EPA-600/R-95-142a
September 1995
Statewide Mapping
of Florida Soil Radon Potentials
Volume 1. Technical Report
by
Kirk K. Nielson, Rodger B. Holt, and Vern C. Rogers
Rogers & Associates Engineering Corporation
P.O. Box 330, Salt Lake City, UT 84110-0330
EPA Interagency Agreement RWFL 93378-01
Florida Department of Community Affairs Contract 94RD-30-13-00-22-003
University of Florida Subcontract (Acct. 1506481-12)
EPA Project Officer: David C. Sanchez
U.S. Environmental Protection Agency
National Risk Management Research Laboratory
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
Prepared for:
Florida Department of Community Affairs
2740 Centerview Drive
Tallahassee, FL 32399
(DCA Project Officer: Mohammad Madani)
and
U. S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before comple ||| |||[ || ||||R III III |||| llll III
1. REPORT NO. 2.
EPA-600/R-95-142a
3 III llll II lllll III III llll ||11 HI
PB9S-104351
4. TITLE AND SUBTITLE
Statewide Mapping of Florida Soil Radon Potentials,
Volume 1. Technical Report
5. REPORT DATE
September 1995
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
KirkK. Nielson, Rodger B. Holt, and Vern C. Rogers
8. PERFORMING ORGANIZATION REPORT NO.
RAE-9226/2-2R1
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Rogers and Associates Engineering Corporation
P. 0. Box 330
Salt Lake City, Utah 84110-0330
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA IAG RWFL 93378-01
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 4/92 - 9/94
14. SPONSORING AGENCY CODE
EPA/600/13
15. supplementary NOTES APPCD project officer is David C. Sanchez, Mail Drop 54, 919/
541-2979. Volume 2 contains Appendices A~P.
16. abstract rep0rt gives results of statewide mapping of Florida soil radon poten-
tials. Statewide maps identify Florida regions with different levels of soil radon po-
tential. The maps provide scientific estimates of regional radon potentials that can
serve as a basis for implementing radon-protective residential building standards
where they are needed. The maps were developed from state soil maps and surface
geology maps, which divided the state into 3, 919 regions with unique combinations of
soil and geologic properties. The potentials of the soil profiles in each region to con-
tribute to indoor radpon levels were calculated and used to classify each map region
into one of seven tiers of radon potential. The maps were validated by comparisons
with more than 1,000 radon flux measurements and with 9, 038 indoor radon measure-
ments from three data sets. The comparisons showed consistency between the mea-
surements and the radon potential maps, with approximately the expected numbers
of outlier points. Field investigations of the outlier data points showed trends asso-
ciating certain construction details with positive or negative biases. Radon potentials
were calculated using a mathematical model of radon generation and transport from
the top 5 m of surface soils into a reference house. Upper confidence limits of 70, 90,
and 95% were defined for the radon potentials, in addition to regional median values.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b. 1DENTIF1ERS/OPEN ENDED TERMS
c. cosati Field/Group
Pollution
Radon
Soils
Mapping
Residential Buildings
Building Codes
Pollution Control
Stationary Sources
Indoor Air
13 B
07B
08G, 08M
08B
13 M
05D
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
105
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
-------
FOREWORD
The U. S. Environmental Protection Agency is charged by Congress with pro-
tecting the Nation's land, air, and water resources. Under a mandate of national
environmental laws, the Agency strives to formulate and implement actions lead-
ing to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, EPA's research
program is providing data and technical support for solving environmental pro-
blems today and building a science knowledge base necessary to manage our eco-
logical resources wisely, understand how pollutants affect our health, and pre-
vent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks
from threats to human health and the environment. The focus of the Laboratory's
research program is on methods for the prevention and control of pollution to air,
land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites and groundwater; and prevention and
control of indoor air pollution. The goal of this research effort is to catalyze
development and implementation of innovative, cost-effective environmental
technologies; develop scientific and engineering information needed by EPA to
support regulatory and policy decisions; and provide technical support and infor-
mation transfer to ensure effective implementation of environmental regulations
and strategies.
This publication has been produced as part of the Laboratory's strategic long-
term research plan. It is published and made available by EPA's Office of Re-
search and Development to assist the user community and to link researchers
with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
11
-------
ABSTRACT
This report documents the characterization of soil radon potentials to be used in state
wide soil radon potential maps of Florida. The maps are designed to show from soil and
geological features the areas that have different levels of radon potential. The soil radon
potential maps have been proposed as a basis for implementing radon-protective building
construction standards in areas of elevated radon risk and avoiding unnecessary regulations in
areas of low radon risk. The soil radon potentials calculated in this report are revised from
previous regional estimates of radon potentials to eliminate boundary faults.
Discrete areas (polygons) on the radon maps were defined from the digital intersection
of State Soil Geographic Data Base (STATSGO) soil map units with digitized geological map
units. The University of Florida GeoPlan Center defined the map polygons using a geographic
information system with Arclnfo format. The GeoPlan Center also partitioned National Uranium
Resource Evaluation (NURE) aeroradiometric data for each polygon.
Radon potentials of each map polygon were estimated from the radon source and
transport properties of the soil profiles that comprise the region represented by the polygon.
Radon source properties include soil radium concentrations and radon emanation coefficients.
Soil radium concentrations were estimated from NURE aeroradiometric data for shallow
horizons (surface to 2 or 2.5m depth), and from geological classifications of the soils for deep
horizons (to 5m depth). Radon emanation coefficients were based on trends from nearly 400
measurements on county-survey soil samples from most counties throughout the state. Radon
transport properties (water contents, radon diffusion coefficients, and air permeabilities) were
estimated from soil profile physical data compiled for the STATSGO soil maps from Soil
Conservation Service (SCS) data bases by the University of Florida Soil and Water Science
Department. Summary soil data files characterized soil densities, particle size distributions,
water drainage curves, high water table depths and durations, and other mechanical and
hydrological data. Radon transport properties were calculated from these data.
Soil radon potentials are quantified as calculated annual average radon entry rates into
a reference house. The radon potentials were computed by mathematically modeling the
reference house, typical of Florida slab-on-grade single-family housing, as if it were located on
each soil profile of each of the radon map polygons. The parameters characterizing the
reference house were held constant for the calculation on each polygon in order to include only
the varied soil effects on each radon potential calculation. The model calculations were based
on the RAETRAD (RAdon Emanation and TRAnsport into Dwellings) model, but were
conducted using a more specialized, benchmarked radon potential cartography algorithm named
RnMAP. Both models calculated radon potentials as the rate of radon entry into the reference
house. Annual units (mCi y"1) were used for the radon potentials to emphasize the long-term
average nature of the radon potential estimates.
i v
-------
Radon potentials were calculated and then averaged for each of several soil profiles in
each polygon, at each of two or three seasonal water table depths. They also were calculated
for low, intermediate, or elevated-radium geology classes for most of the state, and the
applicable geologic classification was used afterward to select the appropriate values to use in
representing each polygon. Separate radon potentials were calculated for the estimated median
radium concentrations and soil conditions, and also for radium and soil conditions corresponding
to the 75%, 90%, and 95% confidence limits for area distributions within each polygon. The
confidence limits were calculated from the geometric means and geometric standard deviations
of radium and soil transport properties for each map polygon. The radium distributions were
estimated from multiple NURE data points in many of the polygons that intersected NURE flight
lines. Radium estimates for polygons not intersected by NURE flight lines were extrapolated
from the overall data for the geologic unit in which the polygon was located. Distributions of
soil radon transport properties were estimated from the multiple soil profiles defined to comprise
each STATSGO map unit.
The resulting radon potentials were partitioned into seven tiers of similar numerical
values for display on the radon potential maps. The tiers corresponded to the 0-0.4, 0.4-1, 1-2,
2-3, 3-6, 6-12, and > 12 mCi y"1 levels of radon potential. This set of tiers provided suitable
range for using a uniform tier scale on all of the radon potential maps. Map polygons finally
were colored according to the appropriate tier classification for intuitive visual interpretation.
This report presents numerical values of the radon potentials computed for each map polygon
for more quantitative interpretations of the maps. A radon potential of approximately 3 mCi y"1
corresponds to approximately 3.9 pCi L1 of soil-related radon in the reference house.
Separate maps were plotted for the median (50%), 75 %, 90%, and 95 % confidence limits
of radon potentials to give a better perspective of radon potentials in a given polygon (region).
Regions with low potentials on both the median and higher-confidence-limit maps have
reasonable assurance of having minimal indoor radon risk. Regions with high radon potentials
on the median and higher-confidence-limit maps conversely have a relatively high probability
of elevated indoor radon levels. Regions with low median radon potentials but high potentials
for higher confidence limits are heterogeneous (low median; high geometric standard deviation)
and may have generally low radon potentials but occasional to frequent anomalies with high
radon potential. Special considerations may be needed to define radon-protective building needs
in these areas.
Comparisons of calculated radon potentials with 2,930 state-wide land-based indoor radon
measurements were consistent with the reference-house indoor radon accumulation rate of 1.3
pCi L1 per mCi y"1 of soil radon potential, and with an ambient outdoor radon concentration of
approximately 0.1 pCi L1. The geometric standard deviation (GSD) between measured indoor
radon levels and those predicted from the maps was 1.9, which is the approximate level of
precision associated with the calculated soil radon potentials. The total variation among
measured indoor radon levels was partitioned to estimate a house variability of approximately
GSD=3.2, an annual-average measurement uncertainty of approximately GSD=2.1, and soil
variabilities averaging approximately GSD=2. Uncertainties are much higher in predicting an
indoor radon level for a particular house than for predicting the median level in the reference
v
-------
house for a given polygon.
The soil radon potential map data were validated by state-wide comparisons with over a
thousand soil radon flux measurements at 330 locations and with 9,038 indoor radon
measurements from three different data sets. The radon flux measurements averaged similar to
the map predictions, but were scattered more widely than the map data (16 below and 18 above
the central 95% range, compared to eight expected for each). The difference in scatter is caused
by inadequate definition of the temporal variations in radon flux that are needed for comparing
the 24-hour flux measurements to annual-average calculated values.
The Geomet land-based data set best represents all regions of Florida and agrees very
well with the map predictions. The middle 95% of the map range included 95.4% of the 2,952
measurements, with 1.9% below and 2.7% above the mid-range, compared to 2.5% expected
for each. The HRS residential and Geomet population data sets do not represent all regions in
Florida, but they were compared with the map predictions anyway. The 2,095 measurements
in the Geomet population-based set averaged slightly lower than map values, while the 3,938
measurements in the HRS residential data set averaged slightly higher than map values.
Over 250 houses with the greatest differences between measured and predicted indoor
radon concentrations were investigated and found to show trends that offer further explanations.
Houses above the 95% mid-range were nearly three times more likely to use slab-on-grade
construction than to have crawl spaces, while the opposite trend was seen for houses below the
mid-range. Similarly, houses above the 95% mid-range were about 50% more likely to use
hollow-block construction than frame construction, and the opposite trend was also seen for
houses below the mid-range. These trends are consistent with model predictions, and account
for potential anomalies on a state-wide level. Considering the variations in both measurements
and map calculations, the measurements give excellent overall state-wide validation of the radon
maps.
ACKNOWLEDGEMENTS
This project was conducted under review and guidance from the Radon Potential
Cartography Committee of the Florida Radon Research Program. Soil survey data bases,
samples, and STATSGO soil mapping data were provided by Randall B. Brown, Willie G.
Harris, and Ronald J. Kuehl of the University of Florida Soil and Water Science Department,
Gainesville, FL. Aeroradiometric data tapes and interpretations of geologic maps were provided
by James K. Otton of the U. S. Geological Survey, Denver, CO. Geologic maps were provided
by Thomas M. Scott of the Florida Geological Survey, Tallahassee, FL. Geographic partitioning
of aeroradiometric and radon measurement data, intersections of soil and geology maps, and
plotting of all radon maps were performed by Stanley D. Latimer and Paul D. Zwick of the
University of Florida GeoPlan Center, Gainesville, FL.
vi
-------
TABLE OF CONTENTS
Section Page No.
Abstract i v
Acknowledgements v i
List of Figures x
List of Tables xiii
1 INTRODUCTION 1-1
1.1 Background 1-1
1.2 Objectives and Technical Approach 1-3
1.2.1 Partitioning of Variations 1 -4
1.2.2 Definition of Radon Map Polygons 1-6
1.2.3 Modeling of Soil Radon Potentials 1-8
1.3 Scope 1-11
2 RADON ENTRY MODELING 2-1
2.1 RAETRAD Numerical Model 2-1
2.2 Radon Potential Cartography Algorithm 2-5
3 RADON SOURCE AND TRANSPORT PARAMETERS 3-1
3.1 Radon Source Parameters 3-1
3.1.1 NUREData 3-1
3.1.2 Soil Radium and Emanation Data 3-2
3.2 Radon Transport Parameters 3-7
4 CALCULATION OF RADON POTENTIALS FOR MAPS 4-1
5 PRODUCTION AND INTERPRETATION OF THE
RADON MAPS 5-1
5.1 Definition of Radon Map Tiers 5-1
5.2 Interpretation of the Radon Maps 5-9
vi i
-------
TABLE OF CONTENTS
(Continued)
Section
Page No.
6 STATE-WIDE VALIDATION OF THE RADON MAPS
6-1
6.1 State-wide Comparisons with Radon Flux Measurements
6.2 State-wide Validation with Indoor Radon Data Sets
6-1
6-6
6.3 Examination of Indoor Radon Anomalies
6.4 Map Validation Summary
6-13
6-19
7 LITERATURE REFERENCES
7-1
APPENDIX A -- COMBINED GEOLOGY AND STATSGO MAPS
A-l*
APPENDIX B - SAMPLE PRINTOUT FROM RNMAP
B-l
APPENDIX C -- RADIUM DATA FROM NURE MEASUREMENTS C-l
APPENDIX D -- RADIUM AND RADON EMANATION
MEASUREMENT METHODS D-l
APPENDIX E -- QUALITY ASSURANCE DATA FOR RADIUM AND
RADON EMANATION MEASUREMENTS E-l
APPENDIX F - RADIUM AND RADON EMANATION
MEASUREMENTS F-l
APPENDIX G -- COMPONENTS OF STATSGO SOIL MAP UNITS G-l
APPENDIX H -- SOIL PROFILE PROPERTIES FOR STATSGO
COMPONENTS H-l
APPENDIX I ~ RADON POTENTIALS OF INDIVIDUAL SOIL
SERIES FROM RNMAP 1-1
APPENDIX J -- CALCULATED RADON POTENTIAL DISTRIBUTIONS
BY MAP POLYGON J-l
(*) All appendices are in Volume 2.
VI 1 1
-------
TABLE OF CONTENTS
(Continued)
Section Page No.
APPENDIX K - RADON FLUX MEASUREMENT PROTOCOLS,
QUALITY ASSURANCE DATA, RESULTS, AND
BIAS STATISTICS K-l
APPENDIX L - MEASUREMENT-MAP BIAS STATISTICS FOR
GEOMET LAND-BASED RADON MEASUREMENTS L-I
APPENDIX M -- MEASUREMENT-MAP BIAS STATISTICS FOR
GEOMET POPULATION-BASED RADON
MEASUREMENTS M-l
APPENDIX N » MEASUREMENT-MAP BIAS STATISTICS FOR
HRS RESIDENTIAL RADON MEASUREMENTS N-l
APPENDIX O -- SITE DATA FROM INVESTIGATIONS OF
POTENTIAL INDOOR RADON ANOMALIES O-l
APPENDIX P -- THE RELATION BETWEEN EMPIRICAL
MEASUREMENTS AND CALCULATED SOIL
RADON POTENTIALS P-l
ix
-------
LIST OF FIGURES
Figure No. Page No.
1-1 Partitioning of Radon Source and House Variations 1-6
1-2 Representation of Soil Layers for Radon Potential Modeling 1-8
1-3 Diffusive and Advective Radon Entry Into the Reference House
from the Underlying Soil Profile 1-10
1-4 Tasks Performed by Cooperating Institutions for Development of
the Soil Radon Potential Maps 1-10
2-1 Relative Soil Gas Radon Concentrations as a Function of House
Size for Five Soils 2-8
2-2 Fitting of the b Coefficients to House Radius for the Five Soils 2-9
2-3 Estimation of the b0 and b, Coefficients in Terms of the Radon
Diffusion Coefficients of the Five Soils 2-10
2-4 Fitting of the g Coefficients to House Radius for Each of the
Five Soils 2-12
2-5 Estimation of the g0 and g, Coefficients from the Diffusion
Coefficients of the Five Soils 2-13
2-6 Variation of Air Velocities with House Radius for the Five Soils 2-14
2-7 Fitting of the f Coefficient to the Air Permeabilities of the
Five Soils 2-15
2-8 Summary Flowchart for the RnMAP Code 2-16
2-9 Modeling of Water Table Annual Distribution From the SCS
High Water Table Depths and Durations 2-17
3-1 Cumulative Probability Distribution of Measured Soil Radium
Concentrations in 779 Florida Soil Samples 3-5
3-2 Comparison of Radon Emanation Coefficients with Soil Radium
Concentrations for 296 Samples of Florida Soils 3-6
x
-------
LIST OF FIGURES
(Continued)
Figure No. Page No.
3-3 Water Capillary Suction Profiles Under and Beside a House for
Normal (16 cm y"1) and Extreme Infiltration 3-9
3-4 Measured Water Matric Potentials at 46 Locations in Florida 3-10
4-1 Distribution of the Random Sums of the Log-Normally
Distributed Variables Ql5 Q2, and Q3 4-6
5-1 Cumulative Probability Distributions of the Soil Radon
Potentials for All Map Polygons in Florida 5-2
5-2 Florida Median (50% Confidence Limit) Map of Soil Radon
Potentials 5-5
5-3 Florida 95 % Confidence Limit Map of Soil Radon Potentials 5-7
5-4 Comparison of Measured Indoor Radon Levels with Mapped Soil
Radon Potentials 5-10
5-5 Comparison of Tier-Averaged Indoor Radon Levels with Mapped
Soil Radon Potentials 5-13
6-1 Illustration of Radon Flux Sampling Locations 6-3
6-2 Distribution of bias statistics for the radon flux measurements 6-5
6-3 Distribution of Geomet land-based radon measurements by
county 6-7
6-4 Distribution of Geomet population-based radon measurements
by county 6-7
6-5 Distribution of HRS residential radon measurements by county 6-8
6-6 Comparison of radon data coverage of Florida counties by data
set 6-8
6-7 Distribution of bias statistics for the Geomet land-based indoor
radon measurements 6-10
xi
-------
LIST OF FIGURES
(Continued)
Figure No. Page No.
6-8 Distribution of bias statistics for the Geomet population-based
indoor radon measurements 6-11
6-9 Distribution of bias statistics for the HRS residential indoor
radon measurements 6-13
6-10 Comparison of floor (a) and wall (b) construction features in
potentially anomalous houses in the 80 cases investigated from
the Geomet land-based data set 6-18
6-11 Comparison of floor (a) and wall (b) construction features in
potentially anomalous houses in the 251 cases investigated
from the land-based, population-based, and HRS residential
data sets 6-19
xi i
-------
LIST OF TABLES
Table No. Page No.
2-1 Definition of Reference House Parameters for Use in Radon
Entry Calculations 2-4
2-2 Results of RAETRAD Analyses for Varying House Sizes and
Soil Textural Classes 2-7
4-1 Geologic Description and Radium Category of Florida Geologic
Units 4-9
5-1 Radon Potential Polygon Distributions Among Seven Tiers 5-3
5-2 Comparison of Calculated and Measured Radon Statistics 5-12
6-1 Radon Flux Measurements Near the FRRP Test Cells in Bartow 6-4
6-2 Statistical Summary of State-Wide Radon Map Validations 6-15
6-3 Gamma-Ray and Radon Measurements of Concretes and
Aggregates 6-17
x i i i
-------
Section 1
INTRODUCTION
1.1 BACKGROUND
Radon (222Rn) gas from the decay of naturally-occurring radium (226Ra) in soils can enter
indoors through building foundations. If enough radon enters and the building is inadequately
ventilated, radon can accumulate to a level that poses significant risks of lung cancer with
chronic exposure. The degree of health risk is proportional to the long-term average level of
radon exposure. The U.S. Environmental Protection Agency (EPA) attributes 7,000 to 30,000
lung cancer fatalities annually to radon, and recommends remedial action if indoor levels average
4 picocuries per liter (pCi L"1) or higher (EPA92a). Indoor radon levels average about 1.25 pCi
L1 in the United States, and about 1% of all U.S. homes exceed 8 pCi L1 (EPA92b).
The Florida Department of Community Affairs (DCA), under the Florida Radon Research
Program (FRRP), is developing radon-protective building standards to reduce radon-related
health risks (San90, SBC90). If integrated into state-wide building codes, the standards could
add an incremental cost for new construction. To minimize economic impacts and still protect
public health, the radon-protective building standards may be applied regionally where soils have
the potential to cause elevated indoor radon. Although radon data are highly variable, regional
trends support the use of a geographic basis for the radon-protective building standards (Coh86,
Nag87, Pea90, EPA92b).
State-wide mapping of soil radon potentials has been proposed for developing a
systematic basis for regional building standards in Florida (Nie91a). Alternative radon mapping
approaches were evaluated in an FRRP workshop (Nie91b), and the present conceptual
approach was selected to best utilize existing resources and to minimize regional bias. The
approach was tested on a detailed scale in Alachua County (Nie91c), and was revised for more
general, systematic applications to larger areas (Nie94a,b). For regional and state-wide radon
1-1
-------
mapping, the methods were further revised to exclude county boundary influences, to eliminate
polygons smaller than one square mile, and to explicitly include soil variations within map
polygons. Using the revised methods, ten regional maps were developed that covered the entire
state of Florida (Nie94b). The regional maps demonstrated several boundary faults when
combined to form a state map, however, and therefore required revision. This report describes
the state-wide revision of the maps to eliminate the regional boundary influences and to correct
an inconsistency in the aeroradiometric data set. For completeness, this report also documents
the technical basis and methods used to generate the present maps.
Soil radon potentials depend mainly on soil radium concentrations, radon emanation
fractions, moisture, air permeability, diffusivity, and density. Indoor air pressures also affect
radon entry rates, and house ventilation affects the extent of radon accumulation. House
properties such as floor and foundation construction and design also affect the relation between
indoor radon levels and soil radon potential. Although indoor radon levels depend on house
conditions as well as soil properties, the effects of soil properties can be separated for mapping
of soil radon potentials by holding the house parameters constant.
Previous radon maps have generally displayed different tiers of measured indoor radon
levels for geographic units such as county or township areas, ZIP-Code areas, or physiographic
or geologic units. As reviewed in the FRRP radon mapping workshop and feasibility study
(Nie91a,b), other mapping approaches also have included numerical radon indices,
aeroradiometric gamma activity, uranium mineralization zones, and surface outcrops of radium-
mineralized geological formations. Although these approaches show where elevated radon has
been or may be observed, they are generally inadequate for undeveloped or sparsely-populated
areas with limited data from previous radon testing. They also are indirect or imprecise
predictors of indoor radon or of radon-protection needs for new construction. Maps aimed at
optimizing testing programs or locating areas of highest observed indoor radon are already
available for Florida (Nag87).
Effective radon-resistant construction methods have been developed (EPA86, Osb88,
Mur90, Bre90, Cla91). Current use of these methods is mostly voluntary or liability-oriented,
1-2
-------
however, because most building regulations have not addressed radon protection. Recent
initiatives to implement radon-protective building codes have raised important policy issues
(Nue90), including altered property values, zoning and enforcement boundaries, and the
confidence with which radon levels can be predicted or generalized. Previous county-scale
radon classifications in Florida (Nag87) have been challenged for failure to correlate with health
effects (Von90). Prevalent theoretical and empirical studies support the predictability,
avoidability, and health effects of indoor radon (EPA92a), and suggest benefits from institutional
controls to limit human exposures. Radon-protective building technology has been demonstrated
for new construction (Nit89) and for remedial action (Fin89, Sco88, Sco89). Active mitigation
systems generally are most effective, but they are more costly and require more maintenance
than passive systems installed during initial construction. Regulations have been successfully
implemented in some places to require radon-resistant features in new construction (Swe86,
Swe90).
1.2 OBJECTIVES AND TECHNICAL APPROACH
The objectives of mapping soil radon potentials in Florida are to provide a sound
scientific basis for implementing radon-protective building standards where needed, and to avoid
the cost of unnecessarily implementing the standards where they are not needed. The measure
of soil radon potential is defined as a calculated annual-average rate of radon entry from soils
into a reference house.
As identified previously (Nie91b,c), several institutional and scientific criteria were
considered in defining the technical approach. First, the maps must identify as precisely as
possible the regions that need radon-protective building features for reduced indoor radon
concentrations. The maps should also avoid political and institutional boundaries (city, county,
etc.) that are not radon-related. The maps should not be restrictively tied to a preconceived
radon standard (i.e., 4 pCi L"1), and they should minimize uncertainties from variations in time,
house design, and occupancy.
1-3
-------
The approach to achieve these objectives was developed in the FRRP Workshop on
Radon Potential Mapping and later feasibility studies (Nie91a,b,c) to satisfy institutional and
scientific goals and objectives. This approach separates the soil radon potential from other
factors that influence indoor radon levels to include only the soil properties that affect the radon
source and its availability. The approach consists of:
a. Regional definition of radon map polygons (geographic areas on a radon map)
from existing soil and geologic maps.
b. Definition of the soil profiles associated with each radon map polygon, and
their associated radon generation and transport properties.
c. Calculation of numeric radon potentials for individual soil profiles, and an
area-weighted average to represent each radon map polygon.
d. Grouping map units with similar radon potentials and plotting the radon map
polygons by color-coded radon potential tiers.
This general approach has been demonstrated in the preceding regional radon maps of
Florida (Nie94b). It was also followed in the present revised state-wide mapping of Florida soil
radon potentials.
1.2.1 Partitioning of Variations
The mapping approach used here separates the soil radon potential from other factors that
influence indoor radon. Besides soil effects, indoor levels also are affected by house
characteristics and by time variations in the soil and house properties. The time variations are
not of interest because only long-term averages are important for radon exposures and their
underlying radon source strengths and house radon resistance. The time variations are
eliminated by using long-term average values for all time-variant parameters such as indoor air
pressure, house ventilation, indoor radon levels, soil water contents, etc. Properties such as
radium concentration, density, and porosity are virtually constant. Others, including soil air
permeability and diffusivity, vary with time, but are dominated by moisture changes. Hence
their values may be estimated from representative soil water contents.
1-4
-------
Long-term average soil water contents under structures can be defined from the drained
field capacity of the soil. This water content commonly is related to a prescribed soil capillary
suction or matric potential (Lut79) that is independent of immediate rainfall conditions or surface
drying. It can be predicted if necessary solely from the soil textural classification and porosity
(Nie92a). The effects of varying water-table depths also are important for the shallow water
tables encountered in Florida, and can be included in defining the soil water profiles from their
capillary suction (Nie92a).
House variations are eliminated similarly to time variations by averaging over the house
variables to define an invariant, reference house. Its properties approximate those of Florida
single-family dwellings, and it can be modeled as if constructed in any source (soil profile)
location. Soil radon potentials then are estimated for each map unit as the calculated rate of
radon entry into the reference house at each defined source location, independent of house and
occupant variations.
The partitioning of soil and house variations, suggested at the FRRP Workshop on Radon
Potential Mapping (Nie91b), is illustrated by the two-stage modeling shown in Figure 1-1. As
illustrated, all radon source and transport parameters are combined with the reference house
parameters in the radon entry model to estimate soil radon potential in units of radon entry rate
(radon activity per unit time). The resulting soil radon potential varies geographically only with
soil radon generation and transport properties, and is independent of house and occupant
variations that further broaden distributions of indoor radon concentrations. The soil radon
potentials can be used separately in a radon balance model to estimate indoor radon
concentrations. However, for mapping purposes, the geographic variation of soil radon potential
is best described independent of house variations. Comparisons with indoor radon levels can
be made by statistically comparing the medians and distributions of indoor data with the
distributions of corresponding soil radon potentials. Such comparisons reveal the effects of
house and occupant variability.
1-5
-------
House
Properties
Indoor
Radon
Cone.
Radon
Entry
Model
Soil
Radon
Potential
Soil
Radon
Potential
Radon
Dilution
& Balance
Model
Soil Profiles:
• Radium-226
• Radon Emanation
• Moisture
• Density & Porosity
• Radon Diffusivity
• Air Permeability
Reference House:
• Size & Shape
• Slab & Footing Design
• Slab Leak Distribution
• Slab Diffusivity & Perm.
O.Cor
Figure 1-1. Partitioning of Radon Source and House Variations.
1.2.2 Definition of Radon Map Polygons
The map polygons used to represent geographic areas with different radon potentials were
defined by the digital intersection of State Soil Geographic Data Base (STATSGO) soil maps
with surface geology maps. This produced a soil radon profile map with polygons that were
independent of institutional boundaries. Each polygon resulting from this intersection was
numbered and characterized from its location and its particular combination of soil and geologic
properties. The STATSGO soil maps (SCS91) provide summary digital coverage of soil units
throughout Florida, based on higher-resolution soil survey data that presently are only digitized
for a few counties in Florida. The STATSGO soil maps were chosen for their more complete,
state-wide coverage and because the higher-resolution soil maps did not significantly improve
estimates of soil radon potentials (Nie92b). The digital STATSGO map files provided by the
University of Florida Soil and Water Science Department defined 165 different soil map units
that occurred in multiple geographic areas that comprised several thousand soil map polygons
in the state-wide STATSGO soil map.
1-6
-------
The surface geology maps used in preparing the soil radon profile map were provided
from newly-revised surface geology studies by the Florida Geological Survey. The maps were
digitized and then intersected with the STATSGO soil maps by the University of Florida
GeoPlan Center using a geographic information system with Arc-Info data formats. The
geologic maps defined 46 geologic map units that occurred in multiple geographic areas that
comprised several hundred geologic map polygons in the state-wide surface geology map. The
U.S. Geological Survey analyzed the state-wide surface geology map and further categorized
each geologic map unit into several tiers of radon production potential based on mineralogy,
bore-hole data, and elevated radon occurrences. These classifications increased the total number
of Florida geologic map units from 46 to 60.
The surface geology maps were revised from the versions used initially (Nie94b) in a few
regions where geologic map faults at political (county) boundaries showed different unit
designations across a boundary. In the cases where the faults could not be explained by known
terrain features (river channels, topographic ridges, etc.), localized map sections were analyzed
and revised by the Florida Geological Survey and the U.S. Geological Survey. The revised
maps were then digitized and intersected with the STATSTO soil maps to obtain the present
version of the soil radon profile maps.
The digital intersection of the soil and geology maps initially formed a soil radon profile
map with more than twelve thousand polygons throughout Florida. Many of these were small,
second-order polygons formed by the two different digital approximations of common
boundaries. The second-order polygons were eliminated by merging all polygons with areas less
than one square mile with adjacent polygons, since they represented border uncertainties rather
than significant geographic areas. The resulting soil radon profile map contained 3,919
polygons, which are illustrated in regional sections by the maps in Appendix A.
Detailed soil profile properties for each map unit were defined from the unit's STATSGO
soil designation, which was defined in turn as an area-weighted combination of several
reference-pedon soil profiles. For each soil profile, Soil Conservation Service (SCS) data
defined the detailed properties of the A, E, B, C, and other soil horizons from the top
1-7
-------
surface to a depth of about 2.0-2.5 m. Deeper soils, extending to a 5-m depth (Figure 1-2),
were characterized by extending the physical properties of the lowest SCS-characterized horizon
to 5 m. Soil radium concentrations were defined from aeroradiometric data provided by the
U.S. Geological Survey for the top interval (from the surface to approximately 2-2.5m depth)
and from geological classification for the lower interval (extending to 5m depth), as described
in more detail in Section 3.
Soil Surface
0 m
Radon source defined from NURE
aeroradiometric data
Radori transport defined from STATSGO
data on the A, E, B, and C Soil Horizon
properties (typically -6 layers)
2.5 m depth
Radon source defined from surface
geology classification
Radon transport defined from deepest
STATSGO soil layer (single layer)
5 m depth
Figure 1-2. Representation of Soil Layers for Radon Potential Modeling
1.2.3 Modeling of Soil Radon Potentials
Soil radon potentials are the calculated rates of radon entry into a hypothetical reference
house that is located over soil profiles defined from invariant or long-term averaged parameters.
The potentials are expressed on an annual basis (mCi y"1) instead of the previously-used short-
1-8
-------
term basis (pCi s"1, Naz89, Rog90) to emphasize the long-term time averaging of parameters.
The potentials were also expressed on an annual basis because of the long-term, chronic nature
of potential radon risks. The radon potentials can be converted to approximate indoor radon
concentrations by dividing by the house volume and its ventilation rate, or by using a more
detailed indoor radon balance model. The conversion of radon potential to indoor radon
concentrations includes the broader uncertainties of house and possibly time variables, and is
best done using a probabilistic approach.
Radon potentials were computed using the RnMAP computer code, a radon potential
cartography algorithm that was developed from the radon entry efficiency model (Nie91d) and
sensitivity analyses with the RAETRAD model (Nie94a). The RnMAP code uses detailed soil
profile properties (density, porosity, water drainage properties, water table, radium
concentration, radon emanation coefficient, radon diffusion coefficient, and air permeability)
defined from SCS data and surrogates to compute the radon generation and transport profiles
beneath the house, and the radon entry rates through its foundation. The approach uses the
complete multi-phase radon theory (Rog91a) to include simultaneous radon transport by both
diffusion and advection through both the intact foundation slab and through a modeled perimeter
foundation crack (Figure 1-3). The reference house is defined to represent a rectangular single-
story slab-on-grade house typical of Florida construction.
The state-wide calculations of soil radon potentials addressed in this report are based on
data and parameter definitions developed cooperatively by several institutions working together
in the Florida Radon Research Program. These institutions and their technical contributions are
summarized in Figure 1-4.
1-9
-------
Concrete
Floor
\
t
t
n
Pressure-Driven
Air Flows;
Advectove Radon
I Transport
Radon Gas Diffusion
5 m
Soil
Layers
House Center Line
(symmetry assumes)
RAE-103688
Figure 1-3. Diffusive and Advective Radon Entry Into the Reference House From
the Underlying Soil Profile.
RADON POTENTIAL
MAP
UFGC, RAE, USGS, UFSWS
Surface
Geology
Map
FGS
Model
RAE
Statsgo
Surface
Soil Map
UFSWS
Soil Radon
Profile
Map
UFGC
UFSWS, RAE
Soil
Parameters
RAE, USGS, UFGC, UFSWS
Radiologic
Parameters
(Lab Meas., NURE)
Figure 1-4 Tasks Performed by Cooperating Institutions for Development of the
Soil Radon Potential Maps (UFSWS: Univ. of Florida Soil and Water Science
Department; RAE: Rogers & Associates Engineering Corp.; UFGC: Univ. of
Florida GeoPlan Center; USGS: U. S. Geological Survey; FGS: Florida
Geological Survey).
1-10
-------
1.3 SCOPE
This report presents the general approach, methods, and detailed basis data used to
prepare state-wide soil radon potential maps of Florida. Section 2 presents the basic radon
modeling, theory, and algorithms used to compute the soil radon potentials. Section 3
describes the radiological data and its analysis to define radon source terms. Section 4
describes calculation of the soil radon potentials for mapping, followed by a description of the
production and intended interpretation of the radon maps in Section 5. Appendices B
through I present detailed tables of the soil and radiological data used to characterize radon
source and transport properties and also details of the laboratory measurements of soil
radium and radon emanation coefficients.
1-11
-------
Section 2
RADON ENTRY MODELING
Radon entry potentials were calculated for each soil profile under each water table
condition using a radon potential cartography algorithm called RnMAP. The RnMAP code is
a one-dimensional approximation of the more detailed RAETRAD code (RAdon Emanation and
TRAnsport into Dwellings, Nie92b; Nie94a), which uses two-dimensional (elliptical-cylindrical)
geometry to calculate radon generation and multiphase transport into houses from soils with
varied moisture contents. This section summarizes the theoretical basis for RnMAP in
estimating state-wide soil radon potentials.
2.1 RAETRAD NUMERICAL MODEL
Conceptually, radon gas is emanated from soil mineral grains, and can be transported
through air-filled soil pores and foundation cracks and pores into the indoor environment.
However, a more detailed model is required to represent the phase interactions and transport
mechanisms of radon gas. For example, the emanated radon gas is distributed between the
aqueous and gas phases of the soil pores and, when dry surfaces are encountered, it may also
be adsorbed onto the solid mineral phase. Radon gas moves primarily by diffusion and
advection mechanisms. Diffusion, driven by radon concentration gradients, is significant in the
aqueous as well as the gas phase because of frequent intermittent blockages of soil pore segments
by water. Advection, resulting from pressure-driven flow of soil gas, carries radon at the
interstitial soil gas velocity. Both mechanisms establish new equilibria of radon concentrations
along the transport route with local aqueous and solid phases in a chromatograph-like process.
The complete description of radon generation and transport is characterized by three
coupled differential equations characterizing radon changes with time in the solid, liquid, and
gas phases. With appropriate parameter definitions, these equations can be reduced to a single,
multi-phase differential equation (Rog91a) that expresses radon concentrations in the air phase
2-1
-------
as they commonly are measured. For steady-state calculations as used for soil radon potentials,
this equation is written as:
V.faDV(Cb/fs) - V[(K/M)(Cb/Q VP] - ACb + RpXE = 0 (1)
where V = gradient operator
4 = p(l-S+SkH)
p = soil porosity (dimensionless: cm3 pore space per cm3 bulk space)
5 = soil water saturation fraction (dimensionless)
kH = 222Rn distribution coefficient (water/air) from Henry's Law (dimensionless)
D = diffusion coefficient for 222Rn in soil pores (cm2 s"1)
Cb = fsCa = 222Rn concentration in bulk soil space (pCi cm"3)
Ca = 222Rn concentration in air-filled pore space (pCi cm"3)
fs = p(l-S+SkH)+pka
p = soil bulk density (g cm"3, dry basis)
ka = kao exp(-bS)
kao = dry-surface adsorption coefficient for 222Rn (cm3 g"1)
b = adsorption-moisture correlation constant (g cm 3)
K = bulk soil air permeability (cm2)
fi = dynamic viscosity of air (Pa s)
VP = air pressure gradient (Pa cm"1)
X = 222Rn decay constant (2.1xl06 s1)
R = soil 226Ra concentration (pCi g"1)
E = total 222Rn emanation coefficient (air + water) (dimensionless).
Equation (1) applies to gas-phase advective radon transport and to combined gas-phase
and liquid-phase diffusive radon transport. The combined-phase diffusive transport is
characterized by appropriate moisture- and porosity-dependent values of the pore-average
diffusion coefficient, D (Rog91b). This approach is important to correctly characterize radon
diffusion in unsaturated soil pores that may have small intermittent water blockages, but that still
2-2
-------
may transmit significant radon flux (Nie84, Rog89). Liquid-phase advective radon transport is
not addressed because it typically is negligible. The radon fluxes between different soil layers
and at the soil surface are calculated as
F = -D fa VCa + (K/ju) VP Ca (2)
where F = bulk flux of 222Rn (pCi cm 2 s1).
For RAETRAD calculations related to mapping, the radon adsorption characteristics of the soils
were ignored, since they are negligible under the moisture conditions typical of Florida.
Accordingly, the value for lc, was defined as zero in equation (1).
To compute radon entry into the reference house, RAETRAD uses an elliptical-
cylindrical form of equations (1) and (2) (2-dimensional gradient operators). This approach
provides a computationally-efficient alternative to three-dimensional algorithms. Three-
dimensional algorithms have sometimes been used (Lou87), but give results similar to two-
dimensional algorithms (Rev91). The present analyses also assume horizontal uniformity of the
soil profiles. Because of the independence of soil air pressures from soil radon concentrations,
RAETRAD computes the solution to Equation (1) in two steps. First, it computes the pressure
gradients required for Equation (1) by separately solving the air flow equation, obtained from
the equation of continuity and the equation-of-state for gases under isothermal expansion
(Yua81):
V*{[K//xp(l-S)] VP]} = 0 (3)
Then, using the resulting arrays of air flow velocities in the radial and vertical directions,
(K//x)VP, RAETRAD similarly solves equation (1) by substituting the computed velocities into
equation (1). The boundary conditions for the finite-difference numerical calculations are:
constant air pressure and radon concentration at the top surface of the house floor; constant air
pressure and radon concentration (but different numerical values) at the top surface of the soil
outside the house; and zero air velocity and radon flux at the center of symmetry, at the outer
radial limit of the finite-difference grid, and at the bottom of the finite-difference grid.
2-3
-------
The reference house, represented in Figure 1-3, was defined to have the approximate
characteristics of Florida slab-on-grade single-family dwellings. The reference house consisted
of a 28 x 54 ft (8.6 x 16.5 m) rectangular structure with nominal properties as listed in Table
2-1. Its volume is based on that of a median U.S. family dwelling (Naz88), and is similar to
that of typical Florida houses (Acr90). A nominal 2.4-m (8-ft) ceiling height was used to
estimate its area, which also is similar to other estimates of Florida floor slab areas (Acr90).
The house ventilation rate is about half the median U.S. house ventilation rate (Naz88), based
on measurements in Florida houses (Cum92). The floor crack location is chosen near the slab
perimeter to approximate a slab/footing shrinkage crack that may occur with floating-slab
construction inside concrete-block stem walls. The stem-wall footing penetrates 2 ft (61 cm) into
the natural terrain, and contains an additional 1 ft (30 cm) of above-grade sandy fill soil beneath
the slab. The fill soil has identical radiological properties to those of the surface soil at the site.
The indoor pressure is typical of that resulting from thermal and wind-induced indoor pressures
in U.S. homes (Naz87), and also of the average indoor pressures measured in a group of 70
Florida houses (Cum92) under average conditions. Concrete slab air permeabilities, radon
diffusion coefficients, and other properties were estimated from data measured on Florida floor
slabs (Rog95).
TABLE 2-1. DEFINITION OF REFERENCE HOUSE PARAMETERS FOR USE IN
RADON ENTRY CALCULATIONS
House Area
143 m2
Fill Soil Thickness
30 cm
House Dimensions
8.6 x 16.5 m
Indoor Pressure
-2.4 Pa
House Length/Width
1.9 (ratio)
Concrete Slab Thickness
10 cm
House Volume
350 m3
Concrete Slab Porosity
0.22
House Ventilation Rate
0.25 h1
Concrete Slab 226Ra • Emanation
0.07 pCi g1
Floor Crack Width
0.5 cm
Exterior Footing Depth
61 cm
Floor Crack Location
slab perimeter
Concrete Air Permeability
lxlO"11 cm2
Crack Area Fraction
0.002
Concrete Rn Diffusion Coeff.
8x10~4 cm2 s1
2-4
-------
2.2
RADON POTENTIAL CARTOGRAPHY ALGORITHM
The detailed RAETRAD numerical model has been used for a previous prototype map
of soil radon potentials in Alachua County (Nie91c), but uses a level of detail that is
unnecessary for mapping soil radon potentials. In order to make the cartographic radon
potential calculations computationally efficient and to better match them to their basis data,
a specialized radon potential cartography algorithm was developed and incorporated into the
RnMAP code. The algorithm was designed to utilize only vertical profile descriptions of soils,
since horizontal uniformity was always assumed within the scale of a building site. RnMAP
still incorporates the full two-dimensional geometry and advective-diffusive transport from
the RAETRAD model, however, by using empirical constants fitted to RAETRAD analyses
for the reference house. In this way, RnMAP preserves for different sites the correct
parametric variations of radon potential, but avoids the duplicative calculation of detailed
horizontal profiles around the same reference house.
The RnMAP code uses the multi-phase, one-dimensional RAECOM model (Radon
Attenuation Effectiveness and Cover Optimization with Moisture Effects, Rog84) to compute
the vertical radon profiles beneath the reference house, with boundary definitions to match
the house-size related surface radon depletion near the house foundation. The rate of radon
entry into the reference house then is computed using the radon entry efficiency model
(Nie91d). The radon entry efficiency model explicitly includes advective and diffusive radon
entry through cracks in the floor slab, and diffusive entry through the intact part of the slab.
Using typical values of concrete permeability to air flow, advective entry through the intact
part of the slab is negligible.
Radon entry into the reference house was modeled using the radon entry efficiency
correlation derived previously (Nie91d), which gives the relationship:
Q = 10 KC, pc Dc Ah)/tc + (Cc ps Ds Ac)/tc + A, Cc v] (4)
where Q = indoor radon entry rate (pCi s"1)
10 = unit conversion
2-5
-------
Cs = area-weighted average sub-slab radon concentration (pCi L"1)
pc = concrete effective total porosity
Dc = concrete radon diffusion coefficient (cm2 s"1)
Ah = house floor area (m2)
tc = concrete floor thickness (cm)
Cc = radon concentration at the base of the floor crack (pCi L1)
ps = effective total porosity of the surface soil layer
Ds = radon diffusion coefficient of the surface soil layer (cm2 s"1)
A,, = floor crack area (m2)
v = air velocity through the floor crack (cm s1)
The three terms in equation (4) correspond respectively to radon diffusion through the
intact floor slab, radon diffusion through the perimeter floor crack, and advective movement of
radon through the perimeter floor crack by pressure-driven air flow. The variables pc, Dc, Ah,
tc, and Ac in equation (4) are defined as invariant properties of the reference house. The
variables ps and Ds are defined directly from the top of the soil profile on which the house is
modeled. The remaining variables, Cs, Cc, and v are defined from empirical fits to numerical
RAETRAD analyses.
The derivations of empirical expressions for Cs, Cc and v in equation (4) involved a series
of RAETRAD analyses of reference houses of varying size over different soil types to establish
the house size-dependence of sub-slab radon concentrations, and their interactions with soil
texture. The analyses utilized a 5 pCi g"1 radium concentration in each of five soil textural types
(sand, sandy loam, loam, clay loam, and clay). The water contents of each of the soils were
defined to correspond to matric potentials of -0.3-bar (-30 kPa) (Nie92a). Soil permeabilities
and diffusivities were defined by default correlations (Rog91b). The results of these analyses
are presented in Table 2-2 in terms of the average sub-slab radon concentrations, the deep-soil
radon concentrations, the radon concentrations beneath the floor crack, and the air velocity
through the crack.
2-6
-------
Table 2-2. Results of RAETRAD Analyses for Varying House Sizes
and Soil Textural Classes
Sub-Slab Radon Concentration (Area-Weighted Average, pCi L"1)
House Minor
Radius (ft.)
Sand
Sandy Loam
Loam
Clay Loam
Clay
4
3762
4797
4772
4329
2823
6
3926
4931
4943
4521
2945
8
4031
5000
5029
4616
2990
10
4104
5044
5077
4668
3007
13
4180
5085
5117
4709
3015
16
4236
5112
5137
4725
3017
20
4292
5134
5148
4735
3017
24
4334
5147
5153
4738
3017
30
4397
5158
5154
4739
3016
Deep-Soil
Rn (pCi L1):
5246
7248
9353
10328
12747
Radon Concentration under the Floor Crack (pCi L"1)
House Minor
Radius (ft.)
Sand
Sandy Loam
Loam
Clay Loam
Clay
4
3641
4408
3846
3094
1490
6
3761
4480
3905
3136
1502
8
3826
4509
3932
3155
1507
10
3862
4524
3946
3165
1510
13
3890
4536
3957
3173
1513
16
3905
4542
3963
3177
1514
20
3917
4547
3966
3179
1516
24
3923
4549
3968
3179
1517
30
3930
4552
3968
3179
1518
Air Velocity Through the Floor Crack (cm s'1)
House Minor
Radius (ft.)
Sand
Sandy Loam
Loam
Clay Loam
Clay
4
3.18x10-'
2.23x10'
5.13x10"*
1.56x10'
4.54xlOb
6
3.35xl0'3
2.40x10''
5.77x10"
1.76x10"
4.52xl0"6
8
3.45x10'
2.48x10-'
6.12x10"
1.85x10-"
4.40xl0'6
10
3.46xio-3
2.52xl03
6.27x10-"
1.90x10"
4.28xl0"6
13
3.48xl0"3
2.55xl03
6.40x10"
1.93x10"
4.11xl0'6
16
3.50xl0"3
2.58xl0"3
6.45x10""
1.94x10"
3.98xl0"6
20
3.51xl0'3
2.59xl0"3
6.47x10"
1.92x10""
3.85xl0"6
24
3.52xl03
2.60xl0"3
6.47x10-"
1.90x10"
3.76xlO"fi
30
3.53xl0"3
2.60xl0'3
6.43x10""
1.87x10""
3.67xl0"6
2-7
-------
The RAETRAD analyses were interpreted by analyzing the ratio of the area-weighted
average sub-slab radon concentration to the deep-soil radon concentration, C,/Cd, as a
function of house size and soil type. Figure 2-1 illustrates the regular trends resulting from
these analyses. The curves in Figure 2-1 were fitted to the following exponential function of
the house minor radius to obtain empirical parameters to represent the RAETRAD analyses:
CyCj = 1 - exp[ -r / (b0 + b, r)] (5)
where Cs = area-weigh ted average sub-slab radon concentration (pCi L"1)
Cd = deep-soil radon concentration (pCi L"1)
r = minor radius of the reference house (m)
b0, b, = empirical fitting constants.
° Sand
. Sandy
Loam
o Loam
o Clay
Loam
x Clay
23456789
House Minor Radius (m)
Figure 2-1. Relative Soil Gas Radon Concentrations as a Function of House Size
for Five Soils.
2-8
-------
The curve fitting was accomplished by transforming the data in Table 2-2 as
b = -r/ln(l-C,/Cd)] (6)
where b = b0 + bj r.
The resulting values of b then were fitted as shown in Figure 2-2 to obtain the illustrated
values of the fitting constants b0 and b; for each of the five soil types.
100
_o
b = 0.22192 + 3.6703r RA2 = 1.000
Clay
b = 0.22849 + 1,5973r RA2 = 1.000
Clay Loam
b = 0.17464+ 1.2254r RA2 = 1.000
Loam
b = 0.16594 + 0.78525r RA2 = 1.000
Sandy Loam
b = 0.39754 +0.51749r RA2 = 0.999
Sand
0 1 23456789 10 11 12
House Minor Radius (m)
Figure 2-2. Fitting of the b Coefficients to House Radius for the Five Soils.
2-9
-------
The fitting constants from Figure 2-2 were further fitted to obtain a continuous
expression for all soils that did not depend on soil type. For this fitting, the five soils were
represented by their radon diffusion coefficients, and were found to exhibit a regular
dependence on the square-root of the diffusion coefficients, as illustrated in Figure 2-3. The
resulting cubic equations, shown in Figure 2-3, were then used to represent the b() and b,
constants for any soil based on its radon diffusion coefficient:
b0 = 0.6178 + 8.368 VF - 164 D + 735.7
(7)
b, = [-0.070645 + 10.6 VD + 12.23 D - 72.25 D3/2]"1
2.0
- 2.0645e-2 + 10.606X + 12.230xA2 - 72.257xA3 RA2 = 1.000
1/b1 ^
o
0.5
bo
Q.
y = 0.18829 + 2.5507x - 49.995xA2 + 224.24xA3 RA2 = 0.982
0.0
0.20
0.15
0.10
0.05
U.00
Square-Root of Diffusion Coefficient
Figure 2-3. Estimation of the b0 and b, Coefficients in Terms of the Radon
Diffusion Coefficients of the Five Soils.
2-10
-------
The average sub-slab radon concentration beneath a reference house was expressed
in terms of the sub-slab radon concentration beneath an infinitely-large slab, so that a one-
dimensional radon calculation (no radial dispersion) could be used to represent a finite-sized
house. This utilized the large-house equivalent to equation (5), which neglects the b0
constant,
where C,g = area-weighted average sub-slab radon concentration for a large house (pCi L"').
Combining equation (5) and equation (8) gives
which is the expression used to obtain Cs from a one-dimensional steady-state RAECOM
analysis (Rog84) for the actual soil profile located beneath a concrete slab corresponding to
the intact house floor.
The radon concentration immediately beneath the perimeter floor crack of the
reference house is consistently lower than the area-weighted average sub-slab value (Table
2-2), despite the advective air flow, due to diffusive losses through the crack. The radon
concentration at the base of the crack therefore was parameterized in a manner
corresponding to the sub-slab average values, producing the fitting constants g0 and g,„ as
shown in Figure 2-4, which result from the relation
Clg/Cd = 1 - expf-l/b,]
(8)
CB = C,B U-exp|-r/(b„ + b,r)]) / 11 - exp(-l/b,)|,
(9)
Ct/Cd = 1 - expf-r / (g0 + g,r)]
(10)
where Cc = radon concentration at the base of the floor crack (pCi L"').
2-11
-------
100
g = 0.21014 + 7.8636r RA2 = 1.000
Clay
g = 9.6342e-2 + 2.7048r RA2 = 1.000
^Clay Loam
g = 8.4772e-2 + 1,8000r RA2 = 1.000 Loam
g = 6.2507e-2 + 1,0034r RA2 = 1.000 Sandy Loam
g = 0.14951 + 0.70507r RA2 = 1.000 Sand
1 1 ¦ ' ' * p ' ' ' I 1 ' 1 » I « » ¦ ' » i ' * * *
4 5 6 7 8 9
House Minor Radius (m)
10 11
Figure 2-4. Fitting of the g Coefficients to House Radius for Each of the Five Soils.
Similarly fitting the constants g„ and gj to the soil radon diffusion coefficients gives
the relationships shown in Figure 2-5, from which the fitting constants are defined as
g0 = 1.184 - 22.56 VD + 157.5 D - 305 D
3/2
3/2 "1-1
g, = (0.03249 + 0.7763 \'D + HID - 405 D,w]
(11)
and the radon concentration at the base of the floor crack is computed as
Cc = C, (1 - exp|-r/(g0 + g,r)l! / [1 - exp(-l/g,jj
(12)
2-12
-------
2.0
y = 3.2487e-2 + 0.77627x + 111,18xA2 - 404.73xA3 RA2 = 1.000
1/91
U.
O
o
O)
0.5
y = 0.36090 - 6.8762x + 48.012xA2 - 92.869xA3 RA2 = 0.987
0.0
U.00
0.05
0.15
0.10
0.20
Square-Root of Diffusion Coefficient
Figure 2-5. Estimation of the g0 and gj Coefficients from the Diffusion Coefficients
of the Five Soils.
The soil air velocities passing through the floor crack also were expressed as a function
of the soil layer immediately beneath the crack. The velocities in Table 2-2 were found to
depend mainly on the soil type (Figure 2-6), with very little dependence on the size of the
house, particularly near the r = 4.9 m radius value. They were therefore fitted to the
permeability function:
v = exp(-fK), (13)
where v = soil air velocity through the crack (cm s"1)
f = empirical fitting constant
K = soil air permeability (cm2).
2-13
-------
-2
-3 .
10
c/>
E
-g 10
cc
O
o A
E 10 4
0)
Q_
c
E 10
CD
>
-5
~—~ ~—o-
^ ^ ft a
-O- O o
X X X X
T ' 1 ' ¦ I '
Sand
-Q
-a Sandy
Loam
Loam
-o Clay
Loam
Clay
10
-6
I ¦ ¦ i ¦ I i i ¦ i I
' ¦ ¦ ¦ .... i
i l„
0 1 23456789 10 11
House Minor Radius (m)
Figure 2-6. Variation of Air Velocities with House Radius for the Five Soils.
The least-squares fit, shown in Figure 2-7, indicates the fitting constant f also is a
function of K, so that the quadratic fitting constants can be incorporated into equation (12)
to give:
v = expl -e1152 K1' 09923 + "¦°0284 ln(K)], (14)
Equation (14) therefore is used to calculate the air velocity in equation (4) for the floor crack.
2-14
-------
26
y = 1.1524 - 0.99337X + 2.8439e-3xA2 RA2 = 1.000
24
22
20
8
6 L~
-24
-22
-20
18
16
14
ln(K)
Figure 2-7. Fitting of the /"Coefficient to the Air Permeabilities of the Five Soils.
The RnMAP computer code implements the above algorithms using equations (4), (9),
(12), and (14). The basic logic of RnMAP is illustrated in Figure 2-8. After reading
descriptive soil profile data, described further in Section 3, the RnMAP code starts a loop over
four main cases of radon source strength for the radon entry calculations. These include a
hot soil and cold soil case for each of two (hot and cold) geology configurations. The hot soil
case was defined for the depth range (from surface to about 2-2.5m) defined by the SCS soil
data, and was set equal to 4 pCi g'1 radium and 0.6 emanation. A radium content of
0.1 pCi g'1 and an emanation coefficient of 0.2 was used for the cold soils. The hot geology
representing the remaining soils to the 5 m depth was defined by 4 pCi g'1 radium and 0.6
emanation, and the cold geology by 0.8 pCi g1 radium and 0.32 emanation, consistent with
the previous measurements and data for Florida (Appendix F).
2-15
-------
Read Input Data File
Ra • E Loop: 4 Cases |
Water Table Loop: 3 Cases
Print Summary & Stop 1
^^End\.
Radon Source
""•nJjoop
^End^
Water Table
s^Loop
Print
Intermediate Results
Print
Intermediate Results
Calculate 1-dimensional
Vertical Radon Profile
Under Slab
Calculate 1-dimensional
Vertical Radon Profile
Outside Slab
Calculate Q = a(RaE) + b
Coefficients, Annual Avg.,
for Hot & Cold Geology
Define Water Contents,
Diffusion Coefficients,
Air Permeabilities
RnMap.FC
Figure 2-8. Summary Flowchart for the RnMAP Code.
2-16
-------
A second loop in the RnMAP code (Figure 2-8) includes three different water table
depths that were defined from the University of Florida/SCS estimate of the seasonal high
water table depth and duration. The annual distribution of water table depths was
estimated, as illustrated in Figure 2-9, as being 2 m deeper than the high water table depth
during most of the part of the year not included in the reported high water table period.
During two one-month transition periods between the high and low water table levels, the
depth was approximated as only one meter deeper than the reported high water table level.
For map units in which the water table was reported as >180 cm, the high water table level
was estimated to be at the three meter depth for half of the year and at the five meter depth
for the remainder of the year.
Soil Surface
0 m
High Water
Table Duration
A "
1 m
...y..
-
1
month |
1 1
.month
«-
A
1 m
y
Annual Period
5 m depth
Figure 2-9. Modeling of Water Table Annual Distribution From the SCS High
Water Table Depths and Durations.
2-17
-------
After starting the two loops (Figure 2-8), the water contents were estimated for each
layer of the soil profile from its distance above the water table. Soil water contents throughout
the top five-meter soil profile were estimated directly from the soil water drainage data supplied
by UFSWS. The height of each layer above the water table was set equal to the soil capillary
suction and used to interpolate the SCS drainage curve data to determine the drained soil water
content. Benchmark numerical calculations with the FEMWATER code (Yeh87, Sul88) indicate
that sub-slab soil water matric potentials are well-approximated by the height above the water
table for wet climates and shallow water tables as occur in most of Florida (Nie94b). Radon
diffusion coefficients and permeabilities were defined for each soil horizon from the soil and
water parameters (Rog91b).
Once inside the main loops (Figure 2-8), RnMAP calculates the vertical one-dimensional
radon concentration profile for an infinitely-large (horizontal extent) slab with the RAECOM
code algorithm (Rog84). The resulting sub-slab radon concentration then is used in equations
(9) and (12) to obtain the parameters needed in equation (4). The air velocity for the crack
similarly is estimated from equation (14) for use in equation (4).
The resulting radon entry rate then is printed with the vertical radon concentration
profiles and related input data, and the code proceeds to the next calculation for a different water
table depth. After completing the calculation for each of the intended water table depths, a
summary of the annual-average radon entry values is printed, and the code proceeds to the next
radon source term case. After completing each of the four radon source term cases and their
summary printouts, the RnMAP code computes and prints linear coefficients for the key output
parameters including the radon entry rates (soil radon potentials). The linear coefficients allow
calculation of corresponding results in terms of any other values of the soil radium
concentrations and emanation coefficients, and thus they serve as the basis to efficiently use the
RnMAP results with the wide variety of soil radium concentrations estimated from National
Uranium Resource Evaluation (NURE) aeroradiometric data. This is described in Section 3.
A sample printout from RnMAP for one soil profile is presented in Appendix B.
2-18
-------
Equations (9), (12), and (14), which comprise the input to equation (4) and the basis
for the RnMAP code were benchmarked against RAETRAD analyses using the parameters
defined in Table 2-1 for the reference house. Although RAETRAD cannot directly compute
an infinitely-large house case, it can directly calculate the sub-slab and deep-soil radon
concentrations that form the basis for equations (5) through (8), from which equation (9) is
defined. For the 4.9-m radius reference house, the C,7Cd ratios were calculated by equations
(5) and (7) for sand, sandy loam, loam, clay loam, and clay. The resulting ratios then were
divided by the corresponding ratios computed by RAETRAD, giving bias ratios of 1.006,1.000,
0.999, 0.994, and 0.998 for the five respective soil types. Their average, 0.999 ± 0.004, shows
no net bias in the equations, and the individual ratios show less than 1% bias for any of the
soils.
Similar benchmarks for the radon concentration at the base of the floor crack also
were examined using the same RAETRAD analyses on the same five soils. In this case, the
ratios of radon at the crack to deep-soil radon (C,/Cd) were computed from equations (10) and
(11). The resulting bias ratios from dividing by their RAETRAD counterparts were 0.999,
0.999, 0.994,1.004, and 0.994 for the same five respective soils. Their average, 0.998 ± 0.004,
again shows no significant net bias, and individual ratios show less than 1% bias for any of
the soils.
Benchmarks of soil air velocities through the crack also used the same RAETRAD
analyses. In this case, the velocities from equation (14) were divided by the RAETRAD
velocities for the same five soils, yielding bias ratios of 1.085, 0.884, 0.957, 1.140, and 0.956.
Their average, 1.004 ± 0.105, indicates no significant net bias. However individual ratios
indicate a typical 10% variation due to uncertainties in the empirical parameterization.
The combined benchmark of all of the empirical equations in equation (4), computed
by RnMAP, against radon entry rates computed by RAETRAD also was analyzed for each of
the five soils. The ratios of RnMAP to RAETRAD radon entry rates were 1.13, 1.02, 1.00,
1.02, and 1.08, respectively. Their mean, 1.05 ± 0.05, indicates an approximate 5% positive
bias of RnMAP relative to RAETRAD, with a nominal 5% uncertainty. Although the bias is
small enough to be acceptable, it also is explainable by the finite-difference method used in
2-19
-------
RAETRAD compared to the exact analytical mathematics used in the RAECOM section of the
RnMAP code. If infinitely-small numerical mesh units were used in RAETRAD, it would
yield a slightly higher radon entry rate that approaches that calculated by RnMAP. The
differences are well within the uncertainty of the other parameters used to represent radon
potentials from soil profile data.
2-20
-------
Section 3
RADON SOURCE AND TRANSPORT PARAMETERS
3.1 RADON SOURCE PARAMETERS
Radon source parameters were defined separately for the upper (surface to approx.
2-2.5m depth) and lower (extending to 5 m depth) zones of the soil profile, as illustrated in
Figure 1-2. The radium concentration in the upper zone was defined from National Uranium
Resource Evaluation (NURE) aeroradiometric data, while that in the lower zone was defined
from geologic estimates and measured radium concentrations. Radon emanation coefficients
were based on measured values as correlated to radium concentrations. This section
describes the NURE data and the measured radium and radon emanation data that were
used to estimate the radon source strengths.
3.1.1 NURE Data
Radon source parameters were defined from NURE aeroradiometric data as a
vertically uniform source profile throughout the upper soil region (see Figure 1-2). The
NURE data, measured on flight lines at 6-mile intervals with data recorded every second,
give a data point corresponding to every 200-ft interval beneath each flight line. Details of
the measurement procedures and results are published (EGG81).
Digital tapes of NURE data provided by USGS were used to digitally intersect the
NURE flight lines with the radon map polygons. Figures A-l through A-8 (Appendix A) show
the flight lines superimposed on radon map polygons for different regions of the state. The
partitioning of the NURE data was performed, with a geographic information system, by the
University of Florida GeoPlan Center. The resulting partitioned NURE data were
transferred to Rogers & Associates Engineering Corp. (RAE). RAE then calculated geometric
means, geometric standard deviations, and the number of points associated with each radon
map polygon. The NURE data, expressed in parts-per-million equivalent uranium (eU), were
3-1
-------
converted to an equivalent 22fiRa concentration in pCi g'1 by dividing by the units conversion
factor of 3.0 pCi g'1 per ppm eU. Table C-l (Appendix C) presents a statistical summary of
the resulting data.
Although Table C-l contains data for most of the map polygons, some were not
intersected by a NURE flight line, and therefore the NURE data did not directly represent
them. These polygons were represented by the overall geometric mean and geometric
standard deviation of the NURE values for the geology unit in which the polygon was
classified. If an entire geology unit was not intersected by NURE flight lines, the inter-line
polygons were defined from corresponding geometric means and standard deviations of NURE
data for their STATSGO soil units.
3.1.2 Soil Radium and Emanation Data
Radon emanation coefficients for defining source terms in the upper and lower zones
of the soil profiles (Figure 1-2) utilized empirical trends in emanation with soil radium. The
trends were defined from radon emanation measurements on 301 samples from 12 counties
in north-central Florida and 95 additional samples from the remaining Florida counties. The
emanation-radium trend was used with NURE radium data to define source terms for the
upper soil zone, and with geology-based estimates to define source terms for the lower soil
zone. The geology-based radium estimates for the lower zone were divided into five
classifications. The first was for low radium levels typical of Florida sands that are not
influenced by elevated-radium mineralization. The second was for intermediate radium levels
with mixed or intermittent mineralization by elevated-radium materials. The third was for
elevated radium levels typical of certain Hawthorn and limestone units. The fourth was for
high radium levels typical of the undisturbed parts of the Bone Valley Formation. The fifth
was for high radium levels typical of the disturbed parts of the Bone Valley Formation.
As reviewed previously (Nie91c), there are few other data resources to characterize
radium levels in Florida soils, and virtually no others that adequately characterize their
associated radon emanation coefficients. Therefore the present radon emanation
3-2
-------
measurements were made using soil samples archived from prior county soil surveys. These
samples, which the University of Florida Soil and Water Science Department made available,
were from documented reference pedon sites that corresponded directly to many of the physical-
property data sets being used from the STATSGO soil data base. The samples therefore were
representative of the STATSGO-defined soil profiles used to calculate soil radon potentials.
Improved, more-sensitive methods were developed (Appendix D) and used for the radon
emanation measurements. They were shown (Appendix E) to be comparable to traditional
emanation measurement methods (Aus78, Tha82, Wil91). Radium concentrations also were
measured (Appendix F) in each sample as part of the emanation measurement procedure.
Although the radium results from these measurements were used only indirectly, the emanation
data and trends were used directly to define radon source terms.
The radon emanation measurements included three sets of samples. The first set, totaling
513 samples (3- or 4- digit sample numbers in Table F-l, Appendix F) included all available
samples from Alachua County and many from Marion and St. Johns Counties. Forty additional
samples in the first set were collected by the U.S. Geological Survey (USGS) and Florida
Geological Survey (FGS) (A1 through A8 series in Table F-l) in supplemental field
investigations for this study. The second set included 128 samples from Citrus, Clay, Duval,
Flagler, Lake, Levy, Nassau, Putnam, and Volusia Counties. The third set included 95 samples
from 38 additional counties. Samples in the third set were selected from University of Florida
archives by gamma screening several hundred samples (four from each county, where available),
and analyzing the samples with gamma intensities corresponding to approximately 1 pCi g"1 or
higher. Radium and radon emanation measurements then were made on the selected samples
by the methods described in Appendix C.
The samples were analyzed first for radium concentration. Radon emanation
measurements then were made on all samples with radium concentrations exceeding 1 pCi g"1
plus a number of the lower-radium samples. Most low-radium samples were excluded because
their radon emanation measurements typically have very high uncertainty. A total of 172
3-3
-------
emanation measurements were performed on the first set of samples, and 29 were performed on
the second set of samples. Emanation was measured on all 98 samples from the third set
because they had already been screened to eliminate low-radium samples.
The results of the radium and radon emanation measurements are presented in Appendix
F. The methods used to make the measurements are described in Appendix D. The sample
moistures reported for the UFSWS samples represent water that was added to the air-dry
archived samples to obtain field-representative radon emanation measurements. The radon
emanation coefficients are generally low for dry samples, but exhibit higher values when a small
quantity of water is present, consistent with previous observations (Nie82, Tha82). Sample
moistures for the USGS/FGS samples represent field (as-received) moistures. Soil series
designations were not provided for several of the sample sets.
The measured radium concentrations for all of the samples were log-normally distributed,
with a geometric mean of 0.56 pCi g"1 and a geometric standard deviation (GSD) of 3.5 (Figure
3-1). The measured radium concentrations ranged from <0.3 pCi g \ the typical analytical
detection limit, to 43 pCi g"1. Approximately 73 percent of the measurements exceeded the
analytical detection limit. Related quality control measurements on duplicate samples, standards,
and blanks are presented in Appendix E.
3-4
-------
100
Geometric Mean = 0.56 pCi/g
Geom. Std. Dev. = 3.52
n = 779
r 211 points
censored
<
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Cumulative Probability (standard deviations)
Figure 3-1. Cumulative probability distribution of measured soil radium
concentrations in 779 Florida soil samples.
Radon emanation coefficients measured on the 296 samples (Tables F-l through F-3
in Appendix F) ranged from 0.01 to 0.85, and averaged 0.44 ± 0.18 (mean ± standard
deviation). Related quality control measurements for the emanation coefficients are
presented in Appendix E. The relation between radon emanation and radium concentration
was explored by plotting these parameters as illustrated in Figure 3-2 (open circles). As
illustrated, the lowest emanation coefficients were only observed at low radium
concentrations (generally <2 pCi g'), and the typically high emanation coefficients (0.4-0.7)
were mainly observed at higher radium concentrations (1-30 pCi g"1). The relation between
emanation and radium concentration is consistent with two types of radium mineralization:
primary mineralization that is of low radium concentration, distributed uniformly in the soil
grains, and secondary mineralization that is of higher radium concentration, localized closer
to pore and grain surfaces.
3-5
-------
c
a>
;o
%
o
O
c
o
CO
a
CO
E
LU
c
o
"O
CO
CC
1 .o ' ¦ * * i * * * * i * * 111111 * 1111 * i * 11 * 11 '\\ * * i
£ ° goo O
°°°o«£ °°°°i ° i
otd8 o°"ft ££ *° °—2.—o
rP 8 pCi/g
/
o .
0 1
* 11 ' ¦ ¦ 1' ¦ * * 1 ¦ ¦ 1 ¦ 1 ¦ * * * * * ¦ 1 * ¦ ' *
2 3 4 5 6 10 15 20 25 30 35 40 45
Radium-226 Concentration (pCi/g)
Figure 3-2. Comparison of radon emanation coefficients with soil radium
concentrations for 296 samples of Florida soils.
The emanation trend lines in Figure 3-2 were defined mathematically for use with
NURE radium values and geology-based deep-zone radium estimates as:
E = min(0.55, 0.15 Ra + 0.20), Ra < 8 pCi g
(15)
E = 0.50
Ra > 8 pCi g !
This approximation avoids the most extreme emanation values but still retains the general
measured trends. Based on the five previously-defined geologic classifications, the radium
concentrations representing the deep-zone soils were defined to be 0.8 pCi g'1 for the low-
radium group, 1.8 pCi g"1 for the intermediate group, 4 pCi g"1 for the elevated-radium group,
3-6
-------
8 pCi g"1 for the undisturbed parts of the Bone Valley Formation, and 20 pCi g"1 for the
disturbed parts of the Bone Valley Formation.
3.2 RADON TRANSPORT PARAMETERS
Radon transport parameters were defined from soil profile properties associated with
each of several reference pedon sites that were defined in the STATSGO data base to
represent a particular STATSGO soil map unit. The reference pedon components of each of
the STATSGO soil map units are listed in Appendix G. As indicated by these data, some of
the soil components occur in more than one STATSGO map units. This illustrates the
generalized nature of the STATSGO soil maps, which represent landscape areas by area-
averaged composites of several different soil profiles. Although the individual profiles can
be associated with more specific soil map units in the higher-resolution (1:24,000) individual
county soil surveys (e.g., see Tho85), they are grouped according to commonly-associated soil
types in the digitized STATSGO files used here.
The dominant soil components identified for each map unit were used to characterize
the radon transport properties for radon potential modeling. Data files were assembled by
the UFSWS from SCS data bases to define the physical properties of each horizon in each soil
component. The data in Appendix H present the physical properties of each soil horizon in
each respective reference pedon as they were used in computing radon transport parameters.
In addition to the data in these tables, other information also was furnished, including the
horizon identifications, complete textural data and classifications, and selected hydraulic
properties. The data in Appendix H constitute the complete soil characterization input to the
RnMAP code, which in turn calculated secondary parameters from these data.
Soil porosities computed in the RnMAP code were calculated from the densities given
in column 2 of the Appendix H tables as:
p = 1 - p/pg
(16)
-------
where p
P
Pi
= total soil porosity
= soil bulk dry density (g cm"3)
= soil specific gravity (nominally 2.7 g cm"3)
Soil water contents were estimated in the RnMAP code for each soil component and each
horizon from the capillary drainage data in Appendix H. The capillary suction corresponding
to the water content for sub-slab conditions was initially estimated by a series of 2-dimensional
water balance calculations with the FEMWATER model (Yeh87, Sul88). These analyses are
summarized for a water table in clay soil 5 m below the house in Figure 3-3. The capillary
suction under the (11 m. wide) house is almost identically equal to the height above the water
table (100 cm water column = -0.1 bar, etc) for a constant infiltration of 16 cm y"1, which is
typical for Gainesville, Florida (Tho64). Even outside of the house footprint the water suction
is well-represented by the height above the water table. At lower infiltration rates and for
sandier soils, the agreement is even better. For higher infiltration rates (3 times the typical
Gainesville estimate), the dashed lines show a larger discrepancy outside the house, but excellent
agreement under the house. The worst outside case shown in Figure 3-4 amounts to only a 1 %
higher water content outside the house than under it.
For deeper water tables, capillary suctions of -0.1 bars to -0.33 bars are commonly used
(Tho85, Lut79) instead of the height above the water table. To examine the frequency of
higher-suction matric potentials, a series of shallow-depth (30 cm) matric potential measurements
was conducted throughout much of Florida. Forty-six measurements were made using a quick-
draw portable tensiometer (Model 2900, Soil Moisture Equipment Corp., Santa Barbara, CA).
They were made in multiple locations in or near Tampa, Ruskin, Sarasota, Punta Gorda,
Bartow, Lakeland, Orlando, Ocala, Gainesville, Lake City, Madison, and Tallahassee.
3-8
-------
Constant
Infiltration:
49 envy
16 crrvy
No Runoff
from House
\|m0.43 bar @ 49 envy Infiltr
V-0.4B bar @ 16 any intttr
0.5 bar
500
0.43 bar
E
3, 400
®
XI
ra
H*
IIUaiAUIAIM
0.4 bar
5 300
ra
5
I 200
XI
<
Clay Soil:
Homogeneous. Isotropic
JZ
OJ
'5
X
100
Water Table
Distance from Center of House to Nearest Exterior Wall (m)
Figure 3-3. Water capillary suction profiles under and beside a house for normal
(16 cm y'1) and extreme infiltration.
The results of the capillary suction measurements (Figure 3-4) suggested two
distributions, corresponding approximately to areas with estimated shallow water tables and
deeper water tables. The shallow-depth case had a geometric mean matric potential of -0.06
bars (GSD=1.1), corresponding to a water table at only 60-cm below the measurement depth.
The other group had a geometric mean matric potential of -0.21 bars (GSD=2.8),
corresponding to approximately a 210-cm water table depth beneath the measurement point.
The upper 7 points in Figure 3-4 were censored due to a suction gauge limit above this point.
About 85% of the open-soil measurements were within the gauge range of <75 centibars, and
78% were within 50 centibars. The medians also approximated the -0.1 to -0.33 bar range
3-9
-------
commonly used for sands and clays. An upper limit of -0.5 bars, therefore, was considered
acceptable for mapping purposes. This is the limit afforded by the maximum 5-m water table
depth represented in Figure 2-9 for the RnMAP calculations. For shallower depths of the
water table, the actual depth is used, as suggested by the data in Figure 3-3.
1
-2.5
•
1 ' ' ' ' ¦
i ¦ ¦1 ¦ i' ¦ ¦' i'
¦ ¦ ¦ 1 ¦ 1 ¦ ¦ 1 ¦ 1 ¦ VI 1 1 1 1 1 1 .
Q r. P a 0_ .
•
y m 21 £ • 10A(0.44x) R*2 - 0.92
\
•
MP - 0.21 bars (GSD-2.8)
A# *
• Guage Limit
Tampa
Orlando
Tampa
¦
Ruskin
Wildwood
Madison
Bartow
Ocaia
Tallahassee
•
Lakeland
Gainesville
SW Orlando
Lake City /
-
m
• • •y4
_3 °
y m 5.6 ' 10A(0.054x) RA2 - 0.93
•
MP = 0.056 bars (GSD=1.1)
•
Tampa
Sarasota
Punta Gorda
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Cumulative Probability (standard deviations)
2.5
Figure 3-4. Measured water matric potentials at 46 locations in Florida.
Soil water contents for each horizon were interpolated from the water drainage curves
in columns 5-15 of the Appendix H tables to obtain the water content for the specific height
of the horizon above the water table. Separate values were computed for each seasonal level,
and were used for separate seasonal radon calculations before averaging, as illustrated by
the sample printout from RnMAP in Appendix B. Interpolation of the water content data in
Appendix H was done on a log(suction) vs. moisture basis. Some of the water contents were
supplied in the SCS data file on a weight-percent basis, and were converted to volume-
3-10
-------
percent for consistency with the majority of the other SCS moisture data. The conversion
between water contents on a weight-percent basis, a volume-percent basis, and fraction of
saturation basis utilized the relationship
100S = M/p = pMJp (17)
where S = fraction of water saturation
Mv = soil water content (volume percent)
Mw = soil water content (dry weight percent)
The high water table depths and durations in columns 16 and 17 of Appendix H were
interpreted to characterize an approximate seasonal distribution of water tables as described
in Section 2 (Figure 2-9). Soil horizon thicknesses in column 18 were used directly to define
layers for the radon transport calculations in RnMAP. Soil particle diameters in column 19
were computed from detailed sieve analysis data in the SCS data files. Using standard sieve
size definitions (SCS75), the arithmetic mean particle diameters were computed as required
by RnMAP for definition of soil air permeabilities (Rog91b).
Soil radon diffusion coefficients were estimated from the water contents and porosities
of the soils using a predictive correlation that is based on 1073 laboratory measurements of
radon diffusion in re-compacted soils at moistures ranging from dryness to saturation
(Rog91b). The soil textures ranged from sandy gravels to fine clays, and their densities
covered the range of most of the Florida soil densities. The correlation exhibited a GSD
between measured and calculated values of 2.0, and had the form
D = D0 p exp(-6Sp - 6S14p) (18)
where D = diffusion coefficient for 222Rn in soil pores (cm2 s"1)
D0 = diffusion coefficient for 222Rn in air (0.11 cm2 s"1).
Soil air permeabilities were estimated similarly from the water contents, porosities,
and particle diameters of the soils using a predictive correlation that was based on more than
3-11
-------
a hundred in-situ field measurements of soil air permeability, including measurements in
Florida (Rog91b). This correlation exhibited a GSD between measured and calculated values
of 2.3, and had the form
K = 104 (p/500)2 d,/:i exp(-12S4) (19)
where K = bulk soil air permeability (cm2)
d = arithmetic mean soil particle diameter, excluding >#4 mesh (m).
3-12
-------
Section 4
CALCULATION OF RADON POTENTIALS FOR MAPS
Soil radon potentials were calculated for radon maps in two steps. The RnMAP code
first computed the individual soil radon potentials for each soil component for which data
were available. The RnMAP calculations used the fundamental soil property data from Table
H-l. The RnMAP code internally averaged the resulting radon potentials over the two or
three seasonal conditions to obtain annual average radon potentials for both low-radium and
elevated-radium geology configurations and for two different (low-radium and elevated-
radium) surface soil radium concentrations. For soils associated with the Bone Valley
Formation, additional RnMAP runs were performed for the two higher-radium categories.
In the second step, the radon potentials RnMAP computed were transferred into a
spreadsheet (MicroSoft EXCEL®) for the following three additional calculations: (a)
interpolation of the RnMAP results to correspond to the NURE-based top-zone radium
concentrations in each polygon (defined from Table C-l); (b) area-weighted averaging of the
resulting radon potentials according to the soil profiles used; and (c) selection of the
appropriate geology class for each polygon for defining the lower-zone radon sources.
To simplify the second step, the soil radon potentials computed by RnMAP were
summarized for each annual-averaged soil profile condition by several pairs of fitting
coefficients. These corresponded to the following equation for a single soil profile:
Q = a • (R-E) + b, (20)
where Q = soil radon potential (mCi y"1)
a,b = fitting coefficients computed by RnMAP
R-E = product of radium concentration (pCi g"1) and radon emanation coefficient.
One a,b pair was computed for the low-radium geology case, one pair for the intermediate-
radium geology case, and another pair for the elevated-radium geology case. Fitting
coefficients were computed for the Bone Valley and disturbed Bone Valley cases for applicable
4-1
-------
soils only. Sensitivity analyses with the RnMAP code showed that when the soil radium
profiles were divided into two parts, as in Figure 1-2, the radon potential was a linear
function of the radium "emanation product for the upper part of the profile, as long as the
radium-emanation product in the lower part was held constant. Furthermore, identical a
values were obtained for all geology definitions, and the b constants were directly
proportional to the radium-emanation product for the lower soil zone. Radium
concentrations used for the lower, geology-dominated soil zone in the RnMAP calculations
were defined as described in Section 3.1. All radon emanation coefficients also were defined
as described in section 3.1.
The a and b radon potential coefficients were entered into a spreadsheet on rows
corresponding to their soil series (profile) name, as illustrated in Tables 1-1 and 1-2. The
constants were ordered alphabetically by soil series name to form a look-up table that was
in turn accessed by a second look-up table. In the second table the fractional area of the
STATSGO unit occupied by each soil series (see Table G-l) was used to calculate the area-
weighted average of the a and b coefficients for each STATSGO unit. With weighted
averaging, the radon potentials for the different soil series were defined from equation (20)
for a STATSGO map unit as:
Qa = 0.011 Fj [aj • (R-E) + bt] (21)
where Qa = soil radon potential for the polygon (mCi y"1)
Fj = normalized area % occupied by the soil component (%, LFj = 100)
a,, bj = RnMAP fitting coefficients for soil series i for the polygon geology designation
R = soil radium concentration (pCi g"1)
E = radon emanation coefficient defined from equation (15).
The soil component areas from Table G-l were used for area-weighted averaging of
the and bj coefficients from Tables 1-1 and 1-2. However, these areas were first normalized
upward to account for the missing minor soil components (generally those <10% of the
STATSGO map unit) that were not defined in the STATSGO data sets (Table H-l).
4-2
-------
The aj and bj coefficients in equation (21), listed in Tables 1-1 and 1-2, are distributed
relatively narrowly for each STATSGO unit. In contrast to the log-normally distributed
NURE radium data, a normal distribution better represents the a^ and bj coefficients.
Therefore, arithmetic means and standard deviations were computed in the area-weighted
averaging of the coefficients to define a simpler form of equation (21). This simpler form
better separates the soil-related coefficients from the radium and emanation parameters. The
equation reads:
Qa = A-(R-E) + B (22)
where A = (XFj aj) / LFj = area-weighted mean of the aj's
B = (ZFj bj) / £Fj = area-weighted mean of the bj's.
The values of bj appropriate to the polygon geology were used in equation (22). Values of aj
were independent of the different geology categories.
The area-weighted averages of the soil coefficients for each STATSGO soil unit finally
were combined with the NURE radium and emanation estimates (see Table C-l) to calculate
the soil radon potential for each map polygon. Combining equation (15) with equation (22)
gave the following equation for the median radon potential for a map polygon:
Qm - Qi + Q2 + Q3 (23)
where Qm = median soil radon potential for a map polygon (mCi y"1)
Qx = 0.15 A R2 R < 2.33 pCi g1
= 0 R > 2.33 pCi g"1
Q2 = 0.20 A R R < 2.33 pCi g1
= 0.55 A R 2.33 < R < 8 pCi g'1
= 0.50 A R R > 8 pCi g'1
Qa =B.
Before calculating and adding the three terms contributing to the radon potential in
equation (23), the normal and log-normal distributions were first made consistent. Since the
radium concentrations (R) typically had very wide log-normal distributions, their variations
4-3
-------
usually dominated the narrower normal distributions of the soil parameters A and B.
Therefore the means and standard deviations of the A and B parameters were used to
compute approximately equivalent log-normal parameters for A and B. Qm was then
calculated using consistent, log-normal distributions for all parameters.
The geometric standard deviations used to approximate the arithmetic standard
deviations for the soil parameters were estimated by linearizing standard expressions (Has75)
as:
gA = 1 + sa/A (24)
gn = 1 + sn/B
where gA = geometric standard deviation of the soil parameters a,
sA = arithmetic standard deviation of af (sA = VZF; a-2 - (ZF; )
gB = geometric standard deviation of the soil parameters b;
sB = arithmetic standard deviation of bj (sn = "V XF,- bf2 - (LFf b,)2 ).
The corresponding geometric means used to approximate the arithmetic means of the soil
parameters were estimated as (Has75):
Ga = A exp(-0.5 [ln(gA)]2) (25)
Gb = B exp|-0.5 |ln(gM)l2)
Soil radium concentrations were defined directly as the geometric means from Table
C-l for estimating the median, or 50% confidence limit of the radon potential for all polygons
having a NURE-defined value. Estimates for polygons without NURE data were defined
using the overall geometric mean for the geology unit associated with the polygon. In the few
cases where the radium for an entire geology class was not defined with NURE data, the
overall geometric mean for the polygon's STATSGO unit was used. Thus, the polygons with
large, identical numbers of points in Table C-l are not necessarily better-defined, but may
be extrapolations to polygons that were not directly intersected by NURE flight lines. The
geometric means and geometric standard deviations of the NURE radium data points were
calculated as:
4-4
-------
G1{ = exp[£ln(i\)/ n]
(26)
and
gR = exp[V{X[ln(r,)]2 - [Iln(r,)J 2/n]) / (n-1) ]
(27)
where GK = geometric mean of the NURE-estimated radium concentrations (pCi g"1)
rf = NURE estimate of radium concentration at a single point (pCi g"1)
n = number of NURE points in a map polygon
gR = geometric standard deviation of the NURE-estimated radium concentrations.
The Qj and Q2 terms in equation (23) were computed by using geometric means (GA
and Gr) directly for the indicated A and R parameters. The resulting Q, and Q2 terms were
therefore medians (geometric means) of log-normal distributions, with associated geometric
standard deviations that were calculated as:
where gQ1, gQ2 = respective geometric standard deviations of Q, and Q2.
The number of degrees of freedom associated with the Q, and Q2 distributions was defined
as the minimum of the degrees of freedom for either their radium or soil parameter
components. Since the soil components were computed as area-weighted averages, their
degrees of freedom were considered large, and the number of degrees of freedom was defined
from the number of NURE points in the radium distribution.
The addition of the Q,, Q2, and Q> components of the soil radon potential posed a more
difficult problem, since the sum of two or more independent, log-normally distributed
variables does not necessarily yield either a normal or log-normal distribution. Although the
central-limit theorem suggests that the distribution of the sums should approach normality
(Dev87), other studies suggest that the sums may be nearly log-normal (Mit68). Tests with
map data indicated that neither distribution adequately describes the median or distribution
of the sums for the necessary range of cases. Figure 4-1 illustrates an example of the three
gQ1 = exp{ V[ln(gA)]2 + 2|ln(gu)]2 }
gq2 = exp{ V[ln(gA)]2 + [ln(gR)J2 }
(28)
(29)
4-5
-------
components of Q for a map polygon calculated previously (Nie93a). As illustrated, the sum
of the medians of the three distributions (0.325) is only about two-thirds of the median of the
distribution of random sums from the Q,, Q2, and Q;) distributions as calculated by Monte-
Carlo methods. For comparison, the sum of the arithmetic means of the three distributions
is 1.5, more than three times higher than the expected median. Although approximate
methods have been proposed for adding independent, log-normally-distributed variables
(Mit68, Bar76), the problem is presently best-addressed by Monte-Carlo calculations (Gil93).
100
Distribution
of Sums
(Error bars = ±1 s.d. variation
in Monte Carlo Calculations)
Q2
.01
Median (GSD)
Q1 = 0.055 (9.5)
Q2 = 0.216 (5.0)
Q3 = 0.054 (1.8)
Sums = 0.486 ± 0.035
o Sum of Arithmetic Means
• Sum of Geometric Means
: Q1
.001
' Q3
.0001
3
2
1
3
0
2
1
Cumulative Probability (standard deviations)
Figure 4-1. Distribution of the random sums of the log-normally distributed
variables Q„ Q2 and Q.,.
For adding the Q,, Q2, and Q:i variables in equation (23), the distribution of the sum
was estimated in the following manner. One hundred points of equal probability were
computed to represent each of the three distributions, as illustrated by the straight lines in
Figure 4-1. The points in each distribution were then randomly shuffled and successive
4-6
-------
single points from each were added together to obtain a 100-point random distribution of the
sums, which is also illustrated in Figure 4-1. This process was repeated nine times for each
sum to reduce the uncertainty in the mean of the sum by a factor of three. The individual
distributions of the sums had a relatively uniform coefficient of variation throughout the
required range of confidence limits (50% to 95%) of about 7.6%, which was reduced to about
2.5% by averaging the nine replicate distributions.
The composite distribution of Q was finally used to estimate soil radon potentials at
various confidence limits. The value for the median, or 50% confidence limit, was defined for
the composite distribution (see Figure 4-1) as the value at zero standard deviations from the
center. The values for 75%, 90%, and 95% confidence limits were defined from the same
composite distribution at positive probability levels defined from student's t-statistic (Nat66)
at the desired confidence level and degrees of freedom. The number of degrees of freedom
used to calculate the t-position in the distribution of Q was estimated from the degrees of
freedom of the Q,, Q2 and Q3 components as:
u = fKf/n,.) J'1 (30)
where u = degrees of freedom for the composite estimate of variability
f; = Qi / XQ,
n; = number of degrees of freedom for component i.
The positions in the Monte Carlo distribution of sums that corresponded to the chosen
"t" values were interpolated between the calculated distribution points to obtain the desired
confidence limits for the radon potentials. For example, the polygon used to create Figure
4-1 was estimated from equation (30) to have u=2021 degrees of freedom, which corresponds
to a t-statistic of 1.282 for a 90% confidence limit. The soil radon potential was therefore
interpolated between points calculated at +1.254 and +1.311 standard deviations above the
median to obtain the 90% confidence limit for the radon potential map.
Soil radon potentials were estimated for the 50% (median), 75%, 90%, and 95%
confidence limits for mapping purposes. The "t" statistics used in the calculations were
4-7
-------
obtained from internal spreadsheet functions found to correspond to standard tables of values
(Nat66). The Monte Carlo calculations also were performed by spreadsheet calculations using
a macro instruction set and the spreadsheet random number generator. It should be noted
that the geometric standard deviations used in equations (24), (27), (28), and (29) reflect
variations among individual NURE observations and individual soil series within a map
polygon. The calculations therefore give confidence limits for individual localized areas in
order to reach the stated radon potentials. Since the spatial resolution of the NURE
measurements is on the order of 0.25 square miles, and NURE variations dominate the total
radon potential variations, the calculated confidence limits potentially apply to maxima over
localized areas as small as 0.25 square miles (0.65 km2). Soil and geology units are much
larger, however, and the one square mile (2.6 km2) minimum polygon size is estimated to
better represent the nominal map resolution.
The calculated radon potentials are presented in columns 7-10 of Table J-l. The
definitions of the geology classes for selecting the appropriate b coefficients for each polygon
are presented in Table 4-1. The geology definitions are based on geologic descriptions and
recommendations from the USGS, and are also summarized in Table 4-1. The resulting
calculated estimates of soil radon potentials were sorted by polygon number and submitted
to the University of Florida GeoPlan Center for plotting on maps.
4-8
-------
Table 4-1
Geologic description and radium category of Florida Geologic Units
Class
Description
Category
Qal
Anastasia Formation
Low
Qall
Alluvium deposits
Low
Qbdl
Beach ridge dune sands
Low
Qbdal
Beach ridge dune sands over Anastasia formation
Low
Qdl
Dune sands
Low
Qhl
Quartz sands and lagoonal deposits
Low
Qkl
Key Largo limestone
Low
Qml
Miami limestone
Low
Qm2
Miami limestone
Intermediate
Qql
Undifferentiated Pleistocene and Holocene coastal deposits
Low
Qrl
Fluvial, lacustrine sands, clay, marl, and peat
Low
Qsul
Undifferentiated shell beds
Low
Qsu2
Undifferentiated shell beds
Intermediate
Qtdl
Dune-like quartz sands
Low
Qtmal
Transitional unit between Qa and Qm
Low
Qtrl
Trail Ridge quartz sands
Low
Qtul
Undifferentiated older quartz sands
Low
Qtu2
Undifferentiated older quartz sands
Intermediate
Qtu3
Undifferentiated older quartz sands
Elevated
Qtuk2
Undifferentiated sands on karstic limestone
Intermediate
Qui
Undifferentiated quartz sand
Low
Qu2
Undifferentiated quartz sand
Intermediate
Quel
Undifferentiated quartz sand from cypresshead
Low
Qull
Lagoonal sands, clay, and shell
Low
Tabl
Alum bluff formation
Low
Tapl
Avon Park formation
Low
Tap2
Avon Park formation
Intermediate
Tel
Cypresshead formation
Low
Tc3
Cypresshead formation
Elevated
Tchatl
Chattahoochee formation
Low
Tcil
Citronelle
Low
Thl
Hawthorn group phosphatic sediments
Low
Th2
Hawthorn group phosphatic sediments
Intermediate
Th3
Hawthorn group phosphatic sediments
Elevated
Tha2
Tampa member, Arcadia formation
Intermediate
That2
Tampa member, Arcadia formation (limestone/dolostone)
Intermediate
Thccl
Charleton Coosawhatchie formation (Hawthorn group)
Low
Thpb4
Bone Valley formation (undisturbed)
High
Thpb5
Bone Valley formation (disturbed)
High
Thpr2
Peace River formation
Intermediate
Thsl
Hawthorn group phosphatic sediments
Low
Ths2
Hawthorn group phosphatic sediments
Intermediate
Thtl
Tampa member, Arcadia formation (phosphatic sediments)
Low
(Continued)
4-9
-------
Table 4-1 (continued)
Geologic description and radium category of Florida Geologic Units
Class
Description
Category
Ticl
Intercoastal formation
Low
Tjbl
Jackson bluff formation
Low
Tko2
Extremely karstified Ocala limestone
Intermediate
Tmcl
Miccosukee formation
Low
Tmc2
Miccosukee formation
Intermediate
To2
Ocala group (Crystal River limestone)
Intermediate
Trel
Residuum on Eocene limestones and siliclastics
Low
Trml
Residuum on Miocene limestones and siliclastics
Low
Trol
Residuum on Oligocene-Miocene limestones and siliclastics
Low
Tsl
Suwannee limestone
Low
Ts2
Suwannee limestone
Intermediate
Tsk2
Karstified Suwannee limestone
Intermediate
Tsml
Undifferentiated Suwannee and Marianna limestone
Low
Tsm2
Undifferentiated Suwannee and Marianna limestone
Intermediate
Tt2
Tamiami formation
Intermediate
Twhl
Weathered Hawthorn group
Low
Twh3
Weathered Hawthorn group
Elevated
4-10
-------
Section 5
PRODUCTION AND INTERPRETATION OF THE RADON MAPS
Soil radon potential maps were produced by displaying each radon map polygon in a
color corresponding to its calculated radon potential. Since the radon potentials for each
polygon were calculated as a statistical distribution, it was possible to prepare separate maps
to show different confidence limits of the radon potentials. Data were therefore prepared for
mapping the median, 75%, 90%, and 95% confidence limits of the calculated soil radon
potentials. To display the numerical radon potentials by corresponding colors, the numerical
data were grouped into several tiers of similar values, and different colors were assigned to
each tier. Although this approach sacrifices the numerical detail associated with each
polygon, it helps illustrate regional trends and anomalies. The numerical details are still
preserved, however, by using Table J-l to examine any particular polygons of interest.
The calculated soil radon potential distributions were compared with indoor radon
measurements to help clarify the general relationship between the calculated radon
potentials and observed indoor radon levels. Although a theoretical, linear relationship is
predicted from simulations of radon entry into the map reference house, large variations
caused by construction and occupancy differences and temporal variations obscure the
relationship. Calculated radon potentials were compared with the Geomet land-based indoor
radon data (Nag87) to help identify the magnitudes of the variations from calculated values
and to attempt to partition them among soil-related and house/occupant-related sources.
5.1 DEFINITION OF RADON MAP TIERS
The boundaries between different tiers of soil radon potential were defined from
analyses of the distributions of all of the calculated radon potentials. Figure 5-1 summarizes
these distributions using cumulative probability plots. The higher confidence limits (i.e.,
90%, 95%) have consistently higher values than the lower ones, as expected, and tend to
exhibit a slightly broader distribution than the values for the 50% confidence limit. As
5-1
-------
illustrated, there are no significant breaks in the distributions that suggest consistent natural
cut points. Therefore an arbitrary but consistent set of tier cut points was defined that would
display the soil radon potentials in several different color categories.
1000
3,919 State-Wide Map Polygons
(121 polygons were water-dominated)
100
o
E
c
-------
polygons that fall into each of the seven tiers at the 50%, 75%, 90%, and 95% confidence
limits. Although these tier definitions provide little separation among most of the lower
radon potentials, they accommodate the higher radon potentials observed in several areas of
the state.
Table 5-1. Radon Potential Polygon Distributions Among Seven Tiers.
Radon
50% Confidence Limit
75% Confidence Limit
Potential
(mCi y"1)
Tier
No. of
Polygons
Percent of
Polygons
No. of
Polygons
Percent of
Polygons
1
0.0 - 0.4
2834
72.3
2036
52.0
2
0.4 - 1
945
24.1
1521
38.8
3
1-2
95
2.4
278
7.1
4
2 - 3
19
0.5
37
0.9
5
3-6
22
0.6
37
0.9
6
6-12
3
0.1
4
0.1
7
>12
1
0.0
6
0.2
Radon
90% Confidence Limit
95% Confidence Limit
Potential
(mCi y1)
Tier
No. of
Polygons
Percent of
Polygons
No. of
Polygons
Percent of
Polygons
1
o
i
o
o
1225
31.3
904
23.1
2
0.4 - 1
1780
45.4
1291
32.9
3
1-2
737
18.8
1435
36.6
4
2 - 3
74
1.9
127
3.2
5
3 - 6
73
1.9
110
2.8
6
6-12
21
0.5
34
0.9
7
>12
9
0.2
18
0.5
It should be recognized that the radon potential maps show the potential soil radon
contribution to the reference house (annual average basis) as if it were located on the soil
profiles in different parts of the state. Indoor radon concentrations also depend strongly on
house and occupant characteristics and their changes in time. Important house variables
include the ventilation (dilution) of radon in indoor air, the ambient indoor air pressure, and
5-3
-------
the house foundation design and construction and its susceptibility to radon penetration.
Models that represent these parameters and their variations are complex. However, a simple
approximate equation estimates the nominal indoor radon concentration that could occur in
the reference house over soil profiles with a given radon potential. The approximate indoor
radon concentration in the reference house is:
Ch = 114 Q / (Vh Xh) + Cout, (31)
where Ch = indoor radon concentration (pCi L"1)
114 = unit conversion (pCi L"1 h"1 per mCi m"3 y*1)
Q = potential radon entry rate (mCi y"1)
Vh = house volume (m3)
Xh = house ventilation rate (h'1)
Cout = outdoor radon concentration (pCi L"1).
This relation suggests that soil with a radon potential of about 3 mCi y'1 would cause
an indoor radon concentration of 3.9 pCi L"1 in the reference house if no other radon sources
were present (using the house volume and ventilation rate in Table 2-1). This gives a ratio
for the reference house of 1.3 pCi L"1 of indoor radon, delivered from the slab and soil, for
every mCi y"1 of calculated soil radon potential. Since ambient levels of radon in outdoor air
generally add an additional 0.1-0.4 pCi L"1 (Ner88) to the levels caused by foundation soils,
the 3 mCi y"1 cut point approximates the 4 pCi L"1 concentration for total indoor radon.
Figures 5-2 and 5-3 illustrate two of the soil radon potential maps for the median (50%
confidence limit) case and the 95% confidence limit case, respectively. Although the 50%
confidence limit map in Figure 5-2 suggests that only a few of the polygon areas have
sufficient radon potential to exceed 4 pCi L"1 in the reference house, soil and house
variabilities cause the reference house to approach and exceed 4 pCi L*1 in many additional
areas at the 95% confidence limit. For this reason, the two maps together give a more
complete picture. The median map in Figure 5-2 shows the commonly-expected levels while
the 95% map shows the levels at the upper end of the range that includes all but the top 5%
of the land areas. As illustrated by Figure 5-3, radon potentials in a much larger part of the
5-4
-------
Q50
Radon Potential
¦ mCi/year < 0.4
¦ mCi/year > = 0.4 and < 1.0
mCi/year >= 1.0 and < 2.0
91 mCi/year > = 2.0 and <3.0
¦ mCi/year > = 3.0 and <6.0
m mCi/year > = 6.0 and < 12.0
¦ mCi/year >= 12.0
¦ Water
Figure 5*2. Florida median (50c/c confidence limit) map of soil radon potentials.
-------
J£
tgya;
k'iS(t
Q95
Radon Potential
mCi/year < 0.4
mCi/year > - 0.4 and < 1.0
mCi/year > = 1.0 and < 2.0
mCi/year > = 2.0 and < 3.0
mCi/year > = 3.0 and < 6.0
mCi/year > = 6.0 and <12.0
mCi/year > = 12.0
Water
&
Figure 5-3. Florida 95c1c confidence limit map of soil radon potentials.
Preceding page blank
-------
state would cause the reference house to have levels that approach, and in some areas exceed, 4 pCi L"1
in a small (approximately 5%) part of polygon land areas.
The interpretation of the soil radon potential maps is conceptually simplified by assuming a
uniform density of residential housing (corresponding to the reference house) in the mapped areas. The
50% confidence limit map then shows that approximately half of the houses in a red area would exceed
4 pCi L"1, and that approximately half in any other color tier would similarly exceed the corresponding
level calculated by equation (31) from the mapped radon potential. The 95% confidence limit map shows
more conservatively the regions in which approximately 5% of the houses would exceed the
corresponding limits. For new construction (of a reference house), any red and pink color tiers on the
two maps show the respective areas in which a 50% and 5% chance of exceeding 4 pCi L1 may be
expected. As implied, the first four color tiers show areas corresponding to lower radon potentials. The
75% confidence limit and 90% confidence limit maps (not shown here) show the corresponding respective
areas in which 25% and 10% of new (reference) houses may exceed corresponding levels under the same
assumptions.
5.2 INTERPRETATION OF THE RADON MAPS
Interpretation of the soil radon potential maps requires an understanding of their various sources
of uncertainty. The large total variations between mapped soil radon potentials and short-term
measurements are illustrated by the scatter plot in Figure 5-4. This plot compares indoor radon
concentrations from the Geomet state-wide land-based radon survey (Nag87) with corresponding soil
radon potentials, and shows the uncertainties associated with the simple linear relationship in equation
(31). The map coordinates for each indoor radon measurement were used in the University of Florida's
geographic information system (GIS) to find the corresponding map polygons, from which the soil radon
potentials were identified. The comparisons utilized 2,930 of the 2,952 measurements in the land-based
data set. The remaining 22 were excluded because they occurred in polygons dominated by water (lakes),
precluding quantitative estimates of their radon potential. The logarithm of
5-9
Preceding page blank
-------
the indoor radon data were fitted by least-squares to the radon potentials to estimate the
slope and intercept for the following equation, which corresponds to equation (31).
Ch = xQ + Cout, (32)
where x = fitted slope of the Ch vs. Q data (pCi L"1 per mCi y 1).
100F
10
.1
T
Points are from Geomet
Land-Based data for the
State of Florida
(n = 2.930)
' oy
O O ,
°°~ 8> So
%y /»
3O03GDO O O
95% C.L.
90% C.L
75% C.L
50% C.L
(Median)
-u-L
' 1 ' ¦ ' * 1
¦ ' 1 ' ¦ ¦ 1
.01
.1 1 10
Soil Radon Potential (mCi/y)
100
Figure 5-4. Comparison of measured indoor radon levels with mapped soil radon
potentials.
The least-squares fit to the median radon potentials yielded the numerical values
*=0.7 pCi L"1 per mCi y"1 and Cout = 0.26 pCi L"1, both of which are reasonable values. The
value x = 0.7 is about 47% lower than the value of 1.3 pCi L 1 per mCi y 1 calculated for the
5-10
-------
reference house in Section 5-1. Inaccuracy in the fitted values is caused in part by the
predominantly low radon potentials, which lower
the soil-related indoor radon to approximately the outdoor levels in many cases. The apparent lower,
narrower distributions of soil radon potentials give a relatively insensitive measure of the value of x. The
value of (1^=0.26 pCi L1 is also within the natural background range, reported from 0.1 to 0.4 pCi L"1,
with large variability (Ner88).
The GSD between the pairs of measured radon levels and the values calculated by equation (32)
from the mapped soil radon potentials was 1.93. This is estimated to be the approximate overall precision
of the mapped soil radon potentials. The total variations among the measured indoor radon levels were
partitioned between geographic soil variations and combined house/occupant and measurement variations
by modifying equation (32) to obtain:
ch = + X50 Qso {exp[\7 (In g/ + (In gj' ] }w, (33)
where X50 = median ratio of indoor radon to soil radon potential (pCi L"1 per mCi y"1)
Q50 = median soil radon potential (mCi y1)
gs = geometric standard deviation from soil profile (location) differences
gh = geometric standard deviation from house and measurement variations
t„ = t-statistic.
The total variations of the measured values from the median line in Figure 5-4 were partitioned
by quadratically subtracting the soil variations, as in equation (33), to find the corresponding value of gh
that matched the observed spread in the data. The geometric standard deviations estimated from NURE
and soil variations for each map polygon were estimated from the calculated soil radon potentials
(Appendix F) as gs=exp[ln(Qp/Q50)/tp] to give average estimates of gj = 1.91 for the 75% confidence
limit, gs = 1.98 for the 90% confidence limit, and gs = 2.05 for the 95% confidence limit. A value of
gh = 4.0 was then found to match the total variation, as illustrated in Figure 5-4 for the calculated
confidence limits. Using a previous estimate of gm = 2.08 for the measurement uncertainty in estimating
annual-average radon concentrations from charcoal canister measurements (Roe90), the value of gh can
be further partitioned to estimate a GSD of approximately 3.2 for house/occupant variations alone. A
5-11
-------
corresponding numerical summary in Table 5-2 shows that the data are reasonably consistent in
approximating the calculated confidence intervals.
Table 5-2. Comparison of Calculated and Measured Radon Statistics
Calculated
Observed
Fraction
Percent
% > C.L.
> C.L.
> C.L.
0.000
50%
1419/2930
48.4%
0.674
25%
800/2930
27.3%
1.282
10%
290/2930
9.9%
1.645
5%
131/2930
4.5%
The precision of median mapped radon potentials (GSD = 1.93) is considerably better than the
total variation among indoor radon levels, and also is better than the partitioned estimates of either soil
or house variations. This fact illustrates the advantage of defining soil radon potentials with the present
approach. It also illustrates the higher uncertainties in trying to predict the indoor radon level in any
particular house compared to the map prediction of the median level in the reference house for a given
polygon.
Averaging the measured radon concentrations in Figure 5-4 by their map tier, as defined in
Section 5.1, gives the geometric means and geometric standard deviations shown in Figure 5-5. As
expected, the medians for each tier are close to the calculated line for the 50% confidence limit. These
analyses illustrate the uncertainties in estimating actual indoor radon levels from soil radon potentials.
The mapped soil radon potentials represent the effects of different soils on the reference house.
For the reference house, the annual-average indoor radon concentration in picocuries per liter (pCi L1)
is approximately 1.3 times the soil radon potential (in mCi y1)- Thus, soil-related radon in the reference
house would average approximately 3.9 pCi L"1 in
5-12
-------
an area with a radon potential of 3 mCi y"1. Average radon concentrations in actual houses
may differ from those in the reference house because they also depend on house and occupant
characteristics. Important house properties include the ventilation and dilution of indoor air,
the indoor air pressure, the foundation design and construction, and its susceptibility to
radon penetration. Some of these properties also change with time, along with soil water
distributions, to alter both the radon potential and its equivalent indoor radon concentration.
The temporal variations preclude direct comparisons of indoor radon measurements with soil
radon potentials without an adequate definition of long-term average values for all significant
soil, house, and occupant parameters.
O
Q.
C
o
"O
CO
CC
o
o
¦a
c
100
10
.1
.01
Points are Geometric Means with
Geometnc Standard Deviations ot
the Land-Based Data tor Florida
Map Tier
Ranges
V
Siue
Green Yellow! Or. Red
95% C.L.
90% C.L.
75% C.L.
50% C.L.
(Median)
Rn = 0.26 + 0.69Q
GSD = 1.93
n = 2.930
.01
.11 10
Soil Radon Potential fmCi/v)
100
Figure 5-5. Comparison of tier-averaged indoor radon levels with mapped soil
radon potentials.
Soil radon potentials, calculated from geometric mean radium concentrations,
correspond to the median, or 50% confidence limit for individual locations in a polygon area.
This means that
5-13
-------
approximately half of the land area represented by a given polygon may have lower radon potentials and
half may have higher potentials. Similar interpretations apply to the other maps (i.e., 95% of the land
area in the 95% confidence limit map has expectedly lower radon potentials while 5% has higher
potentials, etc.). The four maps together give a more complete picture of soil radon potentials than a
single map.
The definition of areas that need radon-resistant building features and of areas that do not depends
strongly on the margin of safety desired. Houses in areas with low soil radon potentials on all four maps
are unlikely to have elevated radon levels. Conversely, houses in areas with high radon potentials on all
four maps are very likely to have elevated radon levels. The 50% confidence limit map shows the most
likely radon potential for a particular location. However, if an area has a low radon potential on the 50%
confidence limit map and a high radon potential on the 95% confidence limit map, the area does not have
a uniformly low radon potential. The area will contain both low radon potential and high radon potential
sections. In such cases, the four maps help estimate the chances of a particular location having a radon
potential that is within the illustrated limits. Housing development in these areas should consider the
potential for elevated radon, and possibly increase the level of radon surveillance, incorporate radon-
resistant construction features, or take other appropriate precautions.
Differences between actual houses and the reference house will alter the potential radon entry rate
for a given soil profile. They also will alter its conversion to an indoor radon concentration, as in
equation (31). For example, if house variations are log-normally distributed with a geometric standard
deviation of 3.0, approximately 16 percent of the houses in a map unit with a radon potential of 3 mCi
y"1 would be expected to exceed a radon potential of 9 mCi y1 (corresponding to nearly 12 pCi L"1 in the
reference house). Additionally, approximately 2.3 percent would be expected to exceed 27 mCi y"1
(corresponding to over 35 pCi L1 in the reference house). The same percentages applied to a map unit
with a radon potential of 0.3 mCi y"1 would indicate lower radon potentials and soil-related concentrations
by approximately a factor of ten.
Interpretations of the soil radon potential maps should also recognize other sources of uncertainty.
These sources include the extrapolation of NURE data to undefined map polygons, averaging from limited
sample analyses to represent geologic radium sources, representing radon map polygons primarily by the
STATSGO soil maps, and estimating seasonal water table depths from high water table data. The maps
5-14
-------
have the advantage of being independent of particular sets of indoor radon data, however, and therefore
they avoid house and occupant variables and biased sampling associated with empirical data sets. Because
of the higher uncertainties associated with predicting the upper confidence limits, the median (50%
confidence limit) radon potentials should be regarded as more reliable, with progressively greater
uncertainty associated with the values computed for higher confidence limits.
Boundaries between map units should be considered approximate because of the gradational nature
of many lithologic and geologic contacts, and because of the imprecisions in defining their locations.
Although the STATSGO soils map constitutes the most widely available geographic basis for defining soil
radon potentials, further variations are inherent within map units. The boundaries between colored map
tiers are based on arbitrary cuts at the indicated values of radon potential. The colored zone boundaries
therefore could change simply with different tier grouping of the calculated radon potentials into more,
less, or different groups. Use of the numerical radon potentials associated with each polygon avoids the
tier definition uncertainties.
5-15
-------
Section 6
STATE-WIDE VALIDATION OF THE RADON MAPS
There is no cost-effective protocol to directly measure soil radon potentials. The soil
radon potential maps were therefore validated on a state-wide basis by comparisons with two
surrogates: soil radon flux measurements and indoor radon concentration measurements.
The soil radon flux measurements hold the potential to better validate the calculated soil
radon potentials because they eliminate the variables associated with house construction and
occupancy. However, their advantage is largely offset by large spatial variations and a
sensitivity to soil moisture that increases temporal variations. Indoor radon measurements
are attractive because they better represent annual-average indoor radon concentrations,
which are the parameter of ultimate interest in using the radon maps. However, the indoor
radon data may vary from map-calculated values because of differences in house construction
and occupancy characteristics compared to the map reference house. Because both radon flux
measurements and indoor radon measurements represent only a 1-3 day period, they both
introduce uncertainty when used to represent annual-average conditions. Despite the
differences between the mapped soil radon potentials and the surrogates, the surrogates
provide the best available basis for validating the radon maps.
6.1 STATE-WIDE COMPARISONS WITH RADON FLUX MEASUREMENTS
The RnMAP calculations of soil radon potential included a corresponding calculation
of bare-soil radon flux (as shown in Appendix B). For initial validation comparisons, the
radon fluxes calculated by RnMAP were compared with radon flux measurements made
throughout the state. The radon flux measurements were distributed to include at least
several locations in each Florida county. After the initial feasibility measurements in
Alachua County, flux measurement sites were selected from criteria that included
accessibility, gamma-ray intensity, and calculated radon potentials. Where possible, locations
were selected from proximate high-radon-potential and low-radon-potential polygons. A
survey of five gamma-ray measurements were then made at 1-m elevations in different
6-1
-------
regions of the polygon using a 5 cm x 5 cm scintillation detector (Model 44-10, Ludlum
Measurements, Inc., Sweetwater, TX). The location with the highest gamma measurement
was used for the radon flux measurements.
Triplicate measurements were made at each location at 10-meter intervals to
characterize local site variability, and also to provide backup in the event that any individual
samplers were disturbed. The gamma-ray activity and soil moisture content also were
measured at the flux measurement sites. Site locations (longitude and latitude) were
determined at the time of sampling with a global positioning system (NAV-5000D, Magellan
Systems Corp., San Dimas, CA). The locations were later analyzed by the University of
Florida GeoPlan Center to confirm which map polygon contained the sampling site.
The protocol for measuring soil radon fluxes is presented in Appendix K, along with
supporting quality assurance data on field blanks and duplicate measurements. A total of
1,041 radon flux measurements were performed at 330 unique locations. These site locations
covered all 67 counties in Florida. The exact locations and results of the radon flux
measurements are presented in Appendix K. Figure 6-1 illustrates the locations of the flux
measurement sites. As illustrated, more emphasis was placed on sampling in areas where
the calculated radon potential was high or highly variable.
Temporal variations in radon flux were examined in addition to the spatial sampling
at the 330 locations by repeating measurements at one site during different seasons of a 17-
month period in 1993-94. The site for the repeated measurements was near the FRRP test
cell structures at the Florida Institute of Phosphate Research in Bartow. The measurements
at this site included individual samples at seven different points around the test cells.
Measurements were made at the same point in each set. The results of these measurements
are summarized in Table 6-1. The geometric standard deviations among the replicate
measurements at each of the seven measurement points were averaged to estimate a time-
dependent GSD among the radon fluxes of 1.6.
6-2
-------
w
>- —;
SUWAMNU ' • £
Sv^ cut*
^UNON r'A$ ' Ct-AYVv
f-.-sf" i
IJtMYCm 1
, • •
yy& £ -
}~i
' # PASCO
*•
•••
A*
••
•• ,
Figure 6-1. Illustration of radon flux sampling locations.
6-3
-------
Table 6-1. Radon Flux Measurements Near the FRRP Test Cells in Bartow.
Soil Radon Flux (pCi m"2 s'1)
Location
3/16/93
6/2/93
10/21/93
2/11/94
6/12/94
8/4/94
GSD
NW
2.3
2.9
2.1
2.7
4.4
0.9
1.7
W
6.1
4.2
6.0
3.4
a
3.5
1.5
SW
5.4
3.4
3.6
4.1
6.4
3.6
1.4
c
4.5
5.6
6.5
3.5
5.2
4.0
1.3
NE
0.6
2.3
1.4
1.7
a
0.7
1.8
E
1.6
7.9
12.5
4.6
a
4.0
2.2
SE
4.3
7.9
9.4
4.8
15.0
14.4
1.7
Average:
1.6
"Not measured.
Considering the temporal variations, site variations (from gamma-ray and replicate
flux measurements), and polygon variations in soil radon potential (Table C-l), the measured
radon fluxes were compared to the mapped soil radon potentials. Figure 6-2 illustrates this
comparison in terms of a bias statistic between the measured and mapped soil radon fluxes.
The bias statistic is defined as:
Z = [ln(Meas) - ln(Map)] / V (In GMeas)z + (In GMap)2 (34)
where Z = measurement-map bias statistic (standard deviations)
Meas = value of the measured parameter (point in time)
Map = value of the mapped parameter (annual average basis)
= uncertainty (GSD) in representing annual average by the measured value
Gr^ap = uncertainty (GSD) of the mapped value.
The estimate of G^^ was calculated to include both temporal and spatial uncertainties as:
GMeas = exp[V £i(ln G-/ ] (35)
where Gj = GSD of component i.
6-4
-------
50
40
30
20
10
RAE Radon Flux Data
Equiv. Normal Distrib.
-5
Mean = +0.12
Std. Dev. = 1.33
n = 330
f i
v/A ///.
%
-'ss.-
'A* --•> /X.
//,' y '/*
/// ' "/
'/A -
.w:
iippi
-4-3-2-101234
Measurement-Map Bias, Z (standard deviations)
Figure 6-2. Distribution of bias statistics for the radon flux measurements.
The distribution of the bias statistics computed from equation (34) for the measured
and mapped radon flux comparisons is illustrated in Figure 6-2. The distribution averaged
0.12, indicating minimal average bias for the overall data set. The standard deviation of the
distribution was 1.33, indicating more scatter in the radon flux measurements than was
predicted from the radon potential map calculations. With the excess scatter, 16
measurements (4.8%) were below the central 95% range of the distribution, and 18
measurements (5.5%) were above it, compared to 8 measurements (2.5%) expected for each.
The extra scatter is attributed primarily to the temporal variations in radon flux, since short-
term (24 h) radon fluxes are being compared to annual-average fluxes from the map
calculations.
6-5
-------
6.2
STATE-WIDE VALIDATION WITH INDOOR RADON DATA SETS
The soil radon potential maps were validated on a general, state-wide basis by
comparisons of their predictions for radon in the reference house to radon levels observed in
actual houses. Three data sets were used for these validations, all of which were based on
short-term charcoal canister measurements of indoor radon. The first, and most
representative was the Geomet land-based data set (Nag87), which contained 2,952 radon
measurements covering all 67 Florida counties. The second data set was the Geomet
population-based data set (Nag87), which contained 2,095 radon measurements in 57
counties. The third data set was the Florida Health and Rehabilitative Services (HRS)
residential data set (Form 1750 data), which contained 3,991 radon measurements in 49
counties.
The distributions of measurements by comity are shown in Figures 6-3 through 6-5
for the land-based, population-based, and HRS residential data sets, respectively. Their
biases are compared in Figure 6-6, which shows the cumulative percentages of measurements
in all Florida counties. A nearly-straight line would show the ideal case, in which each
county has equal sample representation. Although all of the sets show some bias, the radon
flux data and the Geomet land-based data sets have the least bias because they include at
least some measurements in every county. The other sets have many more measurements
in some counties, but leave many others unrepresented.
6-6
-------
*3
5>
*4
M*
~i
»
I
CO
a
on
<*-
~<
cr
o
3
O
o
B
«
«-
to
S3
a
¦
cr
£0
CR
«
a
~i
B3
a
o
a
3
a
S3
§
<5
5
(5
B
CJ*
«<
a
o
§
Number of Measurements in Each County
Jetturson
Liberty
Madison
Lafayette
Calhoun
Union
Holmes
Hendry
Franklin
Washington
Columbia
Hamilton
Bradford
Hardee
Gilchrist
Indian River
Santa Rosa
St. Johns
Okaloosa
Highlands
Sumter
Putnam
Duval
Palm Beach
Citrus
Charlotte
Pinellas
Alachua
Hillsborough
Marion
-------
Cumulative Percent of Measurements in Each County
Cum%4)<1
*1
Number of Measurements in Each County
Jefferson
Liberty
Glades
Madison
Lafayette
Calhoun
Gulf
Union
Holmes
Wakulla
Hendry
Franklin
Washington
Gadsden
Okeechobee
DeSoto
Hamilton
Bradford
Sumter
Citrus
Dixie
Baker
Hardee
Gilchrist
Levy
Suwannee
Flagler
Walton
Marion
Taylor
Highlands
Putnam
Monroe
Columbia
Nassau
Lake
Osceola
St. Lucie
Indian Rh/er
Collier
Jackson
Bay
Hernando
Santa Rosa
Charlotte
Martin
Manatee
Clay
St. Johns
Okaloosa
Brevard
Pasco
Volusia
Escambia
Lee
Duval
Polk
Leon
Seminole
Alachua
Orange
Pinellas
Sarasota
Paim Beach
Broward
Dade
Hillsborough
s 5
H :~
CO CO
CD _
id m
-»¦
X
X I
JSL.
-------
Radon measurements from the Geomet land-based and population-based data sets
were compared directly with the numerical predictions for the reference house from the soil
radon potential map data. The latitude and longitude associated with each measurement
point were associated with their corresponding map polygon by the University of Florida
GeoPlan Center using a GIS system. Each measurement was compared with the median
annual-average indoor radon calculated for the reference house using Cout=0.1 pCi L"1 and
x=1.3 pCi L"1 per mCi y"1 in equation (32). The comparison used the bias statistic defined
by equation (34) with Gmeas = 2.083 for the uncertainty in annual-average indoor radon when
estimated from a single charcoal canister measurement (Roe90). The value of Gmap was
estimated from the median and 95% confidence limits of soil radon potential as
Gm.p^Qss/Qso)1"'645-
The bias statistics calculated for each radon measurement are listed in Appendix L
for the Geomet land-based data set, and in Appendix M for the Geomet population-based data
set. The bias statistics were summarized on a state-wide basis for each data set by fitting
to an equivalent normal distribution. The fitted distribution plot for the land-based data set
is illustrated in Figure 6-7. The land-based data showed virtually no bias for the state-wide
average, with a mean value of only Z=-0.04 standard deviations. The standard deviation of
the Z distribution was 0.99, compared to an ideal value of 1.00, indicating that the observed
variations between measured and mapped values are almost exactly equal to the variations
predicted from map and measurement uncertainties.
6-9
-------
600
P 400
Geomet Land Data
¦ Equiv. Normal Distrib.
Ill
vv' y/s Y//,
/.•/ .*//
•w '//S ///.
ymm
Mean = -0.04
Std. Dev. = 0.99
n = 2952
ill
-4 :3 -2 -1 0 1 2 3 4 5
Measurement-Map Bias, Z (standard deviations)
Figure 6-7. Distribution of bias statistics for the Geomet land-based indoor radon
measurements.
A corresponding distribution plot of the bias statistics for the Geomet population-based
data set is presented in Figure 6-8. The population-based data showed a negative bias of -
0.42 standard deviations, indicating that for the state-wide average, the population-based
measurements were slightly lower than the values predicted by the map for the reference
house. The standard deviation of the Z distribution was 0.99, indicating that the observed
variations between measured and mapped values were again almost exactly equal to the
variations predicted from map and measurement uncertainties.
6-10
-------
500
400
w
-~—<
c
CD
E
-------
Qso = O.OlEpiQjo, (36)
Qs,5 = O.OlEpjCfea (37)
where Q50 = polygon-weighted average median radon potential for a zip code (mCi y"1)
Q95 = polygon-weighted average 95% C.L. radon potential for a zip code (mCi y"1)
Pi = percentage of polygon i in the zip code area.
Comparisons of each measurement in the HRS residential data set were then made
with the average map predictions for their zip code area using Q50 in equation (32) with
Cout=0.1 pCi L"1 and x=1.3 pCi L"1 per mCi y"1. The comparisons are reported in Appendix
N in terms of the bias statistic defined by equation (34). The uncertainties for use in
equation (34) were defined as with the Geomet data sets as Gmeas=2.083 and
-------
900
800
to
700
CD
i 600
$ 500
cu
400
o3 300
.Q
I 200
100
0
HRS Residential Data
Equiv. Normal Distrib.
Mean = +0.51
Std. Dev. = 0.96
n = 3991
/
'//s\fa
\
/// y//-
I !»>
ill;
¦;# 5^
^ V//. '//, '///.
ill#
II
-5-4-3-2-101234
Measurement-Map Bias, Z (standard deviations)
Figure 6-9. Distribution of bias statistics for the HRS residential indoor radon
measurements.
6.3 EXAMINATION OF INDOOR RADON ANOMALIES
Indoor radon anomalies are difficult to positively identify because they can not be
individually distinguished from the small percentage of houses that are statistically expected
to have poor agreement between measurements and map predictions. The approach used
here to search for indoor radon anomalies was to calculate from the bias statistics the
number of houses expected to exceed a prescribed limit, and then to compare the actual
number of houses exceeding the limit with the expected value. If the actual numbers of
houses outside the limits significantly exceeds the expected number, then the excess number
may constitute an anomaly if all the houses above the limit correspond to the reference house
used in developing the radon map. If the potentially anomalous houses have features that
6-13
-------
could otherwise explain their disagreement with the map, however, they may not necessarily
constitute an anomaly.
The search for possible indoor radon anomalies was conducted on a state-wide basis
using all three of the indoor radon data sets. Houses with very large positive or negative Z
statistics were flagged as potential anomalies, and their locations were compiled from their
map coordinates or, in the case of the HRS residential data, from their zip codes. Where the
potential anomalies occurred in regional clusters, the houses were investigated by drive-by
observations. The observations included building type (single-family detached vs. large
building), floor type (slab-on-grade vs. crawl space), wall construction (frame, concrete block,
brick, etc.), soil surface gamma-ray intensity, and other pertinent features (such as number
of stories, terrain slope, evidence of basements, vents from crawl spaces, fireplace vents, and
approximate age). In some regions, the gamma activity near concrete and aggregate
materials was also observed, and samples were analyzed to characterize materials of interest.
The house investigations utilized a global positioning system (NAV-5000D, Magellan
Systems Corp., San Dimas, CA) to locate houses from map coordinates in the Geomet data
sets, and street addresses to locate houses in the HRS data sets. The latitude and longitude
coordinates of the HRS houses were also determined with the global positioning system
during the house investigations to permit a positive identification of the map polygon where
the house was located. To maintain the anonymity and privacy of the data and house
occupants, no contacts were made with house occupants, no addresses from the Geomet data
sets were recorded, and all observations were made from an automobile during an
approximate 1-minute curb-side period, during which a 0.5-minute gamma-ray measurement
was made over an exposed soil or sod surface. The gamma-ray measurements utilized a 5
cm x 5 cm scintillation detector (Model 44-10, Ludlum Measurements, Inc., Sweetwater, TX).
Houses were flagged as potential anomalies for on-site investigation in two stages.
Initially, measurement uncertainties were ignored (Gmeas=0), and Z statistics calculated by
equation (34) were flagged and investigated as potential anomalies if they were below -3 or
above +3. After the magnitude of Gmeas was identified (from Roe90), all of the Z statistics
were re-calculated using Gmeas=2.08 in equation (34), and the potential anomalies were re-
6-14
-------
flagged using a \Z\ >1.96 criterion to define a central range that should contain 95% of the
observations. The latter criterion included many additional houses, but in some cases
excluded houses that had already been investigated. For completeness, all houses that were
investigated are listed in Appendix O with their final Z statistic that includes measurement
uncertainty.
The numbers of houses flagged in each of the data sets as potential anomalies are
listed in Table 6-2 for comparison with the number of houses expected to randomly vary
beyond the IZI=1.96 limit that should include the central 95% of the data. The Geomet
land-based data set, which is the most representative and least biased of the sets, shows
slightly less than the expected number of random negative anomalies (1.9% vs. 2.5%), and
slightly more than the expected number of random positive anomalies (2.7% vs. 2.5%).
Compared to the expected numbers of measurements above and below the central 95% range,
the observed numbers of potential anomalies are not significantly different from the random
variations.
Table 6-2. Statistical Summary of State-Wide Radon Map Validations
Data Set
Potential Potential
No. of Negative Anomalies Positive Anomalies
Comparisons Number Percent Number Percent
Expected:
Observed:
Geomet
Land-Based Data 2,952
Geomet
Population-Based Data 2,095
HRS
Residential Data 3,991
HRS Residential Data
Excluding Large Buildings 3,938
56
84
32
28
2.5%
1.9%
4.0%
0.8%
0.7%
80
31
235
185
2.5%
2.7%
1.5%
5.9%
4.7%
6-15
-------
The population-based data set showed a total of 115 measurements outside the
expected central 95% range, compared to 105 expected from random variations. This
difference is not significant. However, because of the bias noted for this set, most of the
measurements outside the central range are on the low side, contributing to a significant
excess of potential negative anomalies. Whether or not the excess number of low
measurements is caused by actual low-radon anomalies depends on how many of the houses
have significant mitigating features that cause them to be more radon resistant than the
reference house.
The HRS residential data showed a total of 267 measurements outside the expected
central 95% range, compared to 200 expected from random variations. This difference is
significant. However, when 53 of the measurements are removed from the comparison
because they were observed to come from large buildings instead of detached houses, the
number of measurements outside the central 95% range drops to 213. This number is not
significantly different from the 200 expected from random variations. The 53 observations
removed were among the potential anomalies given follow-up investigation (Appendix O). It
is possible that even more large buildings are among the other potential anomalies listed in
Appendix O that were not investigated, and that the comparison could become even closer.
However, because of the bias noted for this data set, most of the measurements outside the
central range are on the high side, contributing to a significant excess of potential positive
anomalies. Whether or not the excess number of high measurements is caused by actual
high-radon anomalies depends on how many of the houses have significant radon entry routes
that cause them to be less radon resistant than the reference house.
In selected regions where potential anomalies were investigated, the gamma-ray
activity near concrete and aggregate materials was also observed. Samples were collected
for radium and emanation measurements at some of these sites. The results of these
observations are listed in Table 6-3. The measurements for the Lee County area suggest
higher radium levels in the aggregates and concretes than were observed at the sites in the
other two counties. Simple model analyses suggest that this magnitude of radium elevation
in concrete could cause incremental increases in indoor radon on the order of 1-2 pCi L"1. In
6-16
-------
houses that would otherwise be just under the 4 pCi L"1 standard, this increase could cause
them to exceed it.
Table 6-3. Gamma-Ray and Radon Measurements of Concretes and Aggregates.
County
Longitude
Latitude
Material
Gamma
(uR h"1)
Radium
(pCi g"1)
Emanation
(%)
Dade
80°29.211'
25°41.387'
Aggregate
8.7
1.7
1.5
Dade
80°21.633'
25°54.772'
Aggregate
6.3
a
a
Dade
80°22.096'
25°52.1591
Aggregate
5.3
a
a
Dade
80°23.999'
25°46.914'
Aggregate
5.9
a
a
Dade
80°29.216'
25°42.452'
Aggregate
5.7
a
a
Dade
80°24.863'
25°37.883'
Aggregate
4.1
a
a
Lee
81°49.180'
26°29.477'
Concrete
...b
4.0
15.5
Lee
81°49.180'
26°29.535'
Concrete
11.6
3.8
3.7
Lee
81°49.180'
26°29.477'
Aggregate
18.2
3.8
4.7
Lee
81°41.658'
26°29.857'
Aggregate
14.5
5.0
4.0
Lee
81°49.519'
26929.787'
Aggregate
17.4
5.1
3.5
Lee
81°45.592'
26°29.457'
Aggregate
14.8
3.1
4.5
Sumter
82°00.477'
28°39.037'
Aggregate
6.8
1.5
10.1
"Not sampled for analysis.
6Not measured
Analyses of the slab and wall construction details observed for the potential anomalies
in Table 0-1 suggests some significant aggravating and mitigating trends may be present in
the potential anomaly houses that were investigated. For example, analyses of the 80 houses
investigated from the Geomet land-based data set (Fig. 6-10) showed that houses above the
95% mid-range were about four times more likely to use slab-on-grade construction than to
have crawl spaces, while the opposite trend was seen for houses below the mid-range.
Similarly, houses above the 95% mid-range were over 40% more likely to use hollow-block
wall construction than frame walls, while the opposite trend was seen for houses below the
mid-range. These trends are consistent with model predictions, which show that crawl spaces
dilute sub-floor radon before it enters houses, and that hollow-block exterior walls may
provide channels for enhanced soil gas entry.
6-17
-------
Similar slab and wall construction trends were observed for the other data sets. When
aggregated, all 251 of the houses from the land-based, population-based, and HRS residential
data sets in Appendix O showed almost identical trends, as illustrated in Fig. 6-11. In this
case, houses above the 95% mid-range were nearly three times more likely to use slab-on-
grade construction than to have crawl spaces, while the opposite trend was seen for houses
below the mid-range. Similarly, houses above the 95% mid-range were 50% more likely to
use hollow-block wall construction than frame walls, while the opposite trend was seen for
houses below the mid-range. These state-wide trends are sufficient to explain the potential
anomalies that are combined with the expected random measurements that fall outside the
central 95% range.
Crawl
Space
on Grade
m Slab
aj 50
100
Negative
Positive
CO
g
"c5
E
o
c
<
c
-------
¦ Crawl Space
E3 Slab on Grade
¦ Frame
^ Block
100
100
Negative
Positive
Negative
Positive
Potential Anomaly
Potential Anomaly
Figure 6-11. Comparison of floor (a) and wall (b) construction features in
potentially anomalous houses in the 251 cases investigated from the land-
based, population-based, and HRS residential data sets.
6.4 MAP VALIDATION SUMMARY
The maps were validated by state-wide comparisons with over a thousand soil radon
flux measurements at 330 locations and with 9,038 indoor radon measurements from three
different data sets. The radon flux measurements averaged similar to the map predictions,
but were scattered more widely than the map data (16 below and 18 above the central 95%
range, compared to eight expected for each). The difference in scatter is caused by
inadequate definition of the temporal variations in radon flux that are needed for comparing
the 24-hour flux measurements to annual-average calculated values.
The Geomet land-based data set best represents all regions of Florida and agrees very
well with the map predictions. The middle 95% of the map range included 95.4% of the 2,952
measurements, with 1.9% below and 2.7% above the mid-range, compared to 2.5% expected
for each. The HRS residential and Geomet population data sets do not represent all regions
6-19
-------
in Florida, but they were compared with the map predictions anyway. The 2,095
measurements in the Geomet population-based set averaged slightly lower than map values,
while the 3,938 measurements in the HRS residential data set averaged slightly higher than
map values.
Over 250 houses with the greatest differences between measured and predicted indoor
radon concentrations were investigated and found to show trends that offer further
explanations. Houses above the 95% mid-range were nearly three times more likely to use
slab-on-grade construction than to have crawl spaces, while the opposite trend was seen for
houses below the mid-range. Similarly, houses above the 95% mid-range were about 50%
more likely to use hollow-block construction than frame construction, and the opposite trend
was also seen for houses below the mid-range. These trends are consistent with model
predictions, and account for potential anomalies on a state-wide level. Considering the
variations in both measurements and map calculations, the measurements give excellent
overall state-wide validation of the radon maps.
6-20
-------
Section 7
LITERATURE REFERENCES
Acr90 Acres. Measurement of Crack and Opening Contribution to Radon Entry (Feasibility Study).
Vol. Ill of Radon Entry Through Cracks in Slabs-on-Grade, Acres International Corp., report
P09314, 1990.
Aus78 Austin, S.R. and Droullard, R.F., Radon Emanation From Domestic Uranium Ores
Determined by Modifications of the Closed-Can, Gamma-Only Assay Method. Washington
D.C.: U.S. Department of the Interior, Bureau of Mines; Report of Investigations 8264, 1978.
Bar76 Barakat, R., Sums of Independent Lognormally Distributed Random Variables, Journal of the
Optical Society of America 66: 211-216, 1976.
Bre90 Brennan, T., Clarkin, M., Osborne, M.C., and Brodhead, B., Evaluation of Radon Resistant
New Construction Techniques, in Proceedings: the 1990 International Symposium on Radon
and Radon Reduction Technology, Vol.2, EPA/600/9-91/026b (NTIS PB91-234450), VIII-1,
1991.
Cla91 Clarkin, M., and Brennan, T., Radon Resistant Construction Techniques for New Residential
Construction. Washington D. C.: U. S. Environmental Protection Agency report EPA/625/2-
91/032, 1991.
Coh86 Cohen, B.L., A National Survey of 222Rn in U. S. Homes and Correlating Factors. Health
Physics 51:175-183; 1986.
Cum92 Cummings, J.B., Tooley, J.J., and Moyer, N., Radon Pressure Differential Project, Phase I,
FRRP, U. S. Environmental Protection Agency report EPA-600-R92-008 (NTIS PB92-
148519), January 1992.
Dev87 Devore, J.L., Probability and Statistics for Engineering and the Sciences, 2nd edition,
Monterey, CA: Brooks/Cole Publishing Company, 1987.
EGG81 E G & G Geometries, Aerial Gamma Ray and Magnetic Survey, Gainesville and Daytona
Beach Quadrangles Florida. Grand Junction, CO: U. S. Department of Energy Report GJBX-
101, March 1981.
EPA86 Environmental Protection Agency, Radon Reduction Techniques for Detached Houses,
Technical Guidance. U.S. Environmental Protection Agency, Air and Energy Engineering
Research Laboratory report EPA/625/5-86/019, 1986.
7-1
-------
EPA92a Environmental Protection Agency, Technical Support Document for the 1992 Citizen's Guide
to Radon. Washington D.C.: Office of Air and Radiation report EPA-400-R-92-011 (NTIS
PB92-218395), May 1992.
EPA92b Environmental Protection Agency, National Residential Radon Survey Summary Report
Washington D.C.: Office of Air and Radiation report EPA-402/R-92-011, October 1992.
Fin89 Findlay, W.O., Robertson, A., and Scott, A.G., Testing of Indoor Radon Reduction
Techniques in Central Ohio Houses: Phase 1 (Winter 1987-88). U. S. Environmental
Protection Agency, report EPA/600/8-89/071 (NTIS PB89-219984), July 1989.
Gil93 Gilbert, R.O., personal communication, Richland, WA: Battelle Pacific Northwest Laboratory,
1993.
Has75 Hastings, N.A.J, and Peacock, J.B., Statistical Distributions, New York: John Wiley & Sons,
1975.
Lou87 Loureiro, C.O., Simulation of the Steady-State Transport of Radon from Soil into Houses with
Basements under Constant Negative Pressure. Berkeley, CA: Lawrence Berkeley Laboratory
report LBL-24378, 1987.
Lut79 Lutton, R.J., Regan, G.L., and Jones, L.W., Design and Construction of Covers for Solid
Waste Landfills. Cincinnati OH: Municipal Environmental Research Laboratory report EPA-
600/2-79-165 (NTIS PB80-100381), 1979.
Mit68 Mitchell, R.L., Permanence of the Log-Normal Distribution, Journal of the Optical Society
of America 58: 1267-1272, 1968.
Mur90 Murane, D.M., Model Standards and Techniques for Controlling Radon Levels Within New
Buildings, in Proceedings: the 1990 International Symposium on Radon and Radon Reduction
Technology. Vol. 3, EPA/600/9-91/026c (NTIS PB91-234468),A-I-2, 1991.
Nag87 Nagda, N.L., Koontz, M.D., Fortmann, R.C., Schoenborn, W.A., and Mehegan, L.L.,
Florida Statewide Radiation Study. Germantown, MD: Geomet Technologies Inc. report IE-
1808; 1987.
Nat66 Natrella, M.G., Experimental Statistics, Washington D.C.: U.S. Department of Commerce,
National Bureau of Standards Handbook 91, 1966.
Naz87 Nazaroff, W.W., Lewis, S.R., Doyle, S.M., Moed, B.A., and Nero, A.V., Experiments on
Pollutant Transport from Soil into Residential Basements by Pressure-Driven Airflow.
Environmental Science and Technology 21:459-466, 1987.
Naz88 Nazaroff, W.W., Doyle, S.M., Nero, A.V., and Sextro, R.G., Radon Entry via Potable
Water, pp. 131-157 in: Radon and Its Decay Products in Indoor Air, W. W. Nazaroff and
A. V.Nero, eds., New York: Wiley & Sons, 1988.
7-2
-------
Naz89 Nazaroff, W.W. and Sextro, R.G., Technique for Measuring the Indoor 222Rn Source Potential
of Soil, Environmental Science and Technology 23:451-458, 1989.
Ner88 Nero, A.V., Radon and Its Decay Products in Indoor Air: An Overview, in Radon and Its
Decay Products in Indoor Air, W.W. Nazaroff and A.V. Nero, eds., p. 1-53, New York:
John Wiley and Sons, 1988.
Nie82 Nielson, K.K., Rogers, V.C., Mauch, M.L., Hartley, J.N., and Freeman, H.D., Radon
Emanation Characteristics of Uranium Mill Tailings, in: Uranium Mill Tailings Management -
V, Ft. Collins, CO: Colorado State Univ., 355-367, 1982.
Nie84 Nielson, K.K., Rogers, V.C., and Gee, G.W., Diffusion of Radon Through Soils: A Pore
Distribution Model. Soil Science Society of America Journal 48:482-487, 1984
Nie91a Nielson, K.K. and Rogers, V.C., Feasibility and Approach for Mapping Radon Potentials in
Florida. Florida Radon Research Program. Research Triangle Park, NC: U. S.
Environmental Protection Agency report EPA-600/8-91-046 (NTIS PB91-217372), July 1991.
Nie91b Nielson, K.K. and Rogers, V.C., Proceedings of the Workshop on Radon Potential Mapping.
Florida Radon Research Program. Research Triangle Park, NC: U. S. Environmental
Protection Agency report EPA-600/9-91-044 (NTIS PB92-115278),November 1991.
Nie91c Nielson, K.K. and Rogers, V.C., Development of a Prototype Map of the Soil Radon
Potentials in Alachua County Florida, Salt Lake City, UT: Rogers & Associates Engineering
Corp. RAE-9127/3-1, October 1991.
Nie91d Nielson, K.K. Rogers, V.C., and Rogers, V., Modeling Radon Generation, Transport, and
Indoor Entry for Building Construction Standards, Salt Lake City, UT: Rogers & Associates
Engineering Corp. RAE-9127/3-2, October 1991.
Nie92a Nielson, K.K. and Rogers, V.C., Radon Transport Properties of Soil Classes for Estimating
Indoor Radon Entry, in: Indoor Radon and Lung Cancer: Reality or Myth ?, F.T. Cross, ed.,
p. 357-372, 1992.
Nie92b Nielson, K.K., Rogers, V.C., Otton, J.K., Brown, R.B., and Harris, W.G., Soil Radon
Potential Mapping and Validation for Central Florida, in Proceedings: the 1992 International
Symposium on Radon and Radon Reduction Technology. Vol. 3, EPA/600/R-93-083c (NTIS
PB93-196210), 1993.
Nie94a Nielson, K.K., Rogers, V.C., Rogers, V., and Holt, R. B. The RAETRAD Model of Radon
Gas Generation, Transport,and Indoor Entry, EPA-600/R-94-198 (NTIS PB95-142030),
November 1994.
Nie94b Nielson, K.K., Holt, R.B., and Rogers, V.C., Soil Radon Potential Mapping of Twelve
Counties in North-Central Florida, EPA/600-R-94-218 (NTIS PB95-159869), December 1994.
7-3
-------
Nit89 Nitschke, I., Radon Reduction and Radon Resistant Construction Demonstrations in New
York. U. S. Environmental Protection Agency, Air and Energy Engineering Research
Laboratory report EPA/600/8-89/001 (NTIS PB89-151476), 1989.
Nue90 Nuess, M., and Price, S., Public Policy Considerations and the Development of a Code for
the Control of Radon in Residences, in Proceedings: the 1990 International Symposium on
Radon and Radon Reduction Technology, Vol. 1, EPA/600/9-91/026a (NTIS PB91-234443),
1-4, 1991.
Osb88 Osborne, M.C., Radon-Resistant Residential New Construction. U. S. Environmental
Protection Agency, Air and Energy Engineering Research Laboratory report EPA/600/8-
88/087 (m\S PB90-274119), 1988.
Pea90 Peake, R. T. , Gundersen, L. C. S. , Schumann, R. R. , James, J., and Ronca-Batista, M,,
Determination of Radon Geologic Provinces in the United States, in proceedings: the 1990
International Symposium on Radon and Radon Reduction Technology. Vol. 3, EPA/600/9-
91/026c (NTIS PB91-234450), C-VI-1, February 1991.
Rev91 Revzan, K.L., Fisk. W.J., and Gadgil, A.J., Modeling Radon Entry Into Houses with
Basements: Model Description and Verification, in Indoor Air, Vol.2, pp. 173-190, 1991.
Roe90 Roessler, C.E., Revell, J.W., and Wen, M.J., Temporal Patterns of Indoor Radon in North
Central Florida and Comparison of Short-Term Monitoring to Long-Term Averages, in
Proceedings: the 1990 International Symposium on Radon and Radon Reduction Technology,
Vol. 1, EPA/600/9-91-026a (NTIS PB91-234443), February 1991.
Rog84 Rogers, V.C. and Nielson, K.K., and Kalkwarf, D.R., Radon Attenuation Handbook for
Uranium Mill Tailings Cover Design, Washington D. C.: U. S. Nuclear Regulatory
Commission report NUREG/CR-3533, 1984.
Rog89 Rogers, V.C., Nielson, K.K., and Merrell, G.B., Radon Generation, Adsorption, Absorption,
and Transport in Porous Media. Washington D. C.: U. S. Department of Energy report
DOE/ER/60664-1; 1989.
Rog90 Rogers, V.C. and Nielson, K.K., Benchmark and Application of the RAETRAD Model, in:
Proceedings: the 1990 International Symposium on Radon and Radon Reduction Technology,
Vol. 2, EPA/600/9-91/026b (NTIS PB91-234450), VI-1, 1991.
Rog91a Rogers, V.C. and Nielson, K.K., Multiphase Radon Generation and Transport in Porous
Materials. Health Physics 60:807-815, 1991.
Rog91b Rogers, V.C. and Nielson, K.K., Correlations for Predicting Air Permeabilities and 222Rn
Diffusion Coefficients of Soils. Health Physics 61:225-230, 1991.
7-4
-------
Rog95 Rogers, V.C., Nielson, K.K., and Holt, R.B., Radon Generation and Transport in Aged
Concrete, U. S. Environmental Protection Agency Report EPA-600/R-95-032 (NTIS PB95-
181590), February 1995.
San90 Sanchez, D.C., Dixon, R, and Williamson, A.D., The Florida Radon Research Program:
Systematic Development of a Basis for Statewide Standards, in Proceedings: the 1990
International Symposium on Radon and Radon Reduction Technology, Vol. 3, EPA/600-9-91-
026c (NTIS PB91-234450), A-I-3, 1991.
SBC90 Florida Code for Radon-Resistant Construction and Mitigation, Southern Building Code
Congress International, Inc., Rev. 1, January 1990.
Sco88 Scott, A.G. and Robertson, A., Follow-Up Alpha-Track Monitoring of 40 Eastern
Pennsylvania Houses with Indoor Radon Reduction Systems (Winter 1987-88), U. S.
Environmental Protection Agency, AEERL report EPA/600/8-88/098 (NTIS PB89-110035),
September 1988.
Sco89 Scott, A.G. and Robertson, A., Follow-Up Alpha-Track Monitoring in 40 Eastern
Pennsylvania Houses with Indoor Radon Reduction Systems (Winter 1988-89), U. S.
Environmental Protection Agency, AEERL report EPA/600/8-89/083 (NTIS PB90-134172),
October 1989.
SCS75 Soil Conservation Service, Soil Taxonomy. A Basic System of Soil Classification for Making
and Interpreting Soil Surveys. Washington D. C.: U. S. Department of Agriculture, Soil
Conservation Service. Agriculture Handbook No. 436, 1975.
SCS91 Soil Conservation Service, State Soil Geographic Data Base (STATSGO) Data Users Guide,
Lincoln, Nebraska: National Soil Survey Center, Soil Conservation Service, U. S. Department
of Agriculture, report, 1991.
Sul88 Sullivan, T.M., Kempf, C.R., Suen, C.J., and Mughabghab, S.M., Low-Level Radioactive
Waste Source Term Model Development and Testing, Washington D. C.: U. S. Nuclear
Regulatory Commission report NUREG/CR-5204, 1988.
Swe86 Swedjemark, G.A., Swedish Limitation Schemes to Decrease Rn Daughters in Indoor Air,
Health Physics 51:569-578, 1986.
Swe90 Swedjemark, G.A. and Makitalo, A., Recent Swedish Experiences in 222Rn Control, Health
Physics 58:453-460, 1990.
Tha82 Thamer, B.J., Nielson, K.K., and Felthauser, K. The Effects of Moisture on Radon
Emanation. Washington D.C.: U.S. Department of the Interior, Bureau of Mines; Open File
Report OFR-184-82, 1982.
Tho64 Thornthwaite, C.W. and Mather, J.R. Average Climatic Water Balance Data of The
Continents. Part VII: United States. Centerton, NJ: Publications in Climatology 17 (3), 485,
1964.
7-5
-------
Tho85 Thomas, B.P., Cummings, E., and Wittstruck, W.H., Soil Survey of Alachua County Florida,
Gainesville, FL: U. S. Department of Agriculture, Soil Conservation Service, 1985.
Von90 Vonstille, W.T. and Sacarello, H.L.A., Assessment of Health Impacts of Radon Exposures
in Florida, in Proceedings: the 1990 International Symposium on Radon and Radon Reduction
Technology, Vol. 3, EPA/600/9-91-026c (NTIS PB91-234450), A-II-4, 1991.
Wil91 Williamson, A.D. and Finkel, J.M. Standard Measurement Protocols, Florida Radon
Research Program, EPA-600/8-91-212 (NTIS PB92-115294), November 1991.
Yeh87 Yeh, G.T., FEMWATER: A Finite Element Model of Water Flow Through Saturated-
Unsaturated Porous Media - First Revision, Oak Ridge, TN: U. S. Department of Energy
report ORNL-5567/R1, 1987.
7-6
------- |