Canadian Nuclear Safety Commission Commission canadienne de sûreté nucléaire Canadian Benthic Data Set Steve Mihok EMRAS II Meeting, Vienna, Austria January 27, 2010 nuclearsafety.gc.ca 1
Thompson et al. (2005) 12 contaminants, As, Cr, Cu, Mo, Ni, Pb, Se, U, V,, Po-210, (N=1,020-2,269) Uranium mining regions with co-located benthos sampling & organic depositional sediments 132 Ontario & Saskatchewan sites 190 genera and/or species Criteria/methods follow Persaud et al. (1992) as used for Ontario LEL / SEL guidelines (As, Cr, Cu, Ni, Pb) 90 th percentile SSLC for each taxon 5 th percentile LEL, 95 th percentile SEL calculated No curve fitting, no dose calculations
Original Methods Weighted percentile as in Persaud et al. (1992), also calculated closest observation percentile (SPSS) Weighted value typically higher Uranium LEL 104 vs 32, SEL 5,874 vs 3,410 ug/g Six data selection criteria Minimum of 10 sites per taxon, lost considerable data Concentration range 2 orders of magnitude (V, Cr x) Spatial range (35 reference sites, 97 contaminated) Mainly benthic species (81% defined as infaunal) Minimum of 20 SSLCs for LEL/SEL calculation Data mean ~30 for SSLCs, LEL/SEL range N=28-59
Interpretation LEL/SEL values highly sensitive to exact percentile calculation method, large effect on Uranium values Good predictions of no impacts with weighted LELs, Poor predictions for many SELs (21 sites) Exploratory data analysis with multiple contaminant indices or data subsets was not informative (As & Ni associated with uranium deposits in Saskatchewan) Radionuclide thresholds differ from ERICA model predictions of potential impacts, Why? Hardly any data on contaminant concentrations in benthos vs sediments, just a few taxa sampled
Benthos - Radiation vs Metals?? % of Community Control Umine site? Sediments Bq / g dw Po-210 LEL range Thompson et al. 2005 0.1 0.6 0.5-0.9 0.6 0.8 Benthos equivalent at 10 µgy/h (ERICA) 0.6 80 600
Example of Uranium SSLC s URANIUM URANIUM N=56 Taxa LEL (ug/g) SEL (ug/g) Ont+Sask + Bancroft Intuitive Weighted 32.0 104.4 3410.0 5874.1 Mean SSLC (ug/g) SD SSLC 693.5 1319.8 1064.9 1683.2 Median sample size for SSLC Mean sample size for SSLC 21 32
Discussion: Dose Calculations Calculations, % Dose Data Issues Notes Approach I E % U-238 Nat uran 49% 11-0 96% Abundance DCF Th-234 =U-238? 0-26 X = Parent? External U-234 Nat uran 49% 12-0 96% Abundance DCF Th-230 =? 10-0 X = Daughter? Measured 20-1 99% Rn-222 @30%? Ra 24-59 X Retention % CRITICAL = % 0-1 74% Site-specific? Bi-210 = 0-13 X Po-210 = % 23-0 70% Site-specific? CRITICAL U-235 (?) Nat uran 2% n/a 96% Abundance DCF I-E example Greer Lake 2004
Greer Lake 2004 PSL2 Chironomid Measured U-238 / Th-230 at equal activity Ra, Pb, Po at 2x U / Th values, Radon estimated @30% Alpha RBE 40, Amiro (1997) DCFs [Tailings present in lake] Rn-222 = 30% Ra Bi-210 (=Pb- 210) Rn-222 = 30% Ra Po-210 (=) Bi-210 (=Pb- 210) Th-230 U-234 U-238 Th-234 (=U- 238) Th-230 U-234 U-238 Th-234 (=U- 238) Po-210 (=)? Internal Dose 2,538 ugy/h External Dose 36 ugy/h
Reference Lakes - Not in equilibrium Schmoo Lake 2002 Po/Pb = 2x U-238, 10x Th-230, 6x Fulton Lake 2004 Po/Pb = 4x U-238 / Th-230 2x Bi-210 (=Pb- Rn-222 210) = 30% Ra Po-210 (=Pb- 210) Bi-210 (=Pb- 210) Rn-222 = 30% Ra Po-210 (=) Th-230 U-234 U-238 Th-234 (=U- 238) U-238 Th-230 U-234 Th-234 (=U- 238) Internal Dose 12.1 ugy/h (alpha RBE of 40) Internal Dose 2.6 ugy/h
Decisions (May need a Radon DCF?) U-238 Th-234 U-234 Th-230 Rn-222 Bi-210 Po-210 U-235 PSL2 Internal Calc Alpha RBE 40, Amiro DCFs, generic benthos = Natural Uranium SA? Not included Included in U-238 = daughter = Measured @30% Ra (~ vert bone) Measured or = radon = Measured or = radon Included in U-238 ERICA Tool Internal + External? RBE = 10, DCFs for insect larvae & bivalve mollusc ~ same? 49% nat uranium (DCFs vary ±) = daughter or parent - Decision? 49% nat uranium = daughter or parent - Decision? = Measured or parent? % retained? Decision critical ~30% missing, calculated @~2x? =, not critical to dose? ~30% missing, calculated @~2x? 2% nat uranium
Discussion: Improving SSLCs? Curve-fitting to obtain SSLCs Chironomus - Arsenic (N=97) Frequency Cumulative % Even this large Frequency 20 18 16 14 12 10 8 6 4 2 0-0.7-0.3 0.1 0.5 0.9 1.3 1.7 2.1 2.5 2.9 More 120% 100% 80% 60% 40% 20% 0% data set is not a good fit to log distribution? 90 th percentile is in the flat part of CDF and may remain as poorly defined? log Sediment concentration
Statistical Interpretation Curve fitting to obtain LEL/SEL thresholds (5%) will also be affected by small sample sizes Bootstrapping for confidence intervals, what to do about outliers (e.g. very high Uranium in Link Lakes) Sensitivity Analysis (sites, SSLCs, taxonomic groups) Multivariate analyses for interactions among metals and radionuclides, fingerprints for severe effects? Prospects for more data Objectives to be defined Path forward on data / analysis J. Garnier LaPlace
A question from WG 4 on Tuesday The Biota modeling group will be estimating dose to fish and benthic invertebrates at Beaverlodge for the next mid-term EMRAS meeting. Is there any interest in modeling population effects, e.g. for lake whitefish and/or fingernail clams (Pisidium spp.)?