Simple Environmental Exposure Models in a GIS Framework

 

By Lesley Hay Wilson

Send e-mail to: hay_wilson@mail.utexas.edu

 

http://www.ce.utexas.edu/stu/haywilli/report.html


 Table of Contents


 

Introduction

 

Risk assessment is fundamental to the development of environmental policy and regulation and in environmental management. One key component of the risk assessment process is environmental exposure assessment. Exposure assessment is the quantitative description of the pathway by which a chemical of concern moves from a source through environmental media to a potential receptor. The assessment describes what happens to the chemical in the environmental media. This includes an analysis and representation of the chemical or physical interaction between the chemical and the media. Examples are the photo-catalyzed oxidation of an airborne chemical; or chemical complexation between a chemical dissolved in groundwater and the aquifer soil material.

By its very nature the process of exposure pathway analysis is a spatial analysis. This project explored exposure assessment in a Geographic Information System (GIS) context. The specific application investigated is the use of a simple analytical model to describe the airborne migration of chemicals of concern from a source to a receptor and the use of a GIS application to track and display the results of the model calculations. 

Environmental modeling is the key analytical component of exposure assessment. The modeling is used to determine the environmental impacts of a release, or the potential impacts of a hypothetical release, from an existing or planned facility. The effects of a release must be communicated in an understandable fashion so that the results can be properly evaluated by the environmental managers and policy makers (Charbeneau, 1997). For this reason, the GIS application framework is well-suited to environmental exposure assessment. This project has provided the opportunity to explore and experiment with capabilities of a GIS for analyzing and tracking exposure pathways. The aspect of environmental exposure assessment that was studied here is the transport from a source to a receptor. The example, described below, assumes an instantaneous release of a chemical to the atmosphere. The procedure gave rise to the development of the predictive tools to describe with maps and concentration data the effects downwind of a source. Further analysis in a site-specific application would include the evaluation of potential receptors, exposure routes at the downwind point and the consequences of the exposure. The algorithm is independent of the nature of the material released, as long as the material does not undergo a chemical or physical reaction in the atmosphere.

 An exposure pathway can be described with four (4) components:

 

Source: The source term is a description of the mechanism by which the chemical of concern is released into the environment. The release may be unintended or may be a regular emission from an industrial or manufacturing facility.

Transport Mechanism: The transport is dictated by the interactions between the specific released material and the environment. Examples of transport mechanisms include wind erosion of soils and atmospheric dispersion of particulates, volatilization of the chemical of concern and atmospheric dispersion, migration through vadose zone soils and dissolution in groundwater, movement across the land surface with storm water or surface water.

Exposure Routes: These are the mechanisms by which a receptor is exposed to the chemical of concern. Examples are dermal contact, ingestion, direct radiation and inhalation.

Receptors: These are the humans, plants and animals that are or may be exposed to the chemicals of concern.

 

Case Example

The situation to be modeled here is an instantaneous release of a gaseous or particulate substance to the atmosphere. Most of the analytical solutions for this transport mechanism are based on the Pasquill-Gifford Model (Mycock et.al., 1995). The process assumes that the release is either a constant point source that creates a "plume" or it is an instantaneous point source release of a known mass that generates a "puff." For this project a puff release was studied. The downwind concentrations are estimated based on the following assumptions (Mycock, et. al., 1995):

 

The applicable equation describing concentrations at ground level given a puff release at height H is:

 

C(x,y,0,t,H) = [QT/(7.875*sx*sy*sz)] exp(-0.5*(x-ut/sx)^2)exp(-0.5*(H/sz)^2)exp(-0.5(y/sy)^2)

Where:

QT = total mass released (g)

sx = longitudinal dispersion coefficient (m)

sy = lateral dispersion coefficient (m)

sz = vertical dispersion coefficient (m)

 H = height of the release (m)

 u= average wind speed (m/sec)

 t= time (sec)

 x,y, = location of the point of interest, relative to the source location.

 (Mycock, et. al., 1995, Chapter 17, equation 25).

 

The maximum concentration at any downwind distance can be determined as the centerline concentration (y=0). The maximum concentration is reached at time t = x/u. The dispersion coefficients are determined based on the atmospheric stability class and the distance from the source. The values are available in many air pollution texts as log-log plots of dispersion coefficient vs. distance. There are six (6) stability classes, each has a different curve on the dispersion coefficient plots.

The atmospheric stability classes range from class A, which is the most unstable, to class F which is the most stable (Till and Meyer, 1983). The classes depend upon the average wind speed at a height of 10 m, the amount of incoming solar radiation during the day and the amount of cloud cover at night. If the conditions are known at the time of the release, then those conditions would be used to determine the stability class, see Table 1, Chapter 17, Mycock, et. al., 1995. If however, the modeling is being conducted as part of an exposure analysis for a prospective evaluation, then the assumption of stability class F will result in the smallest dispersion coefficients and therefore the highest concentrations at the downwind points.

Data are needed for average wind speeds and directions for a particular study location. Several sources were consulted including Stern et. al., 1984, Mycock, et. al., 1995 and Till and Meyer, 1983. The National Oceanic and Atmospheric Administration NOAA web site includes various climate databases and historical records that include wind data. In addition, for Texas, the West Texas A&M University web site includes maps of wind data collected for their alternative energy research. Climate data are generally collected by the USDA Agricultural Experiment Stations for a particular region. Data from the Bushland Texas Conservation and Production Research Laboratory (USDA, 1996) were reviewed for this project.

 

 

 Visual Basic Formulation

The environmental transport is simulated using the concentration prediction model presented above in a windows-based model developed in Visual Basic. Visual Basic programs are event-driven and provide a graphical user interface (GUI). In visual basic an object is either a form (window) or a control. The object has properties associated with it. The properties control the way that an object behaves. Objects respond to events such as mouse clicks (Aitken, 1997). In the case of this application, the model calculation is invoked when the input boxes contain changed values. The code is constructed so that the control objects (i.e., the input boxes) call the dispersion calculation when their values are changed. The code therefore connects the objects to the calculations and provides the user with the convenience of a GUI for input and output.

A calculation form was developed for the gaussian puff equations. Additional specific information about the code is available in the Procedures section below. The output includes the maximum centerline concentration at the given distance and the x and y coordinates of a receptor point entered as a distance from the source. In an emergency planning or bounding calculation scenario, the most important information is this maximum concentration and the time to reach that concentration under various atmospheric conditions. This provides the information needed to determine if the pathway will be important, compared with other pathways, in a detailed analysis.

The code can be access and downloaded from: ftp.crwr.utexas.edu/pub/gisclass/airD/gaussianpuff.exe. The windows driver library file, MSVBM50.ddl is necessary to operate the executable file.

 

Model Runs

The model performance was tested using two scenarios. The scenarios include climate conditions typical for Texas and are based on available information, (USDA, 1996). A hypothetical facility, The Sample Company, was selected in the vicinity of Fort Worth for the test calculations.

A digital raster map was obtained from the Texas Department of Natural Resources and Conservation Commission (TNRCC) web page. The digital raster graphics are 7.5 minute USGS quadrangle maps that are 1:24,000 scale. The TNRCC does not currently have digital maps for all areas in Texas. The Fort Worth sheet was chosen because it was available at this scale. The maps have been developed and projected in the Texas State Plane Mapping System. The features of this projection are included in the data dictionary below.

 

The scenarios are described in the following section. Five (5) regularly spaced points were selected to run the dispersion model and generate the output concentrations. In a specific facility situation, the actual distances and directions to identified receptors would be used. Based on the direction to the actual receptors, the average wind speeds in those directions would be used in the calculations. The time to reach the maximum concentration is a function of the wind speed. It is the atmospheric stability class that dictates the amount of dispersion and consequently, the down-wind concentrations. For these scenarios the release is assumed to be at ground level, this translates to a stack height (H) input to the model of 0 m. This is a conservative assumption, assuring the concentrations calculated are maximal.

 

Scenario No. 1:

Wind Speed = 4.5 m/sec

Wind Direction = North, corresponding angle from north, 180 degrees.

Stability Class = F

Mass Released = 100 kg

Number

Name

Distance (m)

Time to Maximum Concentration (min)

Concentration (g/m3)

Relative Coordinates (x, y)

Actual Coordinates (x, y) (m)

1

Source

0

0.00

100,000

0

0

1239414

1184770

2

Point1

100

0.37

330.7

0

-100

1239414

1184670

3

Point2

500

1.85

3.968

0

-500

1239414

1184270

4

Point3

1000

3.70

0.7994

0

-1000

1239414

1183770

5

Point4

2000

7.41

0.1603

0

-2000

1239414

1182770

6

Point5

5000

18.52

0.0202

0

-5000

1239414

1179770

 

Scenario No. 2:

Wind Speed = 4.5 m/sec

Wind Direction = Southwest, corresponding angle from north, 45 degrees.

Stability Class = D

Mass Released = 100 kg

Number

Name

Distance (m)

Time to Maximum Concentration (min)

Concentration (g/m3)

Relative Coordinates (x, y)

Actual Coordinates (x, y) (m)

1

Source

0

0.00

100,000

0

0

1239414

1184770

2

Point1

100

0.25

39.68

70.71

70.71

1239485

1184841

3

Point2

500

1.24

0.3968

353.55

353.55

1239768

1185124

4

Point3

1000

2.49

0.0567

707.1

707.1

1240121

1185477

5

Point4

2000

4.98

0.0113

1414.2

1414.2

1240829

1186185

6

Point5

5000

12.44

0.0014

3535.5

3535.5

1242950

1188306

 

 

The stability class is used to determine the lateral and vertical dispersion coefficients. The values in the table below were read from the log-log plots of dispersion coefficient as a function of downwind distance (Mycock et. al., 1995).

 

 Distance (m)

Stability Class D

Stability Class F

 

s y (m)

s z (m)

s y (m)

s z (m)

100

8

5

4

2.4

500

40

20

20

8

1000

80

35

38

11

2000

150

50

60

22

5000

320

90

140

32

 

Results

The Procedures section below details the functions performed to obtain the input maps, convert the model results into a point coverage in Arc/Info and attribute the coverage in ArcView. The results from Scenario 1 are presented here.

 

The following map presents the results from Scenario 2.

As expected from the stability classes chosen, the maximum concentrations are higher at the given distances under Scenario 1 with stability class F, than under Scenario 2 with stability class D. The dispersion coefficients are smaller under a more stable atmosphere. The lower wind speed in Scenario 1 also means that the maximum concentrations are reached at a later time than in the higher wind speed case.

Graphs of concentrations vs. distance and time to maximum concentrations were developed in Excel, although a basic charting utility exists in ArcView. Excel is a superior tool for plotting scatter data and in the case of the concentration vs. distance, better for plotting logarithmic data.

Concentration as a function of Distance:

Time to Maximum Concentration as a function of Distance:

 

Several hand calculations were performed for the Scenario 1 input data to verify the model results. The model and hand calculations were consistent. The model accurately performs the required calculations. It is very easy to generate output data for varying input conditions because the model recalculates the results each time an input field is changed. Converting the model results to a point coverage and displaying the results in ArcView is a good way to view the results. The impact of the release and the changes in the concentrations with variations in input parameters are easily compared in the ArcView GUI.

Exposure assessment modeling can be conducted very effectively using visual basic codes and ArcView. Visual basic is user friendly for programming analytical solutions for the calculations. The GUIs and controls are easy to develop and provide a convenient way for a user to develop modeling results. The spatial nature of exposure assessment lends itself to analysis in a GIS application.

 

A New Framework for Exposure Assessment

Through this project, a new model for completing an integrated exposure analysis has been developed. This spatial environmental exposure model is made up of five (5) components:

 

To complete an integrated analysis all of the components are needed. The GIS is the focal point of the model. The exposure calculations must be connected to the GIS to evaluate the consequences to the receptors from the given sources. The GIS is central to the model because it holds the objects, the sources and the receptors, and defines their relationships. The GIS relates the site data to geographic features and provides the ability to display the results of the environmental modeling calculations. The Site Database holds the characterization data, including information about source concentrations, population densities of receptors and transport data, for example, wind speeds and directions. The information stored in the site database is connected to the geographic features as attributes of the data layers in the GIS. In a simple case, the site database might actually reside completely within the attribute tables for the coverages in ArcView. More often, in the case of a large facility evaluation, there will be a separate tabular site database from which subsets of information would be gathered into ArcView. Within ArcView, spatial hydrologic modeling can be performed directly on the data layers using the ArcView Extensions. As has been demonstrated in this project, Visual Basic Codes can be used for screening calculations to connect and relate the sources and receptors through the analytical descriptions of the transport mechanisms. The GUIs are easily developed and analytical solutions are easy to program. For other, more complex pathways, for example to examine explosive releases, or for a subsequent, more detailed analyses of pathways evaluated initially in the visual basic screening calculations, Industry Standard Codes should be used and interfaced with the site database and the GIS. For presenting results to the public, MapObjects may be used to digest results and present them in a map form that can be made available on the web, can be queried for results and does not require ArcView for access. More information about ArcView and MapObjects can be obtained from the ESRI web page. A related discussion of spatial database design appears in Maidment, 1997.

In order to implement this new spatial exposure assessment model, for a specific facility, a number of tasks would need to be accomplished:

 

Future Work

Specific activities to provide additional capabilities for the puff release scenario include:

 

References

Aitken, P., 1997. Visual Basic 5 Programming Explorer, Coriolis Group Books, Scottsdale AZ, 1997.

Charbeneau, R.J. 1997. Groundwater Hydraulics and Pollutant Transport, 1997.

Maidment, D. R., 1997. "Spatial Database Design for Environmental Risk Assessment," unpublished, 1997.

Mycock, J. C., McKenna, J. D., Theodore, L., 1995. Handbook of Air Pollution Control Engineering and Technology, CRC Press, Inc., New York, NY, 1995.

Stern, A. C., Boubel R. W., Turner, D. B., Fox, D. L., 1984. Fundamentals of Air Pollution, Second Edition, Academic Press, Inc., Orlando, FL, 1984.

Till, J. E., Meyer, H. R., 1983. Radiological Assessment, NUREG/CR-3332, ORNL-5968, USNRC, 1983.

USDA, 1996. "USDA - ARS Conservation and Production Research Laboratory, Bushland, TX, Climate Summary, 1996."

 

Appendicies 

The following sections document the electronic data used for this project and the software procedures used to develop the results.

 

 

Data Dictionary

Base Map: 3297-432.tif, 3297-432.tfw

A Digital Raster Graphic (DRG) file is a raster image of a scanned topographic map including the geographic references to a map coordinate system. The DRGs available through TNRCC use the Texas State Wide Mapping System. The scanned maps are the 7.5 minute (1:24,000 scale) USGS maps. The maps are current as of the USGS 7.5 minute Quadrangle Maps of October, 1994.

Spatial Data Organization: Raster

Raster Object Type: Pixel

Row Count: 7500

Column Count: 5750

Horizontal Coordinate System: Planar

Map Projection: Lambert Conformal Conic

Longitude of Central Meridian: -100

Latitude of Projection Origin: 31.1666666

Latitude of First Standard Parallel: 34.91666666

Latitude of Second Standard Parallel: 27.41666666

False Easting: 1000000.0000 m

False Northing: 1000000.0000 m

X-shift: 0.00000

Y-shift: 0.00000

Geodetic Model:

Horizontal Datum Name: NAD83

Ellipsoid Name: GRS1980

 

Visual Basic Model Files

AirD.vbp: The visual basic project file that identifies all of the component modules of the project, the form and the settings.

AirDispersion.frm: The form module of the model. This is the actual GUI that includes the calculation code.

GaussianPuff.exe: The executable version of the air dispersion model.

 

Data Files

Proj2.apr: The ArcView project that includes all of the layouts and tables generated in ArcView for this project.

Scen1.dat: A text file with the coordinates of the source and points of interest for scenario1.

Scen2.dat: A text file with the coordinates of the source and points of interest for scenario2.

Scendata.xls: An excel file of the attributes of Scenario 1 and Attributes of Scenario 2, exported from ArcView and used to plot the model results of concentration as a function of distance and time to maximum concentration as a function of distance.

Scenario1: A point coverage developed in Arc/Info that specifies the x,y locations in the map coordinates for the source and the five downwind points of interest in scenario 1. The file contains no attribute information.

Scenario2: A point coverage developed in Arc/Info that specifies the x,y locations in the map coordinates for the source and the five downwind points of interest in scenario 2. The file contains no attribute information.

Texas: The shape file of the county boundaries of Texas. This was used to locate Tarrant County on the location map layout.

 

Layout Files

Facmap.ps: The ArcView layout that includes the designation of the facility for The Sample Company. The file is formatted as a postscript file for printing from UNIX.

Locmap.ps: The ArcView layout that displays the location of the base map quadrangle in the Texas state map of counties. The file is formatted as a postscript file for printing from UNIX.

Scenres1.ps: The ArcView layout of the results generated for scenario 1. The file is formatted as a postscript file for printing from UNIX.

Scenres2.ps: The ArcView layout of the results generated for scenario2. The file is formatted as a postscript file for printing from UNIX.

 

Image Files

Conctbl.gif: The image file of the graph of maximum concentration vs. downwind distance for scenario 1 and 2.

Expfch.gif: The image file of an example exposure pathway.

Fac.gif: The image file of the facility map.

Locmp.gif: The image file of the location of the base map.

Model.gif: The image file of the Air Dispersion Model Window.

Scenr1.gif: The image file of the scenario 1 results.

Scenr2.gif: The image file of the scenario 1 results.

Spmodel.gif: The image file of the spatial exposure assessment model.

Table1a.gif: The image file of the ArcView table of attributes for Scenario1.

Table2a.gif: The image file of the ArcView table of attributes for Scenario2.

Timetbl.gif: The image file of the graph of time to maximum concentration for scenario 1 and 2.

 

Procedures

Visual Basic Air Dispersion Model

The following sections describe the input parameters and the assumptions used to calculate the output values.

Total Mass Released (g): This required input is the total estimated release of the chemical of concern. For this puff release scenario, the total mass is released instantaneously. For a particular facility or process this information would be derived from the normal operating conditions, the expected emissions, and from the potential accident scenarios for individual processes.

Distance to Receptor (m): The distance included here is used in the calculation as the x value. The distance is along the axis of the wind direction. This would be the distance to potential receptors in a given wind direction. If receptors exist in several directions from the facility, then each direction would be modeled separately with the average wind speed in each direction.

Source Location X: The actual easting value of the source in state plane coordinates (or the relevant projection system of the maps being used) is input here. This would be the location of the particular emission stack on a facility, or the most likely location for a release in an accident scenario.

Source Location Y: The actual northing value of the source in state plane coordinates (or the relevant projection system of the maps being used) is input here. This would be the location of the particular emission stack on a facility, or the most likely location for a release in an accident scenario.

Height of the Release (m): This would be the effective stack height. If this variable is set equal to zero the release is assumed to be at ground level and the downwind concentration will be maximized. Depending on the atmospheric conditions at the time of the release, the effective stack height could actually be higher than the physical stack, if the emission is at an elevated temperature or high velocity. The effective stack height could be lower if the atmospheric temperature actually increases with height (inversion) rather than decreasing with height (lapse), which is usually the case (Mycock, et. al., 1995).

Average Wind Speed (m/sec): Annual average values were chosen from the USDA 1996 report for the two test scenarios run. A value may be chosen as a measured value at the facility of interest at the time of the release, or a historical monthly average value used. For prospective calculations, an historical annual average may be most appropriate. As discussed above, if receptors exist in several directions from the facility, average wind speeds for each of the directions should be used.

Horizontal Angle (degrees from North): This value specifies the wind direction. A south west wind is at a 45 degree angle from North, a north east wind is at a 225 degree angle.

Lateral Dispersion Coefficient (m): Values are obtained from the log-log plots of coefficients vs. distance downwind for a specified stability class. As a first approximation in this model the longitudinal dispersion is set equal to the lateral dispersion value.

Vertical Dispersion Coefficient (m): Values are obtained from the log-log plots of coefficients vs. distance downwind for a specified stability class.

Output:

Maximum Receptor Concentration (g/m3): Equation 25 from Mycock et. al., 1995 is called when the input values are greater than zero and have changed values. The equation is formulated to calculate the concentration maximum by setting y=0 and x= (receptor distance specified in the input) = u*t.

Receptor Location X: The x coordinate of the receptor location based on the receptor distance specified in the input and the horizontal wind direction angle given in the input.

Receptor Location Y: The y coordinate of the receptor location based on the receptor distance specified in the input and the horizontal wind direction angle given in the input.

 

Accessing TNRIS Digital Raster Maps

The Texas Natural Resources Conservation Commission web page includes a clickable map of the available digital 7.5 minute USGS quadrangle maps for Texas. Each map has a USGS serial number that is listed on the map. The maps are located at www.tnrcc.state.tx.us/gis/drgtex.html. The dataset that is downloaded from the web is a zipped file. It contains a readme file, a meta data description file, drg.metadata and the 3297-432.tif file and 3297-432.tfw file. The .tif file is opened in ArcView and the .tfw file contains information needed by ArcView to use the .tif file as a geo-referenced raster image. The 3297-432.zip file was downloaded on a PC, unzipped with the pkunzip utility and ftp to the UNX machines.

 

Building the Point Coverages

The digital raster map, 3297-432.tif, was opened in ArcView as an Image Data Source Theme. The location of the hypothetical release point, The Sample Company, was read-off in projected coordinates:

(x=1239414.32 m, y = 1184770.27 m)

Using the model output values for the x,y locations of the receptors at the input distances and the source coordinates in Text Editor raw data files were created: scen1.dat and scen2.dat.

File: scen1.dat

1 1239414.32 1184770.27

2 1239414.32 1184670.27

3 1239414.32 1184270.27

4 1239414.32 1183770.27

5 1239414.32 1182770.27

6 1239414.32 1179770.27

end

File: scen2.dat

1 1239414.32 1184770.27

2 1239485.03 1184840.98

3 1239767.869 1185123.828

4 1240121.417 1185477.386

5 1240828.515 1186184.500

6 1242949.807 1188305.850

end

Arc/Info was used to build the point coverages of the source location and the downwind receptor locations.

Danube{home/hay_wilson/project}% Arc

Arc: generate scenario1

Generate: input scen1.dat

Generate: points

Generate: quit

Arc: build scenario1 points

Arc: addxy scenario1

Arc:quit

The result of this operation is a point coverage, in the Texas state mapping coordinate system called scenario1. It does not have attribute data at this point.

Danube{home/hay_wilson/project}% Arc

Arc: generate scenario2

Generate: input scen2.dat

Generate: points

Generate: quit

Arc: build scenario2 points

Arc: addxy scenario2

Arc: quit

Attributing the Point Coverages

The attributes generated from the model runs were added to the point coverages in ArcView.

Danube{home/hay_wilson/project}% ArcView3 &

The raster image file and the theme scenario1 were opened in a new View. A new data table was created using the Table icon in the Project Window. Under Edit, Add Field was used to create the attribute fields and Edit, Add Record was used to generate the six (6) records needed. The fields created are:

Name, Number, Distance (m), Time to Max (min), Concentration (g/m3)

Data from the model results were entered in the records.

The Data Table was joined to the Point Attribute Table using the Table icon in the Project Window, selecting Add and choosing Attributes of Scenario1. The Number field in the Data Table is chosen first, then the Scenario# field in the Attributes of Scenario1 is chosen. With these two fields highlighted, the Table/Join command is selected to accomplish the Table connection. The same procedures were followed to join the data for scenario 2 to the Attributes of Scenario2. The Attribute Tables for Scenario 1 and for Scenario 2 are included here.

With the raster image map and the point coverage added to the View, the legend for the point theme was edited to display unique values for concentration data with scaled marker points. The markers decrease in size with decreasing downwind maximum concentrations.

The maps included in this report, listed in the data dictionary, were constructed in ArcView as layouts. For additional information about building base maps and point coverages see Exercise 4 from the GIS class home page.