Soil uranium, soil gas radon and indoor radon empirical ... · Don Appleton (BGS) and Jon Miles (HPA) ... 0 1 2 3 4 5 6 7 8 ... • ASSUMPTIONS: linear relationship between surrogates

© NERC All rights reserved

Soil uranium, soil gas radon and indoor radon empirical relationships in the UK and other

European countries

Don Appleton (BGS) and Jon Miles (HPA)


EUROPEAN GEOGENIC RADON POTENTIAL MAP

• Wide range of numerical (quantitative) data types: indoor radon, soil gas radon, U in soil (measured and estimated from airborne or ground gamma spectrometry), U in rock; U in sediment.

• European radon potential classification based on combining layers of available input variables requires an understanding of the empirical relationships between indoor radon (main hazard to human health) and other numerical data types.

• European radon potential classification will also have to deal with categorical (qualitative) data (e.g. rock type and structural features, such as faults) but this is outside scope of this presentation.


DATA

• Estimated U in <2mm topsoil derived from estimated U (eU) from airborne gamma spectrometry [Estimated <2mm soil U = 1.7216 * HiRES eU mg/kg)]

• U in topsoil (5-20 cm depth) soil geochemical samples (<2mm fraction); potentially better correlation with airborne gamma spectrometry data than subsurface (20-50 cm soil U data)

• Radon concentrations in soil gas• Radon concentrations in homes

Data grouped by geological unit and sub-area (usually 1-km or 5-km grid square)


REGRESSION ANALYSIS

• Spatially and geologically grouped data should be approximately normally distributed to give most robust central estimate for least squares linear regression analysis

• Anderson-Darling normality tests indicate regression modelling should be based on:

1. Arithmetic means of airborne eU and soil U data2. Geometric means of lognormally distributed soil gas

and indoor radon data

• LS regression analysis results have limitations related to uncertainties of x and y axis data AND outliers


MODELS BASED ON AIRBORNE GAMMA

SPECTROMETRY DATA

Estimated U in <2mm topsoil derived from estimated U (eU) from HiRES airborne gamma spectrometry [Estimated <2mm soil U = 1.7216 * HiRES eU mg/kg)]


Figure 1: Least squares regression models between e stimated soil U (derived from HiRES eU) and GM indoor radon (data g rouped by 1-km grid square and geology; thin lines have intercepts set to 5 Bq m -3))

0

50

100

150

200

250

300

350

400

0 1 2 3 4 5 6 7 8

estimated soil U mg/kg

GM

indo

or ra

don

Bq

m-3

DINLM

DINMDMIX

TRIMD

Linear (DINLM)

Linear (DINLM)

Linear (DINMDMIX)

Linear (DINMDMIX)

Linear (TRIMD)

Linear (TRIMD)


INTERCEPTS

• High positive intercepts on y axis of least squares (LS) linear regression models probably due to uncertainties of indoor radon, soil gas radon and soil U data

• UNSCEAR (1993) estimated world mean outdoor radon (10 Bq m-3) and indoor contribution from building materials (6 Bq m-3)

• UNSCEAR intercept of 16 Bq m-3 inappropriate for UK where a value of 5 Bq m-3 is the best estimate for combined indoor contribution from outdoor radon and UK building materials

• LS regression models for UK data therefore forced to intercept y-axis at 5 Bq m-3 (thin regression lines in previous slide)


Figure 2: Least squares regression models between estimated soil U (derived from HiRES eU) and GM indoor radon (data grouped by 1-km grid square and geology) compared with data for Oslofjord region and N. Ireland

0

50

100

150

200

250

300

0 1 2 3 4 5 6 7 8

estimated U mg/kg in topsoil

GM

indo

or r

adon

Bq

m-3

DINLM

DINMDMIX

TRIMD

N. Ireland

Oslofjord

Linear (DINLM)

Linear (DINMDMIX)

Linear (TRIMD)

Linear (N. Ireland)

Linear (Oslofjord)


MODELS BASED ON SOIL CHEMISTRY DATA

U (mg/kg) in <2mm fraction of topsoil determined by XRF


Figure 3: Least squares regression models between U in <2mm fraction of topsoils and GM indoor radon (data grouped by 5-km grid square and geology; dashed lines have intercepts set to 5 Bq m-3)

0

50

100

150

200

250

0 1 2 3 4 5

U mg/kg

GM

indo

or r

adon

(Bq

m-3

)

DINLM

INOLMST

INONS

WESMDMIX

Linear (DINLM)

Linear (DINLM)

Linear (INOLMST)

Linear (INONS)

Linear (INONS)

Linear (WESMDMIX)


MODELS BASED ON SOIL GAS RADON DATA

Geometric mean of soil gas radon concentrations (Bq/L) grouped by geology and grid square


Figure 4: GM soil gas radon vs. GM indoor radon (data grouped by 1-km or 5-km grid square and geology)

0

50

100

150

200

250

300

350

0 50 100 150 200 250 300 350GM Soil gas radon (kBq/m3)

GM

Ind

oor

rado

n (B

q/m

3)

BGS Derby-Notts Carb-Perm1km grouping

BGS Derbyshire Carb Lmst1km grouping (2002-04 data)

BGS Derby-Notts Carb-Perm5km grouping

BGS INONS 1km grouping


Linear (BGS Derby-NottsCarb-Perm 1km grouping)

Linear (BGS Derbyshire CarbLmst 1km grouping (2002-04data))

Linear (BGS Derby-NottsCarb-Perm 5km grouping)

Linear (BGS INONS 1kmgrouping)

Linear (BGS INONS 5kmgrouping)


Figure 5: GM soil gas radon vs. GM indoor radon (data grouped by 1-km or 5-km grid square and geology; intercepts set at 5 Bq m-3)

0

50

100

150

200

250

300

350

400

0 50 100 150 200 250 300 350GM Soil gas radon (kBq/m3)

GM

Indo

or ra

don

(Bq/

m3)

BGS Derby-Notts Carb-Perm 1km grouping

BGS Derbyshire Carb Lmst 1km grouping(2002-04 data)




Linear (BGS Derby-Notts Carb-Perm 1kmgrouping)

Linear (BGS Derbyshire Carb Lmst 1kmgrouping (2002-04 data))

Linear (BGS Derby-Notts Carb-Perm 5kmgrouping)

Linear (BGS INONS 1km grouping)

Linear (BGS INONS 5km grouping)


Figure 5a: Least squares regression models between GM and AM indoor radon (data grouped by 5-km grid square and geology)


Figure 6: Soil gas radon vs. estimated arithmetic mean indoor radon linear regression models for England, Czech Republic and Germany

y = 5.878x + 49.615

y = 2.5749x + 18.229

y = 2.4124x + 87.677

y = 3.6932x + 104.97y = 1.5623x + 123.97

y = 0.9771x + 20.5

y = 0.2786x + 35.429

y = 4.9769x + 35.473

0

100

200

300

400

500

600

700

800

0 50 100 150 200 250 300 350Soil gas radon (kBq/m3)

Arit

hmet

ic a

vera

ge g

roun

dflo

or in

door

rad

on (

Bq/

m3)

Czech; Barnet et al 2008 Tab. 11-2 data

Czech Republic Barnet et al., 2002

Czech Republic: Barnet et al., 2004


BGS Derbyshire Carb Lmst 2002-2004

German NUTS5 units;Barnet et al 2006

Kemski et al., 2009

N. Bohemian Massif; Barnet et al., 2010

Linear (Czech; Barnet et al 2008 Tab. 11-2 data)

Linear (Czech Republic Barnet et al., 2002)

Linear (Czech Republic: Barnet et al., 2004)

Linear (BGS Derby-Notts Carb-Perm 1km grouping)

Linear (BGS Derbyshire Carb Lmst 2002-2004)

Linear (German NUTS5 units;Barnet et al 2006 )

Linear (Kemski et al., 2009)

Linear (N. Bohemian Massif; Barnet et al., 2010)


Indoor radon: soil gas radon ratios•UK (GM) models similar to arithmetic mean SG Czech data

•Both significantly higher than German models based Maximum SG data

•Different house national and regional characteristics will also impact on IR/SG ratios

•Older buildings with ‘leaky’ floors are likely to have higher IR/SG ratios

•Příbram (Czech Rep) has higher IR/SG ratio than average; lower ratios for magmatic rocks, medium ratios for metamorphic, highest ratios for sedimentary rocks.


Figure 7: Soil gas radon vs. estimated arithmetic mean indoor radon linear regression models for England, Italy and Croatia

y = 3.6932x + 104.97

y = 1.5623x + 123.97

y = 2.02x + 69.92

y = 1.3008x - 40.168

y = 4.2388x - 48.496

y = 6.559x - 75.825

0

100

200

300

400

500

600

700

800

0 50 100 150 200 250 300 350Soil gas radon (kBq/m3)

Ari

thm

etic

ave

rage

gro

undf

loor

indo

or ra

don

(Bq/

m3)


BGS Derbyshire Carb Lmst 2002-2004


Italy - lower

Italy - higher

Croatia


Linear (BGS Derbyshire Carb Lmst 2002-2004)


Linear (Italy - lower)

Linear (Italy - higher)

Linear (Croatia)


CONCLUSIONS• ASSUMPTIONS: linear relationship between surrogates for

source of radon (e.g. eU or U) and radon in homes• DATA DISTRIBUTIONS : eU and soil U normally

distributed (therefore use AM for modelling); soil gas and indoor radon log-normally distributed (use GM)

• MEASUREMENT UNCERTAINTIES: impact on slopes and intercepts of LS regression models; slopes generally steeper for fixed intercepts (5 Bq m-3 in UK)

• PERMEABILITY OF THE GROUND: Radon concentrations in houses generated from specific U or soil gas radon concentrations generally increase with the permeability of the ground

• GEOGENIC RADON MAP OF EUROPE: further research required to generate robust empirical relationships for the wide range of numerical input data types (and also for categorical data types).

Documents

Soil uranium, soil gas radon and indoor radon empirical ... · Don Appleton (BGS) and Jon Miles (HPA) ... 0 1 2 3 4 5 6 7 8 ... • ASSUMPTIONS: linear relationship between surrogates