exposure Scenarios
Application Type
Selected Scenarios
lMQOL.crack&.crevice.[a.erosoO..
Pet treatment (liquid)
Lawn (liquid - handwand)
Vegetable garden (dust, powder)
Indoor carpet and room (aerosol)
Lawn (granular - push spreader)
Indoor fogger (broadcast)
Indoor crack & crevice (liquid)
Indoor flying insect killer (aerosol)
Pet treatment (spot-on)
Pet treatment (liquid)
Lawn (liquid - handwand)
Vegetable garden (dust, powder)
Delete Selection
Alter Selection Priority _.
Use arrows to reorder
selections and change
co-occurrence priority.
Cancel
This list contains the applications that
will be simulated. The top scenario has
the highest priority.
Choose and prioritize all exposure
scenarios before continuing.
Continue
Figure 5-8. Specify Exposure Scenarios Screen.
32
-------
5.5.2 Specify Exposure Scenario Details
The list box on the left of Figure 5-9, "Scenarios", displays the scenarios chosen by the user in
the previous screen. Clicking on one of these entries selects that scenario as the current one (and
highlights it). Once selected, the user needs to do the following for each scenario:
1. Specify or edit the application probability.
2. Specify whether co-occurrence will influence this scenario's dates.
3. Click the Specify Inputs button to enter application dates, times, and co-occurrence.
Only when these have been completed for each scenario will the user be permitted to continue. If
the existing values for probability and co-occurrence are acceptable, the user is not required to
edit these values.
When highlighted, the application's probability and co-occurrence status are displayed. The
Application Probability is the probability that this application will be used by an individual that is
in the use group (e.g., has a garden for a garden application).
Note: Co-occurrence can only be
selected in SHEDS if the chemical
application dates are supplied by the
model. If user-supplied chemical
application dates are used in the
Simulation Information Screen, then the
co-occurrence radio button selection
will be grayed out.
If Specify Parameters is selected in the "Scenario
Co-Occurrence" radio box, then the application
dates of other scenarios will be allowed to
influence the application dates of this scenario.
Selecting this here will cause the "Specify Co-
Occurrence" screen to be displayed at the
appropriate time. The "Specify Co-Occurrence"
radio box will be grayed out for the application
with the highest priority since no other scenarios
are allowed to influence it.
For each selected exposure scenario, the user should enter the Application Probability and
specify whether Co-Occurrence should be used. Then clicking on the "Specify Inputs" button
will take the user to the Specify Application Dates and Specify Co-Occurrence screen for the
selected scenario. After completing the information on these screens the user will be returned to
this screen to work on the remaining scenarios.
Scenarios that have been completed are shown in the list box on the right. The user will not be
allowed to continue (click on the Continue button) until the application date and co-occurrence
information has been filled out for all selected scenarios. Clicking on the Continue button will
take the user to the "Specify Re-entry Times" and possibly the "Specify Application Times"
screens.
33
-------
Specify Application Scenarios
Specify Exposure Scenario Details
Scenarios
Pet treatment (liquid)
Lawn (liquid - handwand)
Vegetable garden (dust, powder)
..Aoolication Probability.
0.9000
r
r
Specify Inputs
Select a scenario on the left, enter the base
application probability (if using modeled dates),
indicate whether you want co-occurrence, then click
on Specify Inputs to enter scenario details.
Scenarios Completed
Indoor crack & crevice (aerosol)
Pet treatment (liquid)
Lawn (liquid - handwand)
Vegetable garden (dust, powder)
Cancel
Continue
Figure 5-9. Specify Exposure Scenario Details Screen.
5.5.3 Specify Application Dates: Model Determined
Model determined dates are determined stochastically during the simulation. The dates will
typically vary for each individual simulated. The user enters the blackout period, the probability
vectors for month, day of week, and number of applications. These are used together to
determine the application date or dates. During a simulation the actual dates chosen may be
influenced by other scenarios if co-occurrence was specified.
Typically, the weekday probabilities will be used to specify Mon-Fri for professional applications
or another configuration, possibly Sat-Sun, for home owner applications. The Monthly
probabilities can be used to ensure that lawn or garden applications occur during the appropriate
season. The number of applications can be fixed for each individual by specifying a 1 for the
probability in the appropriate box.
Taken together the probabilities define the likelihood of an application occurring on each day of
the year. For multiple applications the blackout period comes into play. The blackout days
indicate how many days must pass before an additional similar application is allowed.
The probabilities for each vector (weekday, monthly, and number of applications) must total 1.0.
The user can enter probabilities in the first boxes of each probability vector. The last box of the
vector, with a darker border, will always contain the remainder and cannot be edited by the user.
Additional information about entering probability vectors is given in the section "Entering
Probability Vectors" on page 8.
34
-------
Specify Application Scenarios
Specify Application Dates: Model Determined
Indoor crack .crevice (aerosol)
Minimum Days Between Consecutive Applications.
Blackout Davs.
A value of 1 permits applications on consecutive days, a
value of 7 allows an application on the same day of the
next week, etc.
Application Weekday Probabilities
Sat Mon Tue Wed Thur Fri
Sat
.143 .143 .143 .143 .143 .143
Monthly Application Probabilities
Jan Feb Mar
1 .0157
I .0231
1 .0257
Jul Aug Sep
I .1613
I .1401
1 .1007
Apr May Jun
I .0581
I .1937
Oct Nov
I .0631
I .0398
I .1534
OK I
Dec
[ 0.0253J
Probabilities for Number of Applications
1
I .32
7
I .02
2
I .20
8
I .05
3
I -13
9
I -01
4
1 .06
10
1 m
5
1 .05
11
I -01
6
1 D7
12
I 0.07
OK |
Cancel
Continue
Figure 5-10. Specify Application Dates: Model Determined Screen.
35
-------
5.5.4 Specify Application Dates: User Specified
This screen (Figure 5-11) is used to enter user specified application dates. These dates are
specified by day number (with 1 being the first day of the simulation period). There is no
probability associated with these application dates as an application will be made on each of
these dates for all users in the appropriate use groups. That is, all lawn applications will be made
if the individual being simulated has a lawn, and so on for pets and gardens. In addition to the
days of application, the user should select the time. All applications in a scenario will be made at
the same time of day.
Specify Application Scenarios
Specify Application Dates: User Specified
Indoor crack _crevice (aerosol)
Application Days
"
[2pm
Cancel
Next
Figure 5-11. Specify Application Dates: User Specifed Screen.
5.5.5 Specify Co-Occurrence
This screen (Figure 5-12) is used to specify how the application dates of other scenarios affect
the application dates of this scenario. Scenarios must have been specified as being of a higher
priority in the Specify Exposure Scenarios screen (page 32) to affect a scenario. Only those
scenarios of a higher priority are listed in "Application Types of Influence". For purposes of
discussion let us call the scenario we are working with the "current" scenario, and let us call
applications from scenarios with higher priority "previous" applications. In the model code the
dates of the previous applications will be determined first. Note that 'previous' in this context
refers to the order in which the application dates are determined; it does not mean that the
'previous' dates necessarily occur earlier in the simulation.
36
-------
The probability of an application from the current scenario occurring on any day of the year is
determined based on the probability vectors entered earlier for month and day of week. When
working on the current scenario, the application dates for the higher priority scenarios will
already have been determined. The probabilities for the current scenario are altered (multiplied)
by the influence factor for all days within the influence width of previous applications. This
obviously has no effect for those days whose probability is already 0. The influence factor, also
called the co-occurrence probability is discussed and illustrated in the "Application Dates and
Co-occurrence" section of the Technical Manual.
If the influence factor exceeds one, then the likelihood of the current scenario happening near in
time to a previous scenario is increased. If the influence factor is less than one, the likelihood is
decreased. An influence factor of zero means that a 'blackout window' is created around each of
the dates for the previous scenarios, so that the current scenario cannot occur on those dates. The
two scenarios become mutually exclusive for that period of time around an application.
The user must choose an effective combination to continue. That is, at least one application type
must be highlighted and the influence factor must be different from 1. If the user wants to turn
off co-occurrence for this scenario, then enter an effective combination and return to the Specify
Scenario Details screen. On that screen click the No Co-Occurrence choice for this scenario. This
will turn off co-occurrence.
Specify Application Scenarios
Specify Co-Occurrence
Pet treatment (liquid)
Select Other Applications and Their Effect on This Application's Dates.
Application Types Of Influence
( Influence Factor
Influence Width (days)
3
14
The influence factor increases
or decreases the likelihood that
this application will occur on a
date near a previous
application. Only the
applications selected in the list
box are considered when
altering the likelihoods. Dates
must fall within the influence
width for their likelihood to be
altered. Click on help for more
information.
Cancel
Continue
Figure 5-12. Specify Co-Occurrence Screen.
37
-------
Specify Application Times
When model-determined dates are specified, the user will also need to specify probability vectors
for the application times. This screen (Figure 5-13) allows the user to enter one probability vector
for all applications occurring inside and one probability vector for all applications occurring
outdoors. For a fixed application time for each simulated individual simply enter a probability of
1 in the appropriate box. For general information on entering data in probability vectors see the
section "Entering Probability Vectors" on page 8.
Specify Application Scenarios
Specify Application Times
Probabilities for Time of Indoor Application
7am 8am Sam 10am 11am 12pm
~' ~
0
"Iprn 2pm 3pm 4pm 5pm 6pm
if i[ i[ i[ if
Probabilities for Time of Outdoor Application
7am Gam 9am 10am 11am 12pm
_3 LJ LJ LJ LJ LJ
1pm 2pm 3pm 4pm 5pm 6pm
If 1
Cancel
Next
Figure 5-13. Specify Application Times Screen.
38
-------
5.5.6 Specify Re-entry Times
Simulated individuals may be restricted from entering treated areas for a specified number of
hours. This restriction is split into indoor areas and outdoor areas. Within each area the
probability of re-entry is the same for all scenarios affecting that area. The probabilities are
entered on the Specify Re-entry Times screen (Figure 5-14). For general information on entering
data in probability vectors see the section "Entering Probability Vectors" on page 8.
Specify Application Scenarios
Specify Reentry Times
Reentry Time to Treated Indoor Area (probabilities for hours]
0123-4 5-6 7-12
13-24 25-48
1
1
0
0
0
0
0
0
43-96
LJ
Reentry Time to Treated Outdoor Area (probabilities for hours)
0123-4 5-6 7-12
13-24
25-48
1
0
0
0
0
0
0
0
49-96
LJ
OK
Cancel
Return
Figure 5-14. Specify Re-entry Times Screen.
5.6 Specify Concentration-Related Inputs
Clicking on "Specify Concentration-Related Inputs" from the main screen begins a series of
screens through which the user must navigate in their entirety. The exact screens vary depending
on the source-to-concentration options selected earlier. Since these screens may have to be
revisited for each scenario, the scenario being edited is always displayed in the third (blue) title at
the top of the screen. The following screens may be visited:
39
-------
WARNING: Different surface residue collection devices are available (e.g., aluminum plates,
rollers, sleds, hand presses, hand wipes); some collect dislodgeable residues and some collect
total residues. These methods, as well as current methods for obtaining dermal transfer
efficiencies and dermal transfer coefficients, have inherent uncertainties. Thus, it is important
to consider matching the correct type of surface loading with the corresponding transfer factor
(i.e., transfer efficiency or transfer coefficient) when developing inputs for modeling dermal
exposure.
Please see the technical manual (section on Dermal Exposure to Surface Residues) for a
discussion of entering compatible concentrations and transfer factors (e.g., to prevent double
counting of residue transfer from surfaces to skin).
The following screens will be visited once for each scenario selected, depending on the overall
method used. The exception is the Background screen which is only visited once and only for the
Decay and Dispersion and Post-application methods.
Decay and Dispersion: Specify Decay and Dispersion Distributions (Figure 5-15)
Specify Decay and Dispersion Distributions: Background (Figure
5-16)
Post-application: Specify Post-Application Distributions (Figure 5-17)
Specify Post-Application Distributions: Background (Figure 5-16)
Time series: Specify Time Series Input (Figure 5-16)
5.6.1 Specify Decay and Dispersion Distributions
Use of the decay and dispersion method requires that at least two distributions be defined for
each medium affected by a scenario's applications. These distributions define the initial residue
or concentration and the decay rate of the chemical for the medium. These are the first two
distributions given on the screen (Figure 5-15). Additionally, for indoor media the rate of
dispersion to the untreated part of the house must be calculated. To do this, SHEDS requires as
input the distribution for the untreated to treated concentration ratio at the time of maximum
concentration in the untreated area. (See the Technical Manual for more details.) If the user
chooses to ignore dispersion into the untreated area, then this ratio should be set to a point value
of zero.
These distributions need to be specified for each of the media in the simulation. The list of media
will vary depending on the scenario being worked on. Clicking on a specific medium in the
"Media Affected" list box will display the distributions associated with that medium. The two or
three distributions can be edited while their associated medium is highlighted. The third
40
-------
distribution will only be displayed for indoor media. A complete list of media is shown in the list
box on Figure 5-16 and in Table B-2.
Specify Concentration-Related Inputs
Specify Decay and Dispersion Distributions
Indoor crack _crevice (aerosol)
Media Affected
Textured Surfaces
Smooth Surf aces
Dust (indoors]
Air (indoors]
Initial Residue/Concentration of Textured Surfaces (ug/cm2)_
Distribution
UNIFORM
Minimum Maximum
[ 0.059] [ 0.141
Media Specific Decay Rate Distribution (1 /day) _
D istribution Point
POINT
0.1
Ratio of Untreated to Treated At Untreated Maximum (Indoors Only, Unitless]_
Distribution Point
Cancel
Continue
Figure 5-15. Specify Decay and Dispersion Distributions.
5.6.2 Specify Decay and Dispersion Distributions: Background
The background values define distributions of the chemical added to each medium over the
course of the entire run. The background concentrations are added to any chemical on the media
resulting from a chemical application. The chemical concentrations from applications persist
from one time period to the next (altered by decay and dispersion). In contrast, the background
values are estimated daily and used for only that day.
The background concentration for each medium is defined by a single distribution. This same
distribution is used throughout the simulation. Background values can be added to all media that
the model understands, not just those involved in a scenario. The user may disable the addition of
background concentrations by setting each distribution to a point value of 0 (their default value).
The same screen is used for both Decay and Dispersion and Post-application backgrounds.
41
-------
Specify Concentration-Related Inputs
Specify Decay and Dispersion Distributions
Background
Media Affected
Surface residues in vegetable
Residue on pet fur
Suiface residues on lawn
Textured surfaces in untreatec
Smooth surfaces in untieated i
Textured surfaces in treated re
Smooth surfaces in treated roc
Soil in treated outdoor location
Air outside residence
Dust in untreated room
Air in untreated room
Dust in treated room '
Background Concentration for Surface residues in vegetable garden (ug/cm2)
Distribution Point
L
I POINT
0
Cancel
Continue
Figure 5-16. Specify Decay and Dispersion Distributions: Background Screen.
5.6.3 Specify Post-Application Distributions
Post-application distributions might be used when measurement studies have determined the
actual concentrations over time after applications. SHEDS uses four time periods:
1. The day of application (< 1 day).
2. The day after application to 7 days after application (1-7 days).
3. The 8th day after application to the 30th day (8-30 days).
4. The 31st day after application and the remainder of the simulation period.
The user must define these four distributions for each medium affected by each scenario. Once an
application is made for a scenario, these distributions will determine the concentrations on each
medium. When another application is made in the same scenario, the first distribution is used
once again; there is no persistence of chemical from the previous application. The concentrations
on a medium from different scenarios are added when determining exposure of an individual. It
is assumed that background values are included in the distributions; the user cannot define
background concentrations on media not included in the scenarios.
As on the previous screens, the third title, in blue, indicates the current scenario. Clicking on a
medium in the Media Affected list box selects and highlights that medium. The distributions
displayed are related to that medium for the current scenario. The user is not forced to enter data
for each media. Any distributions not entered by the user will retain their previous values
whether desirable or not. Once the distributions have been edited as desired, the user should click
on Continue to go on to the next scenario. If the user is working on the last or the only scenario,
then Continue will take the user back to the main screen.
42
-------
The user is not forced to edit the distributions for all media. Those not edited will maintain the
previously defined values. The Continue button will be active unless one of the distributions
contains an error. Click on Continue to go to the screen to enter data for the next scenario. If this
is the only or last scenario, then the next screen will be used to define background values for all
media.
Specify Concentration-Related Inputs
Specify Post-Application Distributions
Media Affected
ioil in treated outdoor lo
Residue/Concentration < 1 Day Post Application (ug/cm2)_
Distribution Point
\_
POINT
10
Residue/Concentration 1 -7 Days Post Application (ug/cm2) _
Distribution Point
IPOINT
Residue/Concentration 8-30 Days Post Application (ug/cm2)
Distribution Point
~
POINT
Residue/Concentration 31 -365 Days Post Application (ug/cm2)
Distribution Point
POINT
0
Cancel
Continue
Figure 5-17. Specify Post-Application Distributions Screen.
General information on distributions are provided in section "Distributions Supported" on page
11 and information on entering distributions on this screen are discussed in "Individual
Distribution Widgets" on page 11.
5.6.4 Specify Post-Application Distributions: Background
Entering distributions for background concentrations works the same as entering the background
concentrations for decay and dispersion (page 41). In the case of post-application concentrations,
the assumption is that background values for any media with a defined concentration are built
into the concentration distributions. So one is only allowed to enter background concentration
distributions for those media that do not have application related concentrations defined. Media
43
-------
that have post-application concentrations defined will not be listed in the Media Affected list
box.
5.6.5 Specify Time Series Inputs
Time series inputs allow the user to specify the exact concentrations on each medium for specific
dates. This is a convenient method of using data derived from studies of multiple households.
Details on the file format required are discussed in the next section (Time Series File Format).
The burden is on the user to create the file containing time series for all affected media. The
interface expects to be given a single SAS dataset which is copied into the input directory for the
run. The dataset is specified by clicking on Copy Time Series Dataset (Figure 5-18) and then
using the explorer to identify the dataset.
When using time series inputs the Specify Application Scenarios button on the main screen is
grayed out. The model does not simulate specific applications in this case.
5.6.5.1 Time Series File Format
The SAS dataset must contain one variable named Date containing the date in SAS date format.
The rows must contain sequential values for the date from the first specified to the last. The
remaining columns define a time series of concentrations for each of the media. Multiple time
series may be supplied for each medium. If multiple time series are supplied, the model will
choose them at random for each individual. Corresponding time series are always used; if the
first time series is chosen, then the first time series for each medium will be used.
A SAS variable name designates each individual time series. The variable name indicates the
medium by text and an appended number indicates the position in the set for the medium. For
example, ATair_3 is the SAS variable name for the time series for the air in the treated portion of
the third house. A complete list of text names for the media is given in Appendix B, Table B-2.
These time series cannot contain internal gaps, but they may begin after the start of the
simulation period or stop prior to the end of the simulation period. Alternatively, the time series
may begin before the start of the simulation period or extend beyond it. Time series with
different numerical indices may have different start and stop dates, but time series with the same
numerical index must all have the same start date (and the same stop date).
44
-------
Specify Concentration-Related Inputs
Specify Time Series Inputs
Dataset for All Scenarios and All Media
T ime S eries D ataset fwill be cooled to I n. T imeseriesl.
Ti iiiii e S e iri eis D itiaset I I n .Ti m e S e i i es
j^j^jgjj^^j-^jf^jfgjj^-^^j^j^-j^jf^^^f^^ I
Cancel
Continue
Figure 5-18. Specify Time Series Inputs Screen.
5.7 Specify General Exposure and Dose Factors
Clicking this button on the main screen begins a sequence of screens allowing the user to enter or
edit the remaining inputs. These inputs fall into three broad categories:
1. Individual exposure and dose factor variables, typically defined by a distribution.
2. Contact probabilities, defining how frequently an individual comes into contact with
specific media.
3. Correlation between input variables (optional).
Each of these categories is discussed in the sections below.
Initially the user is taken to the Variability Distributions screen (Figure 5-19). From this screen
the user repeatedly visits the Edit Variability Distributions screen (Figure 5-20).
If the user desires to define correlated inputs, clicking on Define Correlated Input Variables will
lead to the following two screens: Select Correlated Variables (Figure 5-22), and Specify
Variable Correlations (Figure 5-25). The first "Select Correlated Variables" screen is used to
select which variables are to be correlated; the "Specify" screen is used to specify the correlation
values defined.
When the previous screens have been visited, the user clicks on the Continue button on the
Variability Distributions screen and is taken to the Contact Probabilities screen (Figure 5-26).
45
-------
This screen is used to repeatedly visit the Edit Contact Probabilities screen (Figure 5-26.
Contact Probabilities Screen.) where the user can specify contact probabilities for the various
media.
5.7.1 Variability Distributions
The variables set using the first two screens in this series are grouped as follows:
Activity-Related (Probabilities related to lawns, gardens, and pets)
Dose-Related (GI and dermal absorption rate, elimination rate, etc.)
Transfer-Related (skin adherence, non-dietary ingestion, etc)
Removal-Related (hand washing frequency, removal efficiencies, etc.)
Baths (time between baths)
Each variable is defined by a distribution, except for the time between baths which is a
probability vector.
The initial Variability Distributions screen (Figure 5-19) is used as a jump point. The user selects
a variable group by clicking on the group name in the Variable Groups list box. The group is
highlighted and the descriptions of each variable in the group are displayed in the center list box.
Clicking on the Specify Inputs button will take the user to the next screen where individual
distributions can be edited. After clicking on Save, the user will be returned to this screen. The
Specify General Exposure and Dote Factors
Variability Distributions
Variable Groups . Groups Completed
6Iiiiftiiili:"r""'~~ """"
Transfer-related
Removal-related
Dose-related
Baths
1 >
Specify Inputs
Activity-related
Transfer-related
Removal-related
Dose-related
Baths
Correlate Input Variables .
Correlate Input Variables Define Correlated Input Parameters
Cancel
Continue
Figure 5-19. Variability Distributions Screen.
46
-------
variable group just edited will be added to the Groups Completed list box if it was not already
there. While one or more variable groups remain to be edited, the background of the Groups
Completed list box will be light blue (Figure 2-3 on page 8) and both the Define Correlated Input
Variables and the Continue button will be disabled. When all groups have been completed the
user may continue contact distribution screens.
If the user returns to the Simulation Information screen and changes options, she will be forced to
return here and enter values for any variables not previously defined. For example if the
simulation options are switched to Transfer Coefficients, the user will need to return here to edit
the Transfer-Related group. Only then will the user be able to continue and run the simulation.
Once all groups are completed, the user will be able to define correlated inputs. By default
correlated inputs are turned off; no input variables are correlated. To turn correlation on click on
the check box labeled Correlate Input Variables. This will enable the Define Correlated input
Variables button. Click on the button to proceed to the two screens used to select and define
correlated inputs. After defining correlated inputs, the user will be returned to this screen. If the
definition of correlated inputs is not completed, input correlation will be turned off when the user
returns to this screen.
Only when all variable groups have been completed will the user be allowed to continue. If
correlation was turned on, the correlated pairs must be defined or correlation must be turned off
before the user may continue. Once enabled, clicking on the Continue button takes the user to the
screens used to define contact probabilities.
5.7.2 Edit Variability Distributions
The user comes to this screen (Figure 5-20) when the Specify Inputs button is clicked from the
Variability Distributions Screen. The screen will always have the same basic layout, but the
specific variables that can be edited will change based on the variable group highlighted when
the Specify Inputs button was clicked. The second title will always indicate the variable group
being edited.
The variable to be edited is chosen in the left hand list box. If the variable has several
distributions associated with it, for different ages perhaps, then a list of conditions will be listed
on the right (see Figure 5-21). A variable is selected by Clicking on the variable description, or
the condition if present. When selected, the current distribution parameters for that variable will
be loaded into the widget. The title of the box surrounding the distribution widget will show the
actual variable name, a vertical bar, and then the description for that variable. Use of the
distribution widget is covered on page 11.
Whenever a valid distribution is loaded or redefined a frequency histogram will be drawn
showing a representative sample for the distribution as defined. To turn this off uncheck the
histogram check box on the right. The mean and standard deviation of the sample will also be
47
-------
shown in the box on the right. It should be emphasized that the sample shown is not used in the
model. The distribution will be sampled during the model run to generate values used in the run.
Some features, particularly due to truncation, are easier to see if the number of bins are increased
using the spin box on the right.
The Print Plot button will route a copy of the plot to the default printer.
To accept the current values click the Continue button. If any of the parameter values are in error,
they will be highlighted in red or yellow and the Continue button will be disabled. Clicking on
the Cancel button returns the user to the Variability Distributions screen without saving any
changes.
More detailed information on editing distributions is given in "Individual Distribution Widgets"
in Section 2.3.5.
48
-------
Edit Variable Distributions
Editing Dose-related
Dataset Beina Edited.
in. Distributions
Variable Descriptions
mass ratio (metabolite/pollutant)
elimination rate from the blood
bioavailabilit1,' traction for surface residues
bioavailability fraction for dust/soil
Gl tract absorption rate per day for surface residues
Gl tract absorption rate per day (or dust or soil
dermal absorption rate per day for surface residues
dermal absorption rate per day for dust or soil
absorption fraction for lungs
Conditions
bioavr I bioavailability fraction for surface residues ([-])
Distribution
ILOGNORMAL j*j
Geo. Mean Geo. Std. D Maximum
[ -3 [ 1-1 f ^ OK |
BIOAVAILABILITY FRACTION FOR SURFACE RESIDUES
bioavailability fraction for surface residues([-] )
Percentage
8
After correcting
errors click on OK or
hit the return key to
force validation
(clear errors).
W Histogram
Distribution Statistics
Stat
N
Mean
StdDev
Value
50000.0
0.8029
0.0755
Display Bins
>.5271746928 0.704484183 0.8522420915 1
0.6306052288 0.7783631373 0.9261210458
Samp1ed Ma 1ues
Cancel
Print Plot
Continue
Figure 5-20. An example of the Edit Variability Distributions Screen.
49
-------
Variable Descriptions
residue-skin transfer efficiency
(transfer coefficient for object mouthinc
surface-skin transfer coefficient for body (unclothed)
object-mouth transfer efficiency
object-mouth contact rate
object-mouth contact area
object-surface concentration ratio
soil ingestion rate (outdoor, direct only]
Conditions
AGE>19
Figure 5-21. Variable with Two Conditions shown.
5.7.3 Correlating Input Variables
All of the variables in this section (General Exposure and Dose Factors) are randomly sampled
from the specified input distributions. In certain cases, the user might want some of these
variables to be correlated with each other. For example, perhaps the hand-to-mouth transfer
efficiency and the object-to-mouth transfer efficiency should have a tendency to track each other,
since the physical/chemical properties should be similar in both cases. This can be achieved in
SHEDS by requesting that these inputs be correlated.
Any of the variables in the General Exposure and Dose Factors section (with the exception of
probability vectors) may be correlated with others. The user selects the subset of input variables
that will be subject to correlation, and then specifies the pair-wise Spearman correlations. If 'N'
variables are selected for correlation, there are N(N-l)/2 distinct pairs. The user does not have to
specify all pairs; any that are not given a definite target correlation are assumed to have a target
correlation of zero.
SHEDS uses a modified NORTA4 method to generate the correlations. The random values for
the selected set of variables are first generated from a multivariate normal distribution with the
correct Spearman correlations. Each normal variate is then transformed to the specified
distribution using a rank-preserving transformation function. Since Spearman correlations
depend only on rank, the Spearman correlations are preserved. The result is that the inputs will
pair-wise exhibit the desired correlations, but will also have the marginal distributions requested
by the user.
It is possible to specify correlations that cannot be achieved. As an example, if two variables are
to have a perfect correlation of 1.0, then each must have the same correlation as the other with
any third variable. Mathematically, a correlation matrix is valid if and only if all its eigenvalues
are non-negative; however, SHEDS-Multimedia requires strictly positive eigenvalues. If the set
of requested correlations is not allowed, a message is printed on the SAS log when the model is
run.
4 The NORTA method is described in the SHEDS technical manual.
50
-------
To be correlated on input, a variable must:
be defined in the distributions file (these are the variables edited through the Variability
Distributions screen and given in Appendix B, Table B-l), and
it cannot be a point or probability vector.
Two screens are used to select and define the correlations. The first screen Select Correlated
Variables (Figure 5-22) is used to choose the variables of interest. After choosing the variables,
clicking on the Continue screen takes the user to Specify Variable Correlations (Figure 5-25)
where specific pairs are assigned target correlation values.
5.7.3.1 Select Correlated Variables
The Select Correlated Variables screen (Figure 5-22) is used to choose which variables will be
used to define correlated pairs in the next screen. It does not matter if variables are selected here
which are not used later. Only valid variables will be displayed in the Input Variables list box.
Move variables to the Correlate These Variables list box to include them in correlated pairs on
the next screen. The single arrow moves the selected variables, the double arrows move all
variables. If the user has previously defined correlated pairs, the right hand list box will initially
be populated only with variables in those correlated pairs.
Clicking on the Continue button will normally take the user to the Specify Variable Correlations
screen. If the user selects less than two variables and clicks on Continue, an error will be
displayed (Figure 5-23), correlation will be turned off, and the user will be returned to the
Variability Distributions screen.
51
-------
Specify General Exposure and Dose Factors
Select Correlated Variables
Input Variables
Correlate These Variables
Bpd^-sufface.fraciignaj. contact .rate
Gl tract absorption rate per day for dust or soil
Gl tract absorption rate per day for surface residues
Hand-surface fractional contact rate
dermal absorption rate per day for dust or soil
dust ingestion rate (indoor, direct only)
elimination rate from the blood
fraction of body unclothed
fraction of surface of one hand that enters mouth
hand mouthing events per hour
maximum dermal loading for body
maximum dermal loading for hands
mean 8 hand washes/day per person
probability of having a dog or cat
soil ingestion rate (outdoor, direct only)
removal efficiency during batWshower
removal efficiency during hand washing
removal efficiency during mouthing
residue-skin transfer efficiency
soil-skin adherence factor
surface-skin transfer coefficient for body (unclothed)
surface-skin transfer coefficient for hand
Cancel
Continue
Figure 5-22. Select Correlated Variables Screen.
You have not selected enough variables
to define any correlations. As a result
no variables will be correlated, To
define variable correlations return to
this screen and select 2 or more variables.
OK
You have not selected any correlated pairs.
As a result no variables will be correlated,
To define variable correlations return to
the Select Variable Correlations screen.
Figure 5-23. Error Displayed if Zero or
One Variables are Selected for
Correlation.
Figure 5-24. Error Displayed if No
Variable Pairs are Selected for
Correlation.
5.7.3.2 Specify Variable Correlations
This screen (Figure 5-25) is used to define or edit variable pairs and assign a correlation value to
the pairs. A correlated pair is represented as two variable names and a correlation value from -1
to 1, inclusive. Note that the numbers in Figure 5-25 are hypothetical and for illustration
purposes only; they are not intended to be recommended values.
52
-------
The variables are ordered. Initially, all but the last variable are displayed in the First Variable list
box. The Second Variable list box will contain all variables below the variable highlighted in the
First Variable list box. The Second Variable list box will also always include the last variable.
The basic method of selecting a variable pair is to select the first variable in the First Variable list
box, select the next variable from the Second Variable list box, specify the correlation value in
the spin box to the right, and click on the Add Pair Button. Once added the variable pair will be
displayed in the Selected Pairs Specified box on the lower portion of the screen.
If a variable pair selected in the upper list boxes was previously specified and is already listed
below, then the Add Pair button will not be enabled. Instead the Replace Pair button is enabled
and will replace the correlation value currently specified with the one previously specified for the
pair.
The user may find it necessary to resize the columns in the Correlated Pairs Specified area. This
area is capable of displaying both variables in the pair and the correlation value. Sometimes the
columns are not initialized such that all three are visible.
To delete a pair, highlight the pair in the Correlated Pairs Specified area. This will enable the
Delete Pair button. Click on this button and the pair will be removed from the list.
Clicking on Continue saves the values defined and returns the user to the Variability
Distributions screen. Once uncertainty is implemented, the user will be returned to the
Uncertainty Distributions screen if doing an uncertainty run.
If no correlated pairs are chosen, then an error screen will be displayed (Figure 5-24), and
correlation will be turned off in the simulation. Begin from the Variability Distributions screen to
turn correlated inputs on once again.
53
-------
Specify General Exposure and Dose Factors
Specify Variable Correlations
First Variable
removal efficiency during hand washing
removal efficiency during mouthing
residue-skin transfer efficiency
soil-skin adherence factor
surface-skin transfer coefficient for body (uncloth
Second Variable
removal efficiency during mouthing
residue-skin transfer efficiency
soil-skin adherence factor
surface-skin transfer coefficient for body (uncloth
surface-skin transfer coefficient for hand
Replace Pair
Correlation
lai
P
Correlated Pairs Specified _
Variable 1 Label
Variable 2 Label
Correlation
residue-skin transfer efficiency
removal efficiency during bath/shower
removal efficiency during hand washing
removal efficiency during bath/shower
soil-skin adherence factor
residue-skin transfer efficiency
residue-skin transfer efficiency
surface-skin transfer coefficient for body (u
soil-skin adherence factor
soil-skin adherence factor
removal efficiency during mouthing
removal efficiency during mouthing
removal efficiency during hand washing
surface-skin transfer coefficient for body (u
surface-skin transfer coefficient for body (u
surface-skin transfer coefficient for hand
surface-skin transfer coefficient for hand
surface-skin transfer coefficient for hand
0.6
0.4
0.4
0.4
0.6
0.6
0.6
0.6
0.6
Cancel
Continue
Figure 5-25. Specify Variable Correlations Screen.
5.7.3.3 Contact Probabilities
During each event in the simulation the model must decide where the individual is and what that
person is touching. This is done in three stages: the first is deterministic; the second and third are
probabilistic.
First, the location, or microenvironment, is decided deterministically based on the person's diary.
For a specific diary-day and event this will always resolve to the same value: inside a residence,
54
-------
outside a residence, inside other (stores, schools, etc.), or inside a vehicle. The SHEDS-
Multimedia version 3 GUI includes only allows data entry for the first two of these
microenvironments.
Second, the events determined to be inside a residence are stochastically assigned to one of three
"sub-microenvironments": treated, untreated, or far away (i.e., neither). A "treated room" in this
context refers to a part of the house that has been or will be treated (have chemical applied)
during the simulation. An "untreated room" is one which is never directly treated, but may
receive chemical contamination from a nearby treated room. The "neither" category refers to
rooms which are sufficiently remote from a treatment location that they will always have zero
chemical concentration. One common interpretation of "neither" is time spent in houses other
than one's own. However, it could also include time in detached buildings at one's home, or
even in parts of one's house that are far enough from a treatment location to never acquire any
chemical. This will depend on many factors, and judgment may have to be used. A conservative
assumption is that all "inside residence" events are either in treated or untreated rooms, in which
case the probability of "neither" should be set to a point value of zero.
Third, the surface being contacted is determined probabilistically. If the microenvironment is
inside a residence, the possibilities are smooth (e.g., vinyl, hard furniture), textured (e.g., carpet,
cloth upholstery), pet, dust, and nothing; if outside, the possibilities are garden, lawn, pet, soil,
and nothing. Only one surface is contacted during a single event. It is assumed that one is always
in contact with the air. To increase the number of contacts, decrease the maximum event time on
the Simulation Information screen (see Figure 5-7).
Figure 5-26 shows the screen used for choosing groups of contact probabilities and editing them.
It functions the same as the Variability Distribution Screen (page 46). The user chooses a
variable group by selecting one in the Probability Variable Groups list box. This will cause the
variables in that group to be displayed in the Variable Descriptions list box. The first group is
used to determine the probabilities used in step two above. The other three determine the
probabilities for step 3 above. The variable groups are:
1. Inside Residence: probability vector assigning an inside individual to a sub-
microenvironment of treated, untreated, or far away (neither).
2. Within Treated Rooms: for an individual that has been determined to be inside and in
a treated room, this probability vector is used to assign the contact surface: smooth,
textured, pet, no surface.
3. Within Untreated Rooms: as previously for people determined to be in untreated
rooms. Possible contact surfaces include: smooth, textured, pet, no surface.
4. Outside Residence: for an individual determined to be outside in the first step, this
probability vector assigns the contact surface: lawn, garden, pet, or no surface.
55
-------
To edit the distributions for the contact probabilities, the user clicks on Specify Inputs. This will
display the Edit Contact Probabilities screen (Figure 5-26) with the proper variables available for
editing. Once those values are edited and the user returns to this screen, the variable group just
edited will be added to the Groups Completed list box. Once all groups have been completed, the
Continue button will be enabled. Clicking on the continue button will complete the entry of all of
the general exposure and dose variables. The user will be returned to the main screen and be
permitted to begin a simulation run.
Specify General Exposure and Dose Factors
Contact Probabilities
r Edit Contact Probabilities
Probability Variable Groups . Groups Completed
waiffiatfeii: ".:::';.:'",:; :.rll
Within Treated Rooms
Within Untreated Rooms
Outside Residence
< >
Specify Inputs
Inside Residence
Within Treated Rooms
Within Untreated Rooms
Outside Residence
Cancel
Continue
Figure 5-26. Contact Probabilities Screen.
5.7.3.4 Edit Contact Probabilities
This screen (almost identical to Figure 5-20) is used to edit the individual contact probability
distributions. The user comes here when the Specify Inputs button is clicked from the Contact
Probabilities Screen. It functions exactly like the Edit Variability Distributions screen (Figure
5-20).
The set of probabilities being edited at one time will form a probability vector of mutually
exclusive possibilities. For Inside Residence three variables can be assigned distributions. Each
variable is a probability, and together they form the probability vector representing the likelihood
of that an individual assigned "Inside Residence" will be assigned to one of the three "sub-
microenvironments": treated, untreated, or far away. During the run a value will be selected from
each of the probability distributions and the three values will be normalized to equal one. For the
remainder of the run this probability vector will represent likelihood of that individual being
assigned to the three "sub-microenvironments".
56
-------
The other probability vectors work the same way with one exception. An individual without a
pet, (or lawn, or garden) cannot contact that surface. In this case those probabilities are set to zero
before the vector is normalized.
The screen will always have the same basic layout, but the specific variables that can be edited
will change based on the variable group highlighted when the Specify Inputs button was clicked.
The third title, in blue, will always indicate the variable group being edited. Below the Help,
Cancel, and Save buttons are a series of container boxes for editing individual variables. The
variable description is given on the container box. The distribution for a variable is defined by
first selecting the desired distribution type and then filling in the individual parameter values.
To accept the current values click on the Save button. If any of the parameter values are in error,
they will be highlighted in yellow and the Save button will be disabled. Clicking on the Cancel
button returns the user to the Contact Probabilities screen without saving any changes.
More detailed information on editing distributions is given in "Individual Distribution Widgets"
in Section 2.3.5.
Run Simulation
Run Name
[Demonstration File (H=100|
5.8 Run Simulation
Only when all information has been specified will the Run Simulation button on the main menu
(see Figure 5-1) be enabled. Clicking on it brings the user to the Run Simulation screen (Figure
5-27). The screen displays the current run name and estimated run time. If 20 or fewer people are
being simulated, then the user will have the option of turning on the diagnostic mode.
"Write Inputs to Excel" generates
a file containing all of the user
editable model inputs. The file is
an XML file, but can be read by
MS Excel 2003 or later. The file
is placed in the install directory
and is named using the run name
and the standard Excel file
extension (.xls).
"Check Input for Errors" forces
two routines to run and check
inputs. Comments and errors will
be displayed to the log which will
be made visible if is not already.
These same checks are also
carried out when the simulation
is run.
,r ..
L*mi£.
.... ...j
FJ!J:5
ChgckJnput For Errorsj
Run Simulation |
Estimated Run Time:
0:06 (Hrs:Mins)
Cancel
Open the log window before running the simulation to
monitor its progress. Typically one can do this by
clicking on the Log Button at the bottom of the
screen. Alternatively select the View menu on the
main SAS window and select Log.
To cancel a run in progress click the Cancel Buttton
on the main SAS tool bar (a circle with an
exclamation mark). On the screen that pops up,
select 'Cancel Submitted Statements', click 'OK1. On
the next screen click 'OK' again.
Continue
Figure 5-27. Run Simulation Screen.
57
-------
Clicking on "Run Simulation" will begin the simulation. Informational messages will be
displayed in the log (Figure 5-28). Among other items, the number of people completed and the
number total are indicated as the run progresses. When the simulation is completed, the final
timing information will be displayed in the log. Clicking the Continue button after run
completion will return one to the main menu where View Results can be selected.
Typically you will want to open the log window before running a simulation. If the log window
has been minimized, clicking the button along the bottom of the main SAS window will open it.
This can be done after the run has started. If not open yet, the log window can be opened by
selecting the View menu on the main SAS menu bar and then selecting Log. This must be done
before the simulation has started.
The correlated pairs are checked for validity at the beginning of the run. If problems are found an
error message will be printed in the log.
The diagnostic mode is only available for simulations with twenty or fewer people. The option
for diagnostic mode does not show up if using the Demonstration File since that has 100
individuals. Typically, a user will not want to run in diagnostic mode. If one is very familiar with
the model it may be helpful in locating errors. Three things happen when the model is run in
diagnostic mode.
1. SAS prints detailed notes to the log detailing how macros were compiled and data steps
completed.
2. Intermediate variables are saved on the datasets. This option is also available on the
Simulation Information screen.
3. Event level datasets and variables are saved for each individual simulated.
Once started, the run can be cancelled by following the instructions on the Run Dialog.
58
-------
1 H Log - (Untitled) ' :^;|sl
Running SHEDS -Mul t imed ia 3, development revision 30 ^j
correlation = 0
job setup
loop setup
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
completed
d iary
concentr
exposure
dose
summary
total
time = 1 .071
time = 3.555
person 1 of 25 in 7.11, job= 11.746
person 2 of 25 in 7.981, job= 19.738
person 3 of 25 in 7.871, job= 27.609
person 4 of 25 in 8.092, job= 35.711
person 5 of 25 in 7.651, job= 43.362
person 6 of 25 in 7.16, job= 50.532
person 7 of 25 in 7.191, job= 57.733
person 8 of 25 in 7.821, job= 65.564
person 9 of 25 in 7.881, job= 73.445
person 10 of 25 in 8.172, job= 81.617
person 1 1 of 25 in 7.951, job= 89.568
person 12 of 25 in 7.131, job= 96.699
person 13 of 25 in 7.701, job= 104.4
person 14 of 25 in 8.332, job= 112.742
person 15 of 25 in 7.44, job= 120.182
person 16 of 25 in 8.122, job= 128.314
person 17 of 25 in 7.871, job= 136.185
person 18 of 25 in 7.712, job= 143.907
person 19 of 25 in 7.681, job= 151.598
person 20 of 25 in 8.282, job= 159.89
person 21 of 25 in 7.661, job= 167.551
person 22 of 25 in 7.47, job= 175.021
person 23 of 25 in 8.292, job= 183.313
person 24 of 25 in 7.391, job= 190.704
person 25 of 25 in 8.312, job= 199.016
time = 100.252
time = 26.197
time = 50.591
time = 22.384
time = 20.971
time = 199.186 for 1 loops of 25 persons
Figure 5-28. Log Screen after Running a Small Simulation.
5.9 View Results
Model results from previous or current SHEDS simulations can be viewed by selecting the View
Results button on the main window. This button will become active once a simulation run has
been made or after a previously run simulation is specified in the Run Name dialog. Pressing this
button opens a dialog (Figure 5-29) allowing the user to select one of the following:
View Results for the Population,
View Results for an Individual,
Each of these buttons opens a new dialog.
59
-------
1 View Results
View Results for the Population |
View Results for an Individual |
I view Diary Pool Sizes
i
Close |
Figure 5-29. View Results Screen.
5.9.1 Additional Outputs and Files
A number of outputs cannot be viewed through the interface. For information on the log files,
export files, and exporting files from SAS for analysis in other software see Appendix A for
more information.
5.9.2 View Results for the Population
The View Results for the Population window is used to generate graphical or tabular results for
output variables for the population as a whole, or for subgroups of the population. The outputs
will utilize the average daily values for the simulation just run or that specified most recently in
the Run Names dialog. When considering the population, the output variables of utilized are the
daily values (of exposure, dose, etc.) for each person, averaged over the simulation period. The
outputs generated examine the variability of these personal mean daily values over the
population.
In general the user will want to work from the top down, selecting the output units and thus the
output dataset of interest, then the sub-population of interest, output type, and specific variables
desired.
The Run Name box shows the name of the current run, while the Output Data Set supplies the
name of the corresponding SAS dataset that will be used to generate the results.
The Output Units menu is used to indicate the units for the output data, either Milligrams per
Kilogram body weight or Micrograms. The appropriate output dataset will be displayed in the
Output Data Set box.
The menus in the Select Population box (Gender, Start Age, Stop Age, Min Rank, and Max
Rank) can be used to select a subgroup of the run population for analysis. Subsetting is based on
60
-------
the fact that each simulated individual is assigned a specific age, gender and bin rank. The entire
simulated population is divided into 100 bins labeled 0 to 99; as nearly as possible, the bins
contain the same number of people. (If the simulated population is less than 100, some of the
bins will be empty and will not appear in the pull down menu.) The bins are ranked based on the
total chemical absorption over the entire simulation period. The user may subset the population
by restricting the age range, gender, bin rank, or any combination of these. For example, the user
may choose to examine only the women or only the children. The pull down menus only display
genders, ages, and ranks that exist in the output dataset. The final selections and count of
individuals will appear on any resulting tables and graphics generated. The Selected Count field
updates after each user selection to display the final number of people in the sub-group.
The Output Type list box includes the analyses that may be performed. The user selects one
option from this list at a time. The options for results analyses in SHEDS-Multimedia version 3
are the following:
Summary Table
CDF (Cumulative Distribution Function)
Box and Whiskers
Contribution by Pathway
Pressing the Display button then performs the analysis and displays the results in a new window.
That window must be closed to return to this dialog to define new outputs.
The Variable Groups and Select Variables menus are used to select the model output variables to
analyze. The Variable Groups menu is used to pick a general category of model variables (such
as dose, exposure, or loading variables). Selecting this variable group will update the Select
Variables menu to include all corresponding model variables. For example, in Figure 5-30 the
Variable Group selected is Absorption: New, so the Select Variables menu has been updated to
include all the corresponding absorption variables. The user then selects one or more variables
from this list to analyze. Clicking on display generates the output. Specific details of each of the
Output Types are covered next.
The dose/exposure terminology varies slightly between the figures generated, the dialog menu,
this manual, and the Technical Manual; the differences are summarized in Table 5-1.
61
-------
Table 5-1. Output Variable Group and Dose/Exposure Terminology Differences
Short Description
Absorption
Exposure
Dose
Elimination
Loading
Pull Down Menu Text
Absorption: New
Exposure: New
Dose (chemical mass in blood)
Eliminated Dose (urine)
Loading (chemical mass on body part)
Technical Manual Term
Absorption or absorbed dose
New exposure
Blood dose
Elimination
Running Exposure
View Results
View Results For The Population
Run Name
[HewRunName
Output Units
Select Population (or subset individuals)
Gender
llMales & Femal -rl
|i _j
Start Age (yrs)
| 6 TJ
Min Rank
| 16 jrj
Selected Count
[ 1
Stop Age (yrs)
| 1S _LJ
Max Rank
I 83 _d
|i ^..jj
Ouput Type
Summary Table
Box and Whiskers
Contribution by Pathway
Output Data Set
[p ut. D a i ly m e a »s_m g kg
Select Variable(s)
Absorption
Absorption
Absorption
Absorption
Absorption
Absorption
Absorption
from lungs (air)
in Gl tract (matter)
in Gl tract (residue)
from body (matter)
from hands (matter)
from body (residue)
from hands (residue)
Variable Groups
(Absorption: New
Display
Close
Figure 5-30. View Results for the Population Screen.
62
-------
5.9.2.1 Summary Table
Selecting this option yields a summary statistics table for the selected variable(s) and specified
model simulation, including sample size, mean, standard deviation, median, 5^,25", 75th, 95th,
and 99th percentiles. An example is shown in Figure 5-31. If multiple variables are analyzed, the
results are displayed on individual rows of the table; the variable name and description (label) are
included on each row. Again, note that the subpopulation that is being examined is printed in the
table window - in this hypothetical example the user is examining males and females between
the ages of 4 and 19.
Absorption: (mg/kg)
Demonstration File N=100 Out.Dailymeans_mgkg
Selection: Males .Females Ages 1 to 20 Ranks 1 to 99
Variable Label
Variable N
Standard
Deviation
The 5th
Percentile
The 25th
Percentile
The 50th
Percentile
The 75th
Percentile
The 95th
Percentile
The 99th
Percentile
Absorption from lungs (air) absLa
Absorption in Gl tract (matter) absGm
Absorption in Gl tract (residue) absGr
Absorption from body (matter) absBm
Absorption from hands (matter) absHm
Absorption from body (residue) absBr
Absorption from hands (residue) absHr 100
0
5.664E-10
0.0004441
2.7116E-9
2.955E-1Q
0.0002507 8.3318E-9
0.0000568 6.9892E-9
00000
0 0 2.103E-32 5.205E-10 3.5606E-9
0 0 0.0000924 0.0007920 0.002725
0 0 1.481E-31 2.8553E-9 1.6948E-8
0 0 5.198E-32 2.512E-10 1.9319E-9
0.0000127 0.0000742 0.0001895 0.0006993 0.001288
9.9019E-6 0.0000240 0.0000670 0.0001604 0.0002930
_J
Close
To print, right click on the
table and select print.
Figure 5-31. Example Summary Table for a Population.
5.9.2.2 CDF
This option yields a cumulative distribution function (CDF) plot for the selected variable(s) and
specified model simulation. The plot illustrates how the percentiles of the selected variable(s)
behave as a function of the variable value. Each selected variable will appear as a separate line
on the plot; a legend appears at the bottom of the plot to identify each variable. An example is
shown in Figure 5-32. In the example, the CDFs for seven absorption variables are plotted (only
three can be seen in this particular case, since all the percentiles for the others were zero in this
simulation.) The plot can be printed by clicking on the Print button. The default printer will be
used and the plot will be formatted for default SAS printer.
63
-------
fi?r Graph Output
View Results
Absorption: (mg/kg)
Demonstration File N= 100 OutDailyneans_mgkg
Selection: Males & Females Ages 1 to 20 Ranks 1 to 99
100
90
80
70
60
50 ^
40
30
20
10
0
0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030
Variable Value
Variable Label
Close
Print
Absorption from body (matter)
Absorption from body (residue)
Absorption from hands (matter)
Absorption from hands (residue)
Absorption from lungs (air)
Absorption in Gl tract (matter)
Absorption in Gl tract Residue)
Use the print button before opening or closing any other graph windows. Activating other
windows deletes the stored graphic from the disk.
Figure 5-32. Example CDF for a Population.
64
-------
5.9.2.3 Box and Whiskers
Selecting the Box and Whiskers option yields a box and whiskers plot for the selected variable(s)
for the current simulation for the defined population subgroup. Multiple variables are shown on
a single plot. An example for seven absorption variables is given in Figure 5-33. Unlike other
plots, the Box and Whiskers plots use the variable names rather than the variable labels. The
longer labels will overwrite one another or be suppressed by SAS. The boxes, whiskers, and
other symbols on the plot are interpreted as follows:
The midlines of the boxes are equal to the median
The plus (+) symbol inside the boxes is equal to the mean
The upper edge of the box is the 75th percentile of the population, while the lower edge is the
25th percentile. Thus, these edges define the width of the interquartile range (IQR).
The whiskers define the maximum or minimum observations that fall within 1.5 times the
IQR, measured from the quartile value (i.e., the 75th or 25th percentile).
The square symbols indicate data points outside of the range defined by 1.5 times the IQR.
65
-------
E*r Graph Output
View Results
Comparison Of Values Across Pathways
Demonstration File N= DO OutDa'lymeans mgkg
Selection: Males ft Females Ages 1 to 20 Ranks 1 to 99
0.0030
0.0025 -
"5J0.0020
^j
£ 0.0015 -
o
M
.O
< 0.0010
0.0005 -
0 -I
1
absBm absBr absGm absGr absHm absHr absLa
Output Variable
Close
Print
Use the print button before opening or closing any other graph windows. Activating
other windows deletes the stored graphic from the disk.
Figure 5-33. Example Population Box and Whiskers Plot.
5.9.2.4 Contribution by Pathway
The Contribution by Pathway option is different from the other options in the Output Type menu.
It does not require (or allow) selection of individual model variables. Only a selection from the
Variable Group menu is required: Dose, Exposure, Absorption, or Loading. Clicking on Display
yields a pie chart showing percent contribution to one of these summary variables (based on
population or population subgroup means) for the following pathways:
66
-------
Body Residue: The chemical mass contained in residues transferred from surfaces to the
skin (excluding the hands). Possible surfaces include smooth surfaces, textured surfaces, the
lawn, the vegetable garden, and pets.
Body Soil/Dust: The chemical present in soil (outdoors) or dust (indoors) that is transferred
onto the skin (excluding the hands).
GI Tract Residue: The chemical mass contained in residues transferred from surfaces to the
gastrointestinal (GI) tract. Some of the GI tract residues may have first transferred to the
hands before being ingested.
GI Tract Soil/Dust: The chemical present in soil (always found outdoors) or dust (always
found indoors) that is transferred into the GI tract.
Hands Residue: The chemical mass contained in residues transferred from surfaces to the
skin on the hands. This is tabulated separately from the body since the model allows
chemical on the hands to subsequently be transferred to the GI tract by hand mouthing, while
chemical on the rest of the body is not transferred in this manner.
Hands Soil/Dust: The chemical present in soil or dust that is transferred to the skin on the
hands. If indirect soil/dust ingestion is chosen, then this can subsequently be transferred to
the GI tract. If direct ingestion were chosen instead, then the amount of soil and dust entering
the GI tract is determined without reference to the amount on the hands.
Inhalation: The amount of chemical mass in air that enters the lungs. The mass is
determined by multiplying the chemical concentration in air by the volume of air inhaled.
Since the latter quantity is time dependent, the inhalation exposure becomes larger for the
longer-duration diary events.
A hypothetical example is given in Figure 5-34. In this example, the largest single contributor to
the variable under consideration (Exposure) is body residue.
67
-------
Graph Output
Percent Absorption: (mg/kg) By Pathway
Demonstration File N= DD Out.Dailymeansjngkg
n: Males & Females Ages 1 to 20 Ranks 1 to 99
12.54%
0.00%
45.08%
42.38%
Close
Pathway
Print
Body Residue
Hands Residue
Gl Tract Residue
OTHER
Use the print button before opening or closing any other graph windows. Activating
other windows deletes the stored graphic from the disk.
Figure 5-34. Example Pie Chart Showing the Contribution by Pathway
for a Population.
68
-------
5.9.3 View Results for an Individual
The second results window accessible from View Results is the View Results for Individual.
Selecting this button brings up the window for performing analysis of the results for a single
individual within a simulation run. Many of these options are similar to those for View Results
for Population. When considering an individual, the output variables of interest are the daily
values for the selected individual. Obviously, a specific simulated individual must be selected.
Additionally, the analysis may be limited to a specific time period. As on the population results
screen, it is best to work from the top down on the left side of the screen, and then choose the
Variable Group and specific variables of interest. All resulting outputs will contain a second title
giving the basic individual statistics and dates considered in the analysis.
The View Results for an Individual dialog is shown in Figure 5-35. The Run Name box shows
the name of the current run, while the Output Data Set supplies the name of the corresponding
SAS dataset that will be used as the basis of the results. The output dataset is dependent on the
units chosen. As in the Results for Population window, one can select genders, an age range, and
a rank range in the Select Population box. In this case, the Select Individuals list box is populated
only with individuals meeting these criteria. The individual list box provides minimal
information on each individual: the identifier used in the simulation, the gender (shown as 'M' or
'F'), age, and the individual's rank or percentile of total absorption. The user clicks on an
individual to select and then analyze data from that individual.
In addition to selecting an individual, the user may also select a particular range of dates to
analyze from the Dates of Interest menu. The Start Date and Stop Date may be selected the
corresponding menus, which are populated with all the dates covered by the current simulation.
The output options for viewing variability results for an individual are listed in the Output Type
list box. These include:
Time-Series;
CDF (Cumulative Distribution Function);
Box and Whiskers;
Contribution by Pathway;
Summary Table; and
Detailed Table.
69
-------
View Results
View Results For An Individual
_. Hun I i-_ntf_-
[NewRtinName
Output Units
illirams Per Kilo
Select Population (or subset individuals)
Gender
1 Males & Femal T
Start Age (yrs)
| 6 Z.
Min Rank
I 1G
Selected Count
f 1
Stop Age (yrs)
[ 1S jd
Man Rank
I 83 jd
Select Individual
: 2 M Age: 18 Rank: 83
ID: 3 M Age: 6 Rank: 66
ID: 4 F Age: 11 Rank: 50
lirs. trr A ___ n n _ .. i . . -ir
Dates Of Interest
Start Date Stop Date
1 01 JAN 2000
J31DEC2QOO ^
OuputType
:T1me-Sairfe«;
CDF
Box and Whiskers
Contribution by Pathway
Summary Table
Detailed Table
Output Data Set
[but.AIIPeisonDaysjngkg
S elect Variable[s]
AbsorDtion I'totall
Absorption
Absorption
Absorption
Absorption
Absorption
Absorption
Absorption
from lungs (air)
in Gl tract (matter)
in Gl tract (residue)
from body (matter)
from hands (matter)
from body (residue)
from hands (residue)
Variable Groups
Absorption: New
Close
Figure 5-35. View Results for an Individual Screen.
70
-------
As in the View Results by Population dialog, the output units may be selected as either
Microgram or Milligrams per Kilogram from the Output Units menu, and the Variable Groups
and Select Variables menus are used to select the model output variables to analyze. As before,
multiple variables may be selected for analysis. The appropriate analysis variables will
automatically be chosen for Contribution By Pathway. All variables will automatically be used
for the detailed table. Each of the output types are described next.
5.9.3.1 Time-Series
Selecting the Time-Series button produces a time-dependent plot of the selected variable(s).
Each data value on the plot is the value for a single day in the simulation. Multiple variables will
be plotted as multiple curves on the same time-series plot; the legend that appears at the bottom
of the plot identifies which variable is associated with each plotted curve. Time-Series
simulations are useful for observing the behavior of dose, exposure, loadings, or absorption in
the days following a pesticide application. An example showing seven absorption variables
(some of them essentially equal to 0) is given in Figure 5-36.
5.9.3.2 CDF
The CDF option for an individual is
similar to that for the population (See
Figure 5-32). The option yields a
cumulative distribution function plot
for the daily values of the selected
variable(s) for the specified individual.
Each selected variable will appear as
a separate line on the plot; a legend
appears at the bottom of the plot to
identify each variable.
5.9.3.3 Box and Whiskers
The Box and Whiskers option for an
individual is similar to that for the
population (see Figure 5-33 on page
66 and the previous discussion), with
the exception that the plot is generated
by examining the daily variable values
for a single person rather than the
mean daily values for the population.
Selecting Box and Whiskers and
clicking on Display will yield a plot
for the selected variable(s) for the
current simulation, selected person,
Dally Time Series for Absorption: (mg/kg)
Demonstration File OuLAIIPersonDays mgkg
Individual M Age: 17 Ranfc84 From OUAN2000 To 3VEC2000
.8
<
0.003
0X1028
0.0026
0.0024
0X1022
0.002
0.0018
0.0016
0.0014
0.0012
0.001-
0.0008
0.0006
0.000+
0.0002
0
OUAN2000 01APR2000 01JUL2000 01OCT2000 01JAN2001
Variable Label
Date
Absorption from body (milter)
Absorption from body (residue)
Absorption from hands ^n alter)
Absorption from hands (residue)
Absorption from lings (air)
Absorption in Gl tract (matter)
Absorption in Gl tract (residue)
Use the print button before opening or ciosing any other graph windows. Activating
other windows deletes the stored graphic from the disk.
Figure 5-36. Example Time-Series for an
Individual.
71
-------
and selected dates. Multiple variables are shown on a single plot.
Percent Absorption: (mg/kg) By Pathway for an Individual
Demonstration Rle OutAJIPersonDaysjngkg
Individual:?? M tgixV Hanfc84 From OUANMOO To 3DEC2000
Pattiway
I Body Residue
I .1 Hands Residue
5.9.3.4 Contribution by Pathway
The interpretation of the Contribution by
Pathway option for an individual is
identical to that for the population. As
before, this option does not require or
permit selection of individual model
variables; only the Variable Group needs
to be specified-Dose, Exposure,
Absorption, or Loading. (The Select
Variables menu will be inactive if
Contribution by Pathway is selected).
Selecting Display then yields a pie chart
showing percent contribution to one of
these summary variables by pathway. See
the earlier discussion on page 66 for
additional details.
5.9.3.5 Summary Table
If Summary Table is selected, clicking on
Display will open a new window
containing the percentiles of the selected
variable(s), with each variable appearing
on its own row in the table. Specifically,
the table contains the sample size, mean,
standard deviation, median (p50), 5th
percentile (p05), 25th percentile (p25),
75th percentile (p75), 95th percentile (p95), and 99th percentile. The age and gender of the
individual being examined appears in the descriptive heading. The resulting table looks very
similar to the summary table for the population (see Figure 5-31), with the exception that the
statistics represent the variation in the variables across days for a single person (rather than the
variation in the averaged daily values across persons).
5.9.3.6 Detailed Table
Selecting the Detailed Table option opens a new window and displays a table that contains
detailed information for all daily average variables generated by the model (Figure 5-38). It is not
required or permitted to select a variable group or specific variables. The detailed table contains
one row for each of the days being analyzed. The table contains the year, month, and day, the
number of diary events for the day, and the values for each model variable related to the Variable
Group. All variables will not fit on the screen at once. The scroll bar at the bottom of the
window can be used to view the variables that appear to the right.
Use the print button before opening or closing any other graph windows.
Activating other windows deletes the stored graphic from the disk.
Figure 5-37. Example Pie Chart Showing the
contribution by pathway for an individual.
72
-------
1 View Results 1
Absorption: (mg/kg)
Demonstration File Out.AIIPersonDays mgkg
Individual:25 M Age: 17 Rank:84 From 01JAN2000 To 31DEC2000
year mo
1 2000
2 2000
3 2000
4 2000
5 2000
6 2000
7 2000
8 2000
9 2000
10 2000
VI
New exposure
ith day numevents on hands
(residue)
1 1 45 0.0005423788
1 2 45 0.0013086779
1 3 35 0.0077826381
1 4 35 0.0090457132
1 5 35 0.0043012457
1 6 35 0.0053764162
1 7 35 0.0057079914
1 8 45 0.001948082
1 9 45 0.0005491 866
1 10 35 0.0011498622
New exposure New exposure New exposure New Gl tract New G
on hands on body on body exposure exposu
(matter) (residue) (matter) (residue) (matter
0 0.0000649554 0 0
0 0.0002119785 0 0
0 0.0021549869 0 0
0 0.0025256305 0 0
0 0.0008109217 0 0
0 0.0010057031 0 0
0 0.0018959812 0 0
0 0.0006239717 0 0
0 0.000408084 0 0
0 0.0004871039 0 0
tract New exposure Hand-to-mouth Hand-to-moutl- H
e - !,,_,. (-. transfer transfer ~J
n lungs (arr) ^^ (ma(M
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 C
0 0 0 Ci
^ |
r| J To print, right click on the
Uose I table and select print.
Figure 5-38. Example Detailed Table for an Individual.
5.9.4 View Diary Pool Sizes
Activity diaries are selected at random from pools that are characterized by different personal and
temporal characteristics. Figure 5-39 provides an example of counts of diaries available for each
diary pool as determined by the user's inputs for the simulations. The user is reminded that
larger diary pools better represent the population.
View Results
Diary Pool Sizes in Diary-Days for Current Run
Season: P-Spring, S-Summer, F-Fall, W-Winter
Weekend: 0 Weekday, 1 Weekend Day
cohort
1 3
2 3
3 3
4 3
5 3
6 3
7 3
8 3
9
10
11
12
13
4
5
8
7 5
8 5
9 5
20 5
abel | gender! subgroup sec
o<2years F 1 P
o<2years F 2P
o<2years F 3S
o<2years F 4$
Q< 2years F 5 F
o< 2 years F 6 F
o < 2 years F 7 W
o< 2years F 8 W
o< 2years M 1 P
o<2years M 2P
o< 2years M 3 S
o< 2years M 4 S
o< 2years M 5 F
o< 2years M 6 F
o< 2years M 7 W
o< 2years M 8 W
o<3years F 1 P
o< 3years F 2 P
o<3years F 3S
o< 3years F 4 S
son weekend [ diarycount |
0 96
1 70
0 52
1 36
0 38
1 30
0 16
1 13
0 82
1 66
0 34
1 34
0 50
1 41
0 22
1 11
0 84
1 75
0 50
1 33
jj
To print, fight click on the
table and select print.
Figure 5-39. Example Diary Pool Size Table.
73
-------
6 SHEDS-Multimedia: Batch Mode
The SHEDS model may also be run in batch mode. Runs in batch mode do not open any SAS
windows. They are usually submitted via the Run dialog on the Start menu in Windows. Batch
mode uses fewer computer resources, which may speed up long runs. This may be appropriate
for overnight runs where the user has no need to view the progress of the job. Multiple batch jobs
may be submitted at one time, also useful for overnight runs. Finally, the GUI does not currently
support the running of SHEDS in sensitivity analysis or uncertainty analysis mode. Such runs
may be conducted in batch mode, or by submitting the appropriate commands in a standard SAS
session.
6.1 Installing Batch Capability
Along with the SAS program "MultimediaS.sas" and the default input files used by the interface,
two additional files ("batchmultimedia.sas" and "multimedia.bat") are needed to run SHEDS-
Multimedia in batch mode.
The file "batchmultimedia.sas" does not require editing. Its contents are:
%Let comma = %index(%quote(&sysparm),%str(,));
%Let dir = %substr(%quote(&sysparm),1,%eval(&comma-l));
%Include "&dir\prg\multimedia3.sas";
%jnultimedia3(&sysparm) ;
The other file is "multimedia.bat", which must have its path names altered to match the SHEDS
installation directory on the user's machine. The default contents of "multimedia.bat" are shown
below.
"C:\Program Files\SAS\SAS 9.1\sas.exe"
"C:\Multimedia\prg\batchmultimedia.sas"
-sysparm "C:\multimedia,%1 %2 %3 %4 %5"
In the batch file, the text should be on a single line with spaces where there are currently line
breaks. The command in this file contains three pathnames which may need to be edited. The
first is the location of the sas.exe file itself. The second gives the location of the
"batchmultimedia.sas" program, which is usually in the \prg directory under the SHEDS
Multimedia installation directory. The third (the part up to the comma) gives the location of the
installation directory itself. The file can be edited in any text editor.
6.2 Defining Inputs for a Batch Run
The main difficulty in submitting a SHEDS run is ensuring that all the necessary inputs are
defined. This is a primary function of the interface itself. Hence, the easiest way to prepare a
batch run is to define and save all the inputs using the interface. Inputs for several runs could be
74
-------
defined and saved without submitting any of these jobs. Alternatively, the user could prepare the
required input files using a standard SAS editor, but this places a greater burden on the user to
ensure that all inputs have been supplied in a consistent manner.
Each set of inputs and settings is saved under a user-supplied job name. For example, suppose
the user chooses the name "jobl". Whether defined through the interface or not, this requires the
addition of a new record to the "runinfo" file found in the SHEDS installation directory, and new
directories called "runsVjobi\input" and "runsVjobi\output" need to be created. Appropriate
versions of the following SAS datasets must be placed in the \input directory:
agegroups
applications
appmedia_decay
appmedia_postapp
contactmedia
cooccurrence
correlations
diarydetails
distributions
modeldates
timeseries
userdates
These steps are automatic if the interface is used to define the job.
6.3 Submitting a Batch Run
To submit a SHEDS run in batch mode, use the Run dialog on the Windows Start menu.
Enter the full path to the "multimedia.bat" file (with quotes needed if the path contains blanks or
spaces), followed by a space and the name of the job to be run. For example,
"C:\SHEDS\multimedia.bat" jobl
Do not quote the name of the job itself, even if it contains spaces. To submit the job "lawn and
garden" (assuming that the "multimedia.bat" file is located in "C:\SHEDS"), type
"C:\SHEDS\multimedia.bat" lawn and garden
and hit the return key to submit the commands.
A black DOS window should appear, along with another SAS window indicating the location of
the source code, output, and log files. The rest of the output from the run should be located in
the \output directory under the job name.
75
-------
6.4 Uncertainty Runs
Only two modifications to the above process are necessary to submit an uncertainty run. First,
the user must ensure that the "distributions" input file contains the data for the uncertainty
clouds. This occupies four variables: "n", "vl_un", "v2_un", and "v3_un". Each record except
the probability vectors can be assigned uncertainty clouds (simply set "n" to zero on the
probability vectors or any other variables without uncertainty clouds). The variable "n"
indicates the number of points in the uncertainty cloud for the given input variable. This is also
the number of items in the list for "vl_un" (and "v2_un" and "v3_un", if needed). For example
if n=3, a valid list for "vl_un" would be ", 0.8, 0.9, 1.0". Each number, including the first, is
preceded by a comma. Spaces can be added as needed to make the data more readable. For point
distributions, only "vl_un" is needed. For triangles, all three are needed. All other shapes
(indicated by the variable "form") require both "vl_un" and "v2_un", but not "v3_un".
The values at the corresponding positions in each of the lists "vl_un", "v2_un", and "v3_un" are
to be used together. For example, for a uniform distribution, "vl_un" contains values for the
lower bound, while "v2_un" contains corresponding values for the upper bound. A method
called the "modified bootstrap approach" for generating suitable values for the uncertainty clouds
is discussed in the SHEDS Technical Manual. Once suitable data for "n", "vl_un", "v2_un", and
"v3_un" have been added to the "distributions" file, these data can remain there, even when
using the file for standard variability-only runs.
The other modification needed for an uncertainty run is to select the number of populations to be
run. For each population, sample values for the distribution parameters are drawn from each
uncertainty cloud. Everyone in the population then draws samples from distributions using these
parameters. Suppose one decides to run 200 populations of 500 persons each. Then the third
variable on the Runlnfo file (labeled "population size") is set to 500, while the fourth variable
(labeled "uncertainty iterations") is set to 200. For variability-only runs, the number of
uncertainty iterations is set to zero.
Uncertainty runs are submitted in exactly the same way as other batch runs.
Note that uncertainty runs may take a long time to complete. A run of 200 populations of 500
persons each takes the same time to complete as a variability-only run of 200x500 = 100,000
persons. This may take several days.
Once the run is complete, the user may run the macro %UncSummary that is built into the
SHEDS-Multimedia code. The user must first ensure that the SAS library "out" points to the
\output directory for the selected job. The %UncSummary macro takes three arguments. The first
is the variable to be analyzed, for which a common choice is "abstot", the time-average of the
total daily absorbed dose across all pathways. The second argument is the choice of units, either
76
-------
"ug" for quantities expressed in micrograms, or "mgkg" for quantities expressed as milligrams of
chemical per kilogram of body mass. The third argument is the lower bound for the output
values. This macro produces two graphs with logarithmic scales, which are used since the
output typically covers several orders of magnitude. However, in some cases the output values
may be zero or very close to zero. To avoid trying to plot such values on a logarithmic scale,
these are replaced by values at the designated lower bound.
Note that the code for producing uncertainty plots is still under development and not considered
to be fully operational in version 3, so the user might have to debug or modify the code if
problems are found.
6.5 Sensitivity Runs
There are various types of sensitivity analysis available in SHEDS, as discussed in the Technical
Manual. The Pearson and Spearman correlation methods, along with the stepwise regression
approach, can be run on any model run (either uncertainty or variability-only), as long as the
"keep intermediate variables" option was selected. This appears as a check box on the "Specify
Simulation Information" screen in the interface, or can be set directly in the Runlnfo file by
setting the variable "Save intermediate variables" to one.
The final method of sensitivity analysis currently available is the percentile scaling approach.
This requires a special model run. Somewhat like uncertainty runs, a percentile scaling run
requires two modifications by the user. First, the variable "pctSensitivity" (which is labeled as
"Sensitivity Run") on the Runlnfo file should be set to one. Second, the user should copy the file
"variables" from the \default directory to the \input directory for the job, and change the setting of
the "sensitivity" column to select the variables to be analyzed. Variables for which sensitivity=l
will be analyzed, while those with sensitivity=0 will not.
Sensitivity runs are submitted exactly like any other runs. The program will perform a run for
the base population, plus two additional populations for each selected variable. Hence, if 10
variables are selected, the job will take 21 times longer than a standard variability-only run.
The SHEDS-Multimedia version 3 code has two built-in macros for processing sensitivity
output. The first, %SensPct, is used for percentile scaling runs. It cannot be invoked via the
interface. Instead, use a standard SAS editor window and submit the macro call. Two arguments
are required: the analysis variable (often "abstot", as for uncertainty analysis) and the units
(either "ug" for micrograms of chemical, or "mgkg" for milligrams of chemical per kilogram of
body mass). Also, the SAS library "out" must point to the \output directory for the selected run.
The %SensPct macro creates a table listing the influences of the input variables selected for
analysis.
The second, %SensCorr, is used with the output from a standard variability-only run or with
output from an uncertainty run. As with the other macros discussed above, the the SAS library
77
-------
"out" must point to the \output directory for the selected run. The %SensCorr macro takes four
arguments. The first is the output variable to be analyzed (often "abstot"), the second is for the
units (either "ug" or "mgkg", as above), the third selects the type of correlation used ("P" for
Pearson, or "S" for Spearman), and the fourth is a switch (use "1" to first take logarithms of the
output variable, or use "0" to use the output variable directly). Since exposure and dose
distributions are often positively skewed, taking logarithms may create more symmetrical
distributions. As for uncertainty runs, the macros for sensitivity runs have undergone preliminary
testing only, and the user may have to identify and/or correct and problems encountered with this
code.
78
-------
Appendix A. Directories and Fi
All of the directories that SHEDS-Multimedia uses are placed underneath the installation
directory which is chosen by the user at install time. By default this is in the user's My
Documents/Multimedia3.14 (or similar) directory. The install directory will be referred to as the
install directory or in the following discussion and diagrams.
A.I Directories
The directories are laid out as follows on installation. The indentation of a directory name
implies that it is within the directory above it.
Data
Prg
Runs
Setup
Default
Diet
Demonstration File
Input
Output
Default and constant input files
Currently unused
SAS catalogs and macros implementing model
All input and output related to specific runs
All input and output related to demonstration run
User edited files for run
Results for demonstration run
Files related to installation and un-installation
The interface will create a directory for each run that the user defines. These directories will
always be named the same as the run name and by default are placed under the Runs directory.
As initially installed the Runs directory contains only one run: Demonstration File. The user has
the option to move the directory for each run to another location.
A.2 Critical Input Files
If using the interface, it is not necessary for the user to know where files are stored. For batch
runs and advanced post-run analysis it becomes important. The table below shows where critical
files are placed at installation or when a run is defined through the interface. All files in the
Runs//Input directory are modified by the interface (except the
contactmedia.sasVbdat dataset). Additional input files (e.g., CHAD diaries) needed for runs can
be found in the /Data/Default/ directory.
/
autoexec, sas
multimedia.bat
SAS start-up file. Many SAS options can be set
here
File used to run SHEDS in batch mode
A-l
-------
runinfo.sasVbdat
File containing run specific information
generated by the interface
Runs//Input/
Required for
Batch Runs
One of These is
Required
One of These is
Required
agegroups . s asVbdat
cooccurrence, s asVbdat
diarydetails.sasVbdat
distributions. sasVbdat
contactmedia. s asVbdat
appmedia_decay.sas7bdat
appmedia_postapp. sasVbdat
timeseries. sasVbdat
modeldates. sasVbdat
userdates. sasVbdat
Defines age and genders to be used in
simulation
Scenario co-occurrence data, if not being used it
may contain no records, but must be present
Contact probabilities
Variable distributions as defined for run
List of contact media and their short names
Decay and dispersion application information
Post- application information
User defined time series concentrations for each
media
Data required for model to determine stochastic
application dates
Data for user defined application dates
A.3 User Specified Output Files
The major output files for a typical variability run, their structure, and variable definitions are
discussed in the SHEDS-Multimedia Technical Manual. Files related to sensitivity and
uncertainty runs are discussed in the section on batch runs in this manual.
If a log file is specified by the user, on the Simulation Information screen, then it is put in
/Runs//Output/saslog.txt. This file can be read in the SAS editor or any
other editor of the user's choosing.
If a user requests a spreadsheet of inputs, on the Run screen, it will be placed in the installation
directory and named after the run name: /
-------
A.5 Exporting SAS Datasets
To facilitate additional analyses in the software of your choosing, you may export data in a
variety of formats. The export wizard can be accessed by selecting a dataset in SAS and then
selecting Export Data from the main SAS File menu. For more information on the export wizard
check the SAS help file index under "exporting data/Export Wizard for". To export data
programmatically, use Proc Export. Again, information on specifics can be found in the SAS
help files.
A.6 Moving Run Files
It is relatively easy to move model results for analysis outside of the interface. The easiest
method is simply to copy the entire output directory to another location and establish a SAS
library pointing to the new directory.
If you want to use existing simulation inputs or outputs or both in the interface it is a little more
difficult and requires some basic knowledge of SAS. Integrating these files on another computer
with SHEDS Multimedia installed is a three step process.
1. Move the entire run directory to the Runs directory on the new computer.
2. Integrate the related records from the Runlnfo file (in the directory.
3. Run the %RunInfoAddDirs macro on the Runlnfo file on the new machine.
A.6.1 Background on the Runlnfo Dataset
The Runlnfo dataset is stored in the directory. When the interface is running the library
is also called Runlnfo, so the dataset is referred to as Runlnfo.Runlnfo. In the windows explorer
it is named /Runinfo.sas7bdat.
The Runlnfo dataset stores user supplied information defining each simulation. The model reads
this information during the run. On installation only one run is defined, the Demonstration File
run. Each run is stored as one record with the run name being the unique identifier; records are
selected based on the run name. When the interface displays the available run names it gets them
from this file. Obviously having records in this file that don't have a corresponding directory in
the /Runs directory will cause problems. If there is no record in this file pointing to an
existing directory, a user will not be able to use the interface to access the inputs or outputs
associated with that run.
A.6.2 Moving the Run Directory
SHEDS-Multimedia should have a Run directory on the old and new machines. Use the windows
explorer to copy the entire run directory for a specific run. That is, copy the directory containing
the run name and every thing under it. If you want to move several runs copy each run directory.
Assuming that the run name was MyRunl you would copy the directory
A-3
-------
//Runs/MyRun 1,
to
//Runs/MyRun 1.
In this case and represent the full paths to the installation directories on
each computer.
A.6.3 Integrating the Old and New Runlnfo Records
There are two possibilities for this. The easiest is that you simply want to replace the Runlnfo file
on the new computer. In this case you will lose access to any previous runs on the new computer.
The other possibility is that you want to integrate one or more runs from the old computer into
the current runs on the new computer.
A.6.4 Replacing the Runlnfo File (Dataset)
If you do not need the previous runs on the new computer then copy the SAS dataset
/runinfo.sas7bdat from the old computer to the new one. To keep things clean you will
want to delete any records representing runs you are not moving to the new machine.
A.6.4.1 Integrating Records into the New Runlnfo File
If you want to maintain access to the previous runs on the new computer then you need to
integrate individual records from the old computer into the Runlnfo file on the new computer.
Use a SAS data step or Proc SQL to merge the desired records from the old Runlnfo file into the
new one. The records being integrated will need to have unique names.
A.6.5 Running the %RunlnfoAddDirs Macro
The Runlnfo dataset stores the full paths to several files and directories. The records for the runs
that were moved to the new machine need to be altered so that these paths point to the proper
location. If you were lucky enough that SHEDS Multimedia is installed on the same path on both
machines, then you can skip this step. The % Runlnfo AddDirs macro will update the path names.
It needs to be run once for each run that was moved. You give it two arguments:
installDir The current (new) installation directory.
runName The run name pointing to the record that needs to be altered.
The macro is in the file /prg/Run!nfoMacros.sas.
A.6.6 Moving a Run: An Example
For this example assume that you used the Windows Explorer to move the run directory
'MyCopyS' to a new machine. You also copied the Runlnfo file from the old machine to the
installation directory on new machine and renamed it OldRunlnfo. The new installation
directory is named c:\Multimedia3.1 IV The following code will integrate the record of interest
from the Runlnfo file and correct the paths in the on the MyCopyS record in the Runlnfo file.
* Define the runinfo library;
A-4
-------
libname runinfo "c:\multimedia3.11\";
Add the record from the old runinfo file to the new one;
This can add multiple records by adding extenting the match list;
data runinfo.runinfo;
set runinfo.runinfo
runinfo.OldRunlnfo (where=(runname in ("MyCopyS")));
run;
Including the RunlnfoMacros.sas file compiles the macros;
%include "c:\multimedia3.11\prg\RunInfoMacros.sas";
Finally, redefine the installation directory in all the paths;
%RunInfoAddDirs(installDir=C:\multimedia3.11\, runname=MyCopy3) ;
A-5
-------
Appendix B. Reference Tables
B.I Variable Names
Table B-1. Variable Names From Distributions File
Group
Activity-related
Activity-related
Activity-related
Baths
Chemical properties
Chemical properties
Chemical properties
Chemical properties
Direct-ingestion
Direct-ingestion
Dose-related
Dose-related
Dose-related
Dose-related
Dose-related
Dose-related
Dose-related
Dose-related
Other
Other
Other
P rob Vector
P rob Vector
P rob Vector
P rob Vector
Transfer Coefficients
Transfer Coefficients
Transfer Coefficients
Variable
washprob
hm_freq
f_uncloth
bathdays
dermaxh
metab_ratio
object_ratio
dermaxb
ingestion_indoor
ingestion_outdoor
absr_gr
absr_gm
bioavm
bioavr
elimr_blood
absfjung
absr_dr
absr_dm
has_lawn_p
has_pet_p
has_garden_p
timeofdayjndoor
timeofday_outdoor
re-entry_indoor
re-entry_outdoor
tc_m
tc_b
tc_h
Label
mean # hand washes/day per person
hand mouthing events per hour
fraction of body unclothed
maximum number of days between baths
maximum dermal loading for hands
mass ratio (metabolite/pollutant)
object-surface concentration ratio
maximum dermal loading for body
dust ingestion rate (indoor, direct only)
soil ingestion rate (outdoor, direct only)
Gl tract absorption rate per day for surface residues
Gl tract absorption rate per day for dust or soil
bioavailability fraction for dust/soil
bioavailability fraction for surface residues
elimination rate from the blood
absorption fraction for lungs
dermal absorption rate per day for surface residues
dermal absorption rate per day for dust or soil
probability of having a lawn
probability of having a dog or cat
probability of having a vegetable garden
probability vector for hour of indoor application
probability vector for hour of outdoor application
probability vector for re-entry time indoors
probability vector for re-entry time outdoors
transfer coefficient for object mouthing
surface-skin transfer coefficient for body (unclothed)
surface-skin transfer coefficient for hand
Units
1/day
events/hr
[-]
days
ug/cm2
[-]
[-]
ug/cm2
mg/hour
mg/hour
1/day
1/day
[-]
[-]
1/day
[-]
1/day
1/day
[-]
[-]
[-]
[-]
[-]
[-]
cm2/hr
cm2/hr
cm2/hr
B-1
-------
Transfer
Transfer
Transfer
Transfer
Transfer-
Transfer-
Transfer-
Transfer-
Transfer-
Transfer-
Transfer-
Transfer-
Efficiency
Efficiency
Efficiency
Efficiency
related
related
related
related
related
related
related
related
om_transfer
transfer_dermal
om_freq
om_area
remv_bath
remv_dry
remv_mouth
remv_wash
contactb
contacth
adherence
hm fraction
object-mouth transfer efficiency [-]
residue-skin transfer efficiency [-]
object-mouth contact rate events/hr
object-mouth contact area cm2
removal efficiency during bath/shower [-]
removal efficiency during events without water 1 /hr
removal efficiency during mouthing [-]
removal efficiency during hand washing [-]
Body-surface fractional contact rate 1/hr
Hand-surface fractional contact rate 1/hr
soil-skin adherence factor mg/cm2
fraction of surface of one hand that enters mouth [-1
B.2 Media Descriptions
Table B-2. Media Descriptions And Symbols
Media Description
Media Symbol
(medsymbol)1
Examples
Indoor air
Air outside the residence
Residue on indoor smooth surfaces
Residue on indoor textured surfaces
Indoor dust
Residue on lawn
Residue on vegetable garden
Residue on pet fur
The model uses these internally
ATair, AUair
AYair
RThard, RUhard
RTsoft, RUsoft
MTdust, MUdust
RYlawn
RYveg
Rpet
Anone, Mnone, Rnone
Wood floor, laminate counter tops
Carpet, upholstered furniture
1 The capital letters Y, T, and U indicate yard, treated, and untreated.
B-2
-------
Appendix C. Probability Density Functions
The explicit probability density functions (pdf s) utilized by SHEDS are listed below. Note that
some of these may have alternate parametrizations, so the user must be careful when obtaining
distributions from the literature. The expressions "Exp," "Sqrt," "Log," and "F" refer to the exponential,
square root, natural logarithm, and gamma functions, respectively.
C.I Beta
The beta distribution in SHEDS has a lower bound of zero, an upper bound of one, and two shape
parameters vl and v2. The restrictions are vl>0 and v2>0. When vl< v2 then the mean is below l/2 and
the distribution is positively skewed, whereas when vl > v2 the mean is above l/2 and the distribution is
negatively skewed. For vl=v2 the mean is at l/2 and the shape is symmetric. The PDF is
Equation E-l
p(x)= xvl"1(l-x)v2"1r(vl+v2)/(r(vl)r(v2)), forOl and v2>l, the PDF of the beta has a single peak, away from the bounds. When
0vl
C.3 Gamma
The gamma distribution in SHEDS is bounded below by zero and has two parameters, the shape
parameter vl and the scale parameter v2. The restrictions are vl>0 and v2>0. The shape parameter vl
controls the appearance of the PDF. Shape parameters less than one lead to a monotonically decreasing
form with the highest probability at zero. If vl=l, then the gamma is identical to an exponential that starts
at zero and has a mean given by the gamma parameter vl. If vl>l, then the gamma somewhat resembles
the lognormal, rising from zero to a peak probability, and then gradually declining with an overall positive
C-l
-------
skewness. The mean of the gamma is at |a = vl v2, and the standard deviation is o = v2 Sqrt(vl). The
PDF of the SHEDS gamma is
Equation E-3
p(x) = v2"vl x vl-1 Exp(-x/v2) / T(vl), for x>0
C.4 Lognormal
The lognormal in SHEDS is bounded below by zero and has two parameters, the geometric mean
GM (vl) and the geometric standard deviation GSD (v2). The restrictions are vl>0 and v2>l. Many
variables in exposure science are approximately lognormally distributed, so its use is fairly common. If a
variable 'x' has a lognormal distribution, then log(x) has a normal distribution.
The geometric mean (GM) of a lognormal distribution is also its median. Log(GM) is the mean of
the distribution of log(x). Log(GSD) is the standard deviation of log(x). Since standard deviations must
be positive, then Log(GSD)>0, which implies GSD> 1. The PDF of the SHEDS lognormal is
Equation E-4
p(x) = Exp[(-l/2) (Log[x/vl] /Log[v2])2 ] / (x Sqrt[2 n ] Log[v2]), for x>0
If GM and GSD are given, then the lognormal has arithmetic mean and standard deviation
Equation E-5
H = GM Exp[(l/2) (Log(GSD))2 ],
G = GM Sqrt[Exp( [Log(GSD)]2) (Exp( [Log(GSD)]2)-!)].
If the user knows the arithmetic mean \i and arithmetic standard deviation G of the lognormal
instead of the GM and GSD, then these can be converted as follows:
Equation E-6
GM = u
GSD = Exp(Sqrt(Log(l + o2/)/))).
If instead, one has the mean |iiog and standard deviation ciog of log(x), then use
Equation E-7
GM = Exp(^iog),
GSD = Exp(oiog).
C.5 Normal
This is the normal or Gaussian distribution commonly used in statistics. The normal has two
parameters: the mean (vl) and the standard deviation (v2), with v2>0. Note that the normal is unbounded,
so it is a good idea to provide lower and upper truncation points to prevent physically impossible values
from being returned. The PDF of the normal is
Equation E-8
C-2
-------
p(x) = Exp[-(x-vl)2/(2 v22)] / (Sqrt[2 TC] v2)
C.6 Point
A point value means that the same value is always returned. This is sometimes called a fixed or
constant form. The point has one numeric argument (vl) which is the value that is to be returned. The
mean is vl and the standard deviation is zero. The sampling frequency does not matter for points. While
points are technically discrete, here they are classified with the continuous distributions since they are
applied to variables that are expected to reside on a continuous scale, but happen to be assigned no
variability.
C.7 Triangular
The triangular distribution has a probability density function (PDF) that is shaped like a triangle.
The three parameters locate the vertices, with vl=minimum, v2=peak, v3=maximum. The restrictions are
vl<= v2 <= v3, with vl0 and v2>0. The Weibull has slightly
different properties from a gamma, but there is a strong overall resemblance. When the shape parameter
vl < 1, the Weibull is monotonically decreasing. For vl=l, it reduces to an exponential. For vl>l, it rises
to a peak and then declines gradually in a long tail. The mean is |a = v2 F(l+l/vl), and the standard
deviation is o = v2 Sqrt[F(l+2/vl) - (F(l+l/vl) )2]. Here T' is the mathematical gamma function, not
the gamma distribution. The gamma function is a generalization of the factorial function to non-integer
arguments; for integers, F(l+n) = n!. The PDF of the Weibull distribution is
C-3
-------
Equation E-ll
p(x) = vl v2"vl xvl4 Exp [-(x/v2)vl], for x>0.
C.10 Discrete Probability Density Functions
Bernoulli variables have only two possible outcomes, for example, the outcome of yes/no tests.
The usual statistical notation would be (pi, p2), where pi and p2 are between 0 and 1 and sum to 1.
However, the SHEDS code implements Bernoulli variables by having the user specify only the probability
of a "yes". The probability of "no" is implied by 1 - "yes".
Multinomial variables allow more than two possible outcomes. The usual statistical notation would
be (pi, p2, ..., pn), where each pas between 0 and 1 and the pi's sum to 1. In SHEDS, multinomial
variables are called probability vectors. Unlike Bernoulli variables, the user must supply the entire set of
probabilities.
C-4
-------
vvEPA
United States
Environmental Protection
Agency
(8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA600/R-08/118
September 2008
www.epa.gov
Recycled/Recyclable Printed with
vegetable-based ink on paper that
contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |