radon in groundwater - Mark- och vattenteknik - KTH

More documents

Recommendations

Info

Radon in groundwater - Influencing factors and prediction methodology for a Swedish environmentStatistical analyses of data - InterpretationsUnivariate analysis using the method ofKruskal-Wallis ANOVA by ranks was successfulin evaluating the relative influence ofthe subclasses of factors investigated inrelation to the radon concentration ingroundwater. Compared to visual data mining,quantitative units (median radon concentrations)could be computed for thevarious classes (Table 2), as a result of whichit was easier to interpret the factors. Classselection was based on expert judgementand the different median values were computedwith an appreciable significance level.For the bedrock variable, granitic rocks wereassociated with a much higher radon concentration(540 Bq/l) than the other rocktypes. This observation tallies very well withprevious studies on radon and its relationshipwith geological materials (Choubey andRamola, 1997; Knutsson and Olofsson,2002; Przylibski et al., 2004). Granite containson average a higher concentration ofuranium, the parent element of radon, and inSweden, granitic rocks can contain morethan 5 ppm of uranium (Åkerblom andLindgren, 1997). The content of uranium inbedrock computed by flight measurementsprovided a rough prediction for radon concentrationsin groundwater. As can be observedin Table 2, radon concentration ingroundwater increases with increasing uraniumcontent in the bedrock. However, ananomaly was encountered for the highesturanium class (> 8 ppm), since the observedradon concentration was low. One possibleexplanation could be that the uranium mineralsare not always located very near togroundwater-bearing fractures and hencecannot contribute to increasing the radonconcentration in the groundwater.There exists a relationship between thealtitude of the well and the correspondingradon concentration in groundwater. Radonconcentrations were relatively higher at lowaltitudes than at high altitudes. This observationcould be clarified by the fact that anyradon emanated at high altitude flows togetherwith the groundwater by convectionto lower terrains, provided the travellingdistance is short and the radon does notdecay (within 3.8 days). Alternatively, itcould instead be argued that parent elementsof 222 Rn ( 238 U and 226 Ra) leach into groundwaterat high altitudes, get transported bythe water and re-deposit on the surfaces ofthe fracture at lower altitudes and continueto decay from these spots to increase concentrationof radon in groundwater. It wasdifficult to interpret the factor of relativealtitude within 100 m, since varying radonconcentrations were observed at differentslopes.Another result related to geological propertiesrevealed that the type of soil overlyingthe bedrock could have some importance.One plausible explanation could be theinfluence of the permeability property of thesoils. Permeable soils allow infiltration readilyand this leads to a dilution of the radonconcentration in groundwater in the subsurface.It could therefore be observed thatsand and gravel are associated with slightlylower radon concentrations than clay andsilt. Another observation made possiblethrough univariate statistical analysis relatesthe land use factor to the radon concentrationof the wells. The original land use datawere classified into summer houses, permanenthouses and other land uses, such asforests and fields. In the latter a small waterusage was assumed and it was observed thatradon concentrations had a tendency to berelatively low when water was in continuouscirculation, such as in permanent houses. Asimilar observation was made by Knutssonand Olofsson (2002). Regarding the factorof distance from a fracture line or zone, theresults of univariate statistical analysis weredifficult to interpret. On the other hand, theinfluence of that factor could be observed invisual data mining on 3D images. Fracturesmight be filled with minerals containinguranium and therefore the nearer a well islocated to a fracture zone, the higher theprobability of the groundwater containing anincreased concentration of radon. The reasonwhy statistical analysis was less successfulat showing the relationship might beattributed to the class selections of distance.13
Kirlna Skeppström TRITA LWR.LIC 2032The assumed fracture zones are also givenby the lineament maps and the position ofthe zones and their water conducting propertiesare uncertain.Results of principal component analysis(PCA) are presented in Table 3. The methodanalyses only quantitative factors and inorder to explain 86% of the variance in theoriginal dataset, five components (PC) wereneeded. Although it was possible to see thevarious loadings coming from each factor ina component, it was often difficult to interpretthese components and the interpretationwas to some extent subjective.PC 1 was principally loaded by the factors ofaltitude and the relative altitude within100 m and 500 m. This component wasinterpreted as a property that guides thetransport of radon in water. The factor ofaltitude and its derivatives indirectly providedinformation about the flow paths andflow speed of groundwater. The secondprincipal component (PC 2) was moderatelyloaded by uranium content in the bedrock,distance to fracture zone and radon concentrationin groundwater. PC 2 was interpretedas an enrichment of uranium in the fracturezones. Uranium content in the bedrock,radon concentrations in groundwater anddifference in altitude carried the highestloading in PC 3. That component might bean indicator of radon and uranium transportin groundwater, as a result of the heightdifference. PC 4 was loaded by a singlefactor and was probably related directly tothe fracture fillings. The fifth principal component(PC 5) was retained as the emanationof radon from uranium, since the loadingscame from these parameters.Table 2: Results of Kruskal-Wallis ANOVAby ranksFactorsSoil at well locationBedrock at welllocationLand use at welllocationUranium content(ppm) in bedrockAltitude (m a.s.l)Relative altitudewithin 100 m (%)Distance fromfracture (m)Clay/SiltTillRockClassesSand/gravelMetasedimentaryFelsic gneissesMafic rocksGraniteSummer housesPermanent housesFields, forests0-22-44-66-8> 80-2020-4040-60>600-2020-4040-6060-8080-100200RadonMedianvalue (Bq/l)280252200224150240150540190163370190260495410195240240147125200240250250210220210205230Table 3: Principal component loadings for quantitative factors . Significant loadings are marked inboldPC 1 PC 2 PC 3 PC 4 PC 5Radon content in groundwater (z 1 ) -0.093 0.493 0.427 0.159 0.723Altitude of the well (z 2 ) 0.586 -0.114 0.129 -0.258 0.235Relative altitude within 100 m (z 3 ) 0.432 0.280 -0.375 0.331 -0.079Relative altitude within 500 m (z 4 ) 0.608 0.238 -0.116 -0.014 -0.007Difference in altitude (z 5 ) 0.246 -0.538 0.354 -0.325 0.146Uranium content in rock (z 6 ) 0.106 0.370 0.666 -0.045 -0.627Distance to fracture zone (z 7 ) -0.140 0.428 -0.280 -0.831 0.01114
Page 2 and 3: Kirlna Skeppström TRITA LWR.LIC 20
Page 9: Kirlna Skeppström TRITA LWR.LIC 20
Page 12 and 13: Radon in groundwater - Influencing
Page 30: Radon in groundwater - Influencing

radon in groundwater - Mark- och vattenteknik - KTH

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?