14 CO 2 measurement in the NOAA/ESRL Global Cooperative Sampling Network: An update on measurement and data quality Scott Lehman INSTAAR, University of Colorado at Boulder Jocelyn Turnbull, Chad Wolak - NOAA & INSTAAR, Boulder John Miller, Pieter Tans - CIRES & NOAA/ESRL, Boulder John Southon, UC Irvine, California
outline 14 C systematics, sensitivity to F ff and C ff quantification of C ff from 14 CO 2 and CO 2 sampling and measurement protocols measurement repeatability measurement programs representative results NWR surface (relatively clean FT as bg ) CMA vertical profiles (isolation of C ff and C bio ) preliminary results 14 CO 2 intercomparison (J. Miller) future effort
secular trends 1000 800 tree rings Vermunt meas. JFJ fit (m_m) NWRfit (ind) excess 14 C 1000 800 1/λ 14C = 8033 yr! 14 C 600 400 assimilation 600 400 fossil fuel CO 2 Δ = -1000 200 Suess effect 200 0 1900 1920 1940 1960 1980 2000 Δ=Δ 14 C= [ 14 C C ]x 14 C C year ( ) ( ) sample standard 1 0 1000 ratio to standard is δ 13 C corrected for mass dependent fractionation Δ is decay corrected for date of collection data: Levin and Kromer
secular trends excess 14 C 200 180 NWR fit (ind) JFJ fit (m_m) 200 180 160 160 assimilation 140 140! 14 C 120 100 120 100 Suess effect 80 60 80 60 Δ=Δ 14 C= [ 14 C C ]x 14 C C ( ) ( ) sample standard 1 1000 40 40 1985 1990 1995 2000 2005 2010 year dδ obs /dt = ~ -5.5 /yr dδ ff /dt = ~ -10.5 /yr dδ oce /dt + dδ bio /dt = ~0 /yr dδ prod+nuc /dt = ~ +5.7 /yr F ff ( Suess effect ) once again dominating secular trend data: Levin and Kromer analysis: Naegler et al. 2006; Turnbull et al. in press
sensitivity to C ff 1/λ 14C = 8033 yr fossil fuel CO 2 Δ = -1000 (i.e. zero 14 C content) ambient atm. (& other CO 2 sources) Δ = ~ +55 ΔΔ ff-atm C atm = -1055 380ppm = ~-2.8 /ppm - detection of ~1 ppm for recently-added fossil-fuel derived CO 2 (C ff ) requires measurement precision of ~2 - detection is unbiased if other contributions to tropospheric 14 C distribution are small
C ff v. 14 CO 2 in TM5 annual@350 m annual@350 m F ff is 7 GTC/yr global (1.6 GTC/yr US) distributed according to EDGAR2000, model represents all terms influencing tropospheric 14 C budget (except nuclear industry) distribution of C ff dominates Δ 14 CO 2 signal over NH land areas (figures scaled according to mass balance relation of 2.8 /ppm) ΔΔ 14 C CBL up to ~10, implying scientifically meaningful Δ 14 CO 2 precision ~2
quantification of C ff C obs = C bg + C ff + C r + C p Δ obs C obs = Δ bg C bg +Δ ff C ff +Δ r C r +Δ p C p and setting Δ p = Δ bg : C ff = C obs (Δ obs - Δ bg ) C r (Δ r - Δ bg ) - Δ ff - Δ bg Δ ff - Δ bg right hand term is correction to C ff of ~0.2-0.5 ppm, C r and Δ r can be determined independently (i.e. quantifiable) Turnbull et al. 2006 C ff detection effectively limited by quadrature sum uncertainty of two 14 C measurements (Δobs, Δbg) psuedo-lagrangian analysis framework (rel. obs to bg) is itself an uncertainty
sampling ~2-3 L parent air (0.4-0.6 mg C) to glass ( 13 C-proven ) paired NOAA-ESRL network flasks (~2.2L) filled in series (several minutes flush/fill time) NOAA-ESRL PFPʼs (aircraft) (700 cc at 40 psi) sequentially-filled pairs (several minutes flush/fill)
experimental CO 2 extraction via simple cryogenic procedure (following Zhao et al. 1997) graphitization CO 2 + 2H 2 (Fe) (625 C) C + 2H 2 O hi-count AMS measurement (UC-Irvine) repeatability evaluated in control tank clean dry air from NWR (NWT std, 2002.9) process blank evaluated in synthetic air spiked with dead CO 2
measurement UCI AMS MC-SNICS source 40-position carousel 8 primary stds (Ox-I) 2 secondary stds (Ox-II) 4 NWT std control samples (single cylinder) 1 process blank ( 14 C-free synthetic air) 25 authentic samples 1 wheel/10d at current sampling rates w/in network
measurement (future) UCI AMS MC-SNICS source 40-position carousel 8 primary stds (Ox-I) 0 secondary stds (Ox-II)? 2 NWT std control samples 3 NWT 3 control samples (new cylinder) 3 NWT 4 control samples (new cylinder) 1 process blank ( 14 C-free synthetic air) 23 authentic samples? relatively short life time of control tanks (6 yrs on NWT std ) multiple, staggered tanks required for long term surveillance
atom counting typically, 15 acquisition cycles at 50k 14 C events each (n = 750,000 cts.) statistical limit on precision ~1.2 typical single-sample precision after standardization ~1.7 ( i.e. AMS system error of ~1.2 ) repeatability..
1-σ repeatability wheel stdev! 14 C 2 1 0 80 date pres. 4/04 240 260 3/06 280 300 11/08 320 340 360 5 5 NWTstd 1" stdev 4 4 OxI 1" stdev 3 3 2 1 0 80 John Southon, UCI! 14 C! 14 C 78 76 74 72 70 68 66 40 38 36 34 32 30 28 28 26 Ox1 >.4mgC 26 240 260 280 300 320 340 360 wheel no. wheel NWTstd>.4mgC assign uncertainty as the greater of the 1-σ NWT std repeatability or the 1-σ single sample analytical precision 78 76 74 72 70 68 66 40 38 36 34 32 30 NWTstd: since 241: 2.03 since 289: 1.95 NBS OxI: since 241: 1.99 since 289: 1.75 (wheel means normalized)
(in)stability 78 NWTstd, wheel mean & stdev. y = 73.357 + 0.018013x R= 0.24317 78 76 76! 14 CO 2 74 72 74 72 70 70 0 10 20 30 40 50 60 70 wheel order?
CU/NOAA 14 CO 2 measurement sites NWR ( 3523 masl, rel. clean cont. FT) NA aircraft sampling sites ( ) CMA, NHA, BNE, LEF NA tower sites ( ) WKT, LEF, SCT, WGC WGC NWR Asia WLG, UUM, TAP (Lin An, Shangdianzi) Asia North America BNE SCT see Turnbull ICDC poster for recent results
NWR (3523 masl, 40.5 N, 105.6 E) 80 NWR raw clean NWRfit (clean) 80 70 70! 14 CO 2 60 50 60 50 40 40 30 30 2003 2004 2005 2006 2007 2008 2009 date local pollution flagged based on CO (CO a >15 ppb) scrubbed record can be used to provide representative bg meas. in psuedo- Lagrangian analysis framework (as m_m or harmonic fit) [i.e. Turnbull et al. JGR 2007]
3-ht sampling CMA 80 70 CMA NWRfit 3962m 2134m 305m 80 70! 14 CO 2 60 50 late summer 60 50 40 40 30 30 2005 2006 2007 2008 2009 2010 date Δ 14 CO 2 difference signal is large w.r.t. measurement precision data also resolves limitations of analysis framework - for example, enhanced N -> S transport of 14 C-rich air in late summer
isolation of C ff and C bio, CMA above (below) 2400m C obs = C bg + C ff + C bio - NWR fit (ΔC) obs = (ΔC) bg + (ΔC) ff
isolation of C ff and C bio, CMA above (below) 2400m C obs = C bg + C ff + C bio (ΔC) obs = (ΔC) bg + (ΔC) ff includes correction for ΔC r C bio large even in winter C ff detectable year round, even above CBL C ff & C bio from same sample but independently useful: - constrain F ff, F bio - evaluate other C ff tracers
International Atmospheric 14CO2 Inter-comparison to evaluate measurement precision, repeatability, and laboratory inter-comparability to aid integration of regional and global observations of 14CO2
initial participants: EXTRACTION UC-Irvine (US) Scripps (US) CU-INSTAAR (US) CIO - Groningen (Netherlands) CAOS-Tohuku (Japan) ANSTO (Australia) MEASUREMENT UC-Irvine Keck AMS LLNL-CAMS UC-Irvine Keck AMS Groningen-AMS Nagoya-AMS ANSTO progress round two of flask inter-comparison complete interested parties may contact John Miller (NOAA)
preliminary 14 CO 2 intercomparison results A B 2005 Experts Recommendation for lab intercomparability was 1 per mil. Labs 3,4,6 within 2 per mil range for A and B. Blue= Round1; Red= Round2, with error bars indicating reported error bars of measurements Black = unweighted means of both rounds with error bars indicating std. dev. of values. Grey lines = weighted means across all labs. No outliers excluded. from John Miller CIRES/NOAA
future estimates of regional US emissions ratios for various long-lived anthropogenic trace gases (R = T/C ff ) introduction of additional surveillance tanks renewed effort to reduce sources of variance automation of CO 2 extraction (C-Rex) improved handling of primary standard expansion of graphitization capability additional measurement partners continue 14 C ICP, w/ greater frequency THANK YOU