Tracers: RT Isotope Hydrology Shortcourse Residence Time Approaches using Isotope Tracers Prof. Jeff McDonnell Dept. of Forest Engineering Oregon State University 1 McGuire, OSU Tracers: RT Outline Day 1 Morning: Introduction, Isotope Geochemistry Basics Afternoon: Isotope Geochemistry Basics ‘cont, Examples Day 2 Morning: Groundwater Surface Water Interaction, Hydrograph separation basics, time source separations, geographic source separations, practical issues Afternoon: Processes explaining isotope evidence, groundwater ridging, transmissivity feedback, subsurface stormflow, saturation overland flow Day 3 Morning: Mean residence time computation Afternoon: Stable isotopes in watershed models, mean residence time and model strcutures, two-box models with isotope time series, 3-box models and use of isotope tracers as soft data Day 4 Field Trip to Hydrohill or nearby research site © Oregon State University 2 McGuire, OSU Tracers: RT © Oregon State University How these time and space scales relate to what we have discussed so far 3 Bloschel et al.,McGuire, 1995 OSU Tracers: RT © Oregon State University This section will examine how we make use of isotopic variability 4 McGuire, OSU Tracers: RT Outline What is residence time? How is it determined? modeling background Subsurface transport basics Stable isotope dating (18O and 2H) Models: transfer functions Tritium (3H) CFCs, 3H/3He, and © Oregon State University 85Kr 5 McGuire, OSU Tracers: RT Residence Time Mean Water Residence Time (aka: turnover time, age of water leaving a system, exit age, mean transit time, travel time, hydraulic age, flushing time, or kinematic age) tw=Vm/Q For 1D flow pattern: tw=x/vpw where vpw =q/f Mean Tracer Residence Time tC (t )dt Residence time distribution I tt 0 C (t )dt I 0 © Oregon State University g (t ) CI (t ) C (t )dt CI Q / M I 0 6 McGuire, OSU Tracers: RT Why is Residence Time of Interest? It tells us something fundamental about the hydrology of a watershed Because chemical weathering, denitrification, and many biogeochemical processes are kinetically controlled, residence time can be a basis for comparisons of water chemistry Vitvar & Burns, 2001 © Oregon State University 7 McGuire, OSU Tracers: RT Tracers and Age Ranges Environmental tracers: © Oregon State University added (injected) by natural processes, typically conservative (no losses, e.g., decay, sorption), or ideal (behaves exactly like traced material) 8 McGuire, OSU Tracers: RT Modeling Approach Lumped-parameter models (black-box models): System is treated as a whole & flow pattern is assumed constant over modeling period Used to interpret tracer observations in system outflow (e.g. GW well, stream, lysimeter) Inverse procedure; Mathematical tool: The convolution integral © Oregon State University 9 McGuire, OSU Tracers: RT Convolution y (t ) h(t ) x(t )d A convolution is an integral which expresses the amount of overlap of one function h as it is shifted over another function x. It therefore "blends" one function with another It’s frequency filter, i.e., it attenuates specific frequencies of the input to produce the result Calculation methods: Fourier transformations, power spectra Numerical Integration © Oregon State University 10 McGuire, OSU Tracers: RT The Convolution Theorem f (t ) * g (t ) F ( )G( ) { f (t ) * g (t )} f ( x) g (t – x) dx exp( i t ) dt Proof: f ( x) g (t x) exp(–i t ) dt dx Trebino, 2002 © Oregon State University f ( x){G ( exp(–i x)} dx Y()=F()G() and |Y()|2=|F()| 2 |G()| 2 f ( x) exp(–i x) dxG ( F ( G ( We will not go through this!! 11 McGuire, OSU Tracers: RT Convolution: Illustration of how it works g() = e -a x() Step e -(-a g(-) Folding 1 e -a(t- g(t-) t Displacement 2 x()g(t-) Multiplication 3 Integration 4 t y(t) t © Oregon State University Shaded area t 12 McGuire, OSU Tracers: RT Example: Delta Function f (t ) t a) f (t u ) (u – a) du f (t a) Convolution with a delta function simply centers the function on the delta-function. This convolution does not smear out f(t). Thus, it can physically represent piston-flow processes. Modified from Trebino, 2002 © Oregon State University 13 McGuire, OSU Tracers: RT Matrix Set-up for Convolution = length(x) = [length(x)+length(h)]-1 =S = x(t)*h y(t) =0 © Oregon State University 14 McGuire, OSU Tracers: RT Similar to the Unit Hydrograph Precipitation Excess Precipitation Infiltration Capacity Excess Precipitation Hydrographs for Event 2500 Time Flow 2000 1500 1000 500 Tarboton 0 0 © Oregon State University 1 Time(hrs) 2 3 15 McGuire, OSU Tracers: RT Instantaneous Response Function Excess Precipitation P(t) Unit Response Function U(t) 2500 2000 1500 1000 500 0 0 1 2 3 4 Event Response Q(t) 2500 2000 1500 Q ( t ) P ( ) U ( t ) d 1000 500 0 0 1 2 3 Tarboton © Oregon State University 16 McGuire, OSU Tracers: RT Subsurface Transport Basics 17 McGuire, OSU Tracers: RT Subsurface Transport Processes Advection Dispersion Sorption Transformations Modified from Neupauer & Wilson, 2001 © Oregon State University 18 McGuire, OSU Tracers: RT Advection Solute movement with bulk water flow t=t1 t2>t 1 t3>t 2 FLOW Modified from Neupauer & Wilson, 2001 © Oregon State University 19 McGuire, OSU Tracers: RT Subsurface Transport Processes Advection Dispersion Sorption Transformations Modified from Neupauer & Wilson, 2001 © Oregon State University 20 McGuire, OSU Tracers: RT Dispersion Solute spreading due to flowpath heterogeneity Modified from Neupauer & Wilson, 2001 © Oregon State University FLOW 21 McGuire, OSU Tracers: RT Subsurface Transport Processes Advection Dispersion Sorption Transformations Modified from Neupauer & Wilson, 2001 © Oregon State University 22 McGuire, OSU Tracers: RT Sorption Solute interactions with rock matrix FLOW Modified from Neupauer & Wilson, 2001 © Oregon State University t=t1 t2>t1 23 McGuire, OSU Tracers: RT Subsurface Transport Processes Advection Dispersion Sorption Transformations Modified from Neupauer & Wilson, 2001 © Oregon State University 24 McGuire, OSU Tracers: RT Transformations Solute decay due to chemical and biological reactions MICROBE CO2 Modified from Neupauer & Wilson, 2001 © Oregon State University t=t1 t2>t1 25 McGuire, OSU Tracers: RT Stable Isotope Methods 26 McGuire, OSU Tracers: RT Stable Isotope Methods and 2H in precipitation at temperate latitudes Seasonal variation of 18O Variation becomes progressively more muted as residence time increases These variations generally fit a model that incorporates assumptions about subsurface water flow Vitvar & Burns, 2001 © Oregon State University 27 McGuire, OSU Seasonal Variation in 18O of Precipitation air temperature (°C) 18O (per mil SMOW) Tracers: RT © Oregon State University 0 Neversink watershed, 1993 - 1996 -5 -10 -15 -20 20 10 0 -10 Jan-93 Vitvar, 2000 Jan-94 Jan-95 Jan-96 28 McGuire, OSU Tracers: RT Seasonality in Stream Water Deines et al. 1990 © Oregon State University 29 McGuire, OSU Tracers: RT Example: Sine-wave Oxygen-18 (per mil) -2 Cin(t)=A sin(t) Cout(t)=B sin(t+j) 6 g(t) 0 -3 4 Mean = 235 d 2 Precipitation or recharge signal -4 x 10 0 0 1000 2000 Time [days] -6 -8 -10 Streamflow signal -12 -14 1200 © Oregon State University T=-1[(B/A)2 –1)1/2 1300 1400 1500 Time [days] 1600 1700 1800 30 McGuire, OSU Tracers: RT © Oregon State University Convolution Movie 31 McGuire, OSU Tracers: RT Transfer Functions Used for Residence Time Distributions 32 McGuire, OSU Tracers: RT Common Residence Time Models -3 0.012 6 D/vx=5 D/vx=3.5 D/vx=2 D/vx=1 D/vx=0.1 D/vx=0.2 D/vx=0.5 D/vx=0.01 D/vx=0.05 D/vx=0.005 0.01 g(t) 0.008 0.006 3 0.002 1 1 1.5 Normalized time (t/T) © Oregon State University 4 2 0.5 2 eta=1 eta=1.25 eta=1.5 eta=1.75 eta=2 eta=2.25 eta=2.5 eta=2.75 eta=3 5 0.004 0 0 x 10 0 0 0.5 1 1.5 2 Normalized time (t/T) 33 McGuire, OSU Tracers: RT Piston Flow (PFM) Assumes all flow paths have transit time All water moves with advection Represented by a Dirac delta function: 1 0.8 g (t ) (t T ) g(t) 0.6 0.4 0.2 0 0 1 2 3 4 t/T © Oregon State University 34 McGuire, OSU Tracers: RT Exponential (EM) Assumes contribution from all flow paths lengths and heavy weighting of young portion. Similar to the concept of a “well-mixed” system in a linear reservoir model 0.16 0.14 0.12 g (t ) T 1 exp(t / T ) g(t) 0.1 0.08 0.06 0.04 0.02 0 0 © Oregon State University 2 4 6 t/T 8 10 12 35 McGuire, OSU Tracers: RT Dispersion (DM) Assumes that flow paths are effected by hydrodynamic dispersion or geomorphological dispersion Arises from a solution of the 1-D advection-dispersion equation: C 2C C t D x 2 v x 0.01 0.008 4D p t g (t ) T g(t) 0.006 0.004 1/ 2 2 t T 1 t exp 1 T 4 D p t 0.002 0 0 © Oregon State University 2 4 6 t/T 8 10 36 McGuire, OSU Tracers: RT Exponential-piston Flow (EPM) Combination of exponential and piston flow to allow for a delay of shortest flow paths h ht g (t ) exp h 1 T T 0.2 g(t) 0.15 for t T (1-h1, and g(t)=0 for t< T (1-h-1) 0.1 0.05 Piston flow = 1 0 0 2 4 6 8 10 1 h 12 t/T © Oregon State University 37 McGuire, OSU Tracers: RT Heavy-tailed Models Gamma t 1 t / g (t ) exp ( ) Exponentials in series t t 1 g (t ) exp exp T1 T2 T1 T2 © Oregon State University 38 McGuire, OSU Tracers: RT Exit-age distribution (system response function) Unconfined aquifer EM: g(t’) = 1/T exp(-t‘/T) Confined aquifer PFM: g(t’) = (t'-T) DM DM EM EM EPM EM PFM PFM Maloszewski and Zuber Kendall, 2001 © Oregon State University 39 McGuire, OSU Exit-age distribution (system response function) cont… Tracers: RT Partly Confined Aquifer: EPM: g(t’) = h/T exp(-ht'/T + h-1) g(t’) = 0 for t‘≥T (1 - 1/h) for t'< T (1-1/ h) DM Kendall, 2001 © Oregon State University Maloszewski and Zuber 40 McGuire, OSU Tracers: RT 0 g(t) -10 0.15 0.1 -15 0.05 -20 0 0 MRT = 12 months D/vx = 0.3 0.2 g(t) -5 -10 0.15 0.1 -15 0.05 -20 0 0 -10 -15 -20 0 MRT = 6 months D/vx = 0.05 0.2 -5 0.15 0.1 0.05 20 40 60 Time (months) © Oregon State University MRT = 6 months D/vx = 0.3 0.2 -5 g(t) O-18 (per mil) O-18 (per mil) O-18 (per mil) Dispersion Model Examples 80 0 0 20 40 60 80 Time (months) 41 McGuire, OSU Residence Time Distributions can be Similar Tracers: RT 0.08 EPM: 21% Piston MRT = 10.5 mon. g(t) 0.06 0.04 DM: Dp = 0.36 MRT = 8.5 mon. EPM 12% Piston MRT = 9.5 mon. DM: Dp = 0.27 MRT = 10.5 mon. 0.02 0.00 0 © Oregon State University 500 1000 Time (d) 1500 0 500 1000 1500 Time (d) 42 McGuire, OSU Tracers: RT Uncertainty Freq 30 0.15 10 0 10 FittedTransfer Functions 12 16 30 0.1 0.05 20 10 0 140 0 0 14 Piston Flow % Freq Function weighting [g(t)] 0.2 20 160 180 MRT 0.5 1 1.5 2 2.5 Normalized time [t/T] 3 3.5 4 -8 tracer content Convolution Output Obs -9 -10 Simulation Results with Optimized Parameters -11 40 45 50 55 60 65 70 75 time © Oregon State University 43 McGuire, OSU Tracers: RT © Oregon State University Identifiable Parameters? 44 McGuire, OSU Tracers: RT Review: Calculation of Residence Time Simulation of the isotope input – output relation: t C (t ) g (t ) Cin (t ) d 0 Calibrate the function g(t) by assuming various distributions of the residence time: 1. 2. 3. © Oregon State University Exponential Model Piston Flow Model Dispersion Model 45 McGuire, OSU Tracers: RT Input Functions t C (t ) g (t ) Cin (t ) d 0 Must represent tracer flux in recharge Weighting functions are used to “amount-weight” the tracer values according recharge: mass balance!! Methods: Winter/summer weighting: in (t ) N i Pi N P i 1 Lysimeter outflow General equation: C (t ) © Oregon State University g ( )w(t ) 0 in (t )d g ( )w(t )d 0 i Cin Cin i i where w(t) = recharge weighting function 46 McGuire, OSU Tracers: RT Models of Hydrologic Systems Model 1 Cin Cout Model 3 Cin Model 2 Upper reservoir 1- Cout g Direct runoff Cin 1- g 1- g g Cout Lower reservoir Maloszewski et al., 1983 © Oregon State University 47 McGuire, OSU Tracers: RT Soil Water Residence Time Stewart & McDonnell, 2001 © Oregon State University 48 McGuire, OSU Tracers: RT Rietholzbach watershed, Switzerland Mean baseflow residence time = 12.5 mo -10 -11 18O measured values -12 dispersion model Rietholzbach watershed, Switzerland Mean groundwater residence time = 28 mo -9 -10 -11 -12 measured values -13 -14 -8 (per mil SMOW) -9 18O (per mil SMOW) -8 Example from Rietholzbach -13 exp/piston-flow model 1994 1995 1996 1997 -14 dispersion model exponential model 1994 1995 1996 1997 Vitvar, 1998 © Oregon State University 49 McGuire, OSU Tracers: RT Model 3… Stable deep signal Uhlenbrook et al., 2002 © Oregon State University 50 McGuire, OSU Tracers: RT How residence time scales with basin area 172 degrees E New Zealand 42 degrees S Figure 1 © Oregon State University 51 McGuire, OSU Tracers: RT Contour interval 10 meters Digital elevation model and stream network Figure 2 © Oregon State University 52 McGuire, OSU 1.0 Tracers: RT M15 catchment (2.6 ha) 0.5 Median sub-catchment size = 1.2 ha 0.0 Frequency 1.0 1 10 K catchment (17 ha) 0.5 100 Median sub-catchment size = 8.2 ha 0.0 1.0 1 10 PL14 catchment (80 ha) 0.5 100 Median sub-catchment size = 3.9 ha 0.0 1.0 1 10 100 Bedload catchment (280 ha) 0.5 Median sub-catchment size = 3.2 ha 0.0 Figure 3 1 10 100 Sub-catchment size ha © Oregon State University 53 McGuire, OSU Mean tritium age years Tracers: RT K (17 ha) 2 Bedload (280 ha) PL14 (17 ha) M15 (2.6 ha) 1 0 2 4 6 8 Median sub-catchment size ha Figure 4 © Oregon State University 54 McGuire, OSU Tracers: RT RIF -7 Low -3.5 0 High Scale 500 m © Oregon State University 55 McGuire, OSU Tracers: RT Determining Residence Time of Old(er) Waters 56 McGuire, OSU Tracers: RT What’s Old? No seasonal variation of stable isotope concentrations: >4 to 50 years Methods: Tritium (3H) 3H/3He CFCs 85Kr © Oregon State University 57 McGuire, OSU Tracers: RT Tritium Historical tracer: 1963 bomb peak of 3H in atmosphere 1 TU: 1 3H per 1018 hydrogen atoms Slug-like input 36Cl is a similar tracer Similar methods to stable isotope models Half-life (l) = 12.43 Tritium Input © Oregon State University 58 McGuire, OSU Tracers: RT Tritium (con’t) Piston flow (decay only): tt=-17.93[ln(C(t)/C0)] Other flow conditions: t C (t ) Cin (t ) exp(lt ' ) g (t t ' )dt' 0 © Oregon State University Manga, 1999 59 McGuire, OSU Tracers: RT Deep Groundwater Residence Time Spring: Stollen t0 = 8.6 a, PD = 0.22 3H-Input 3H-Input-Bruggagebiet 1000 sim. 3H-Konzentrationen 3 H-Messungen mit analyt. Fehler H-Input im Bruggagebiet H [T.U.] 3 20 3 100 15 3 H-Konzentrationen [T.U.] 25 10 1950 1960 1970 1980 1990 2000 10 1992 1993 1994 Zeit [Jahre] Time [yr.] Uhlenbrook et al., 2002 © Oregon State University 1995 1996 1997 1998 1999 Time [yr.] lumped parameter models 60 McGuire, OSU Tracers: RT 3He/3H As 3H enters groundwater and radioactively decays, the noble gas 3He is produced Once in GW, concentrations of 3He increase as GW gets older If 3H and 3He are determined together, an apparent age can be determined: 3 * He 1 tt l ln 3 1 H © Oregon State University 61 McGuire, OSU Tracers: RT Determination of Tritiogenic He Other sources of 3He: 30 Atmospheric age (years) solubility (temp dependent) Trapped air during recharge 20.5 years Radiogenic production ( decay of U/Th20 series elements) 3He/3H 4He and Determined by measuring 10 other noble gases 0 1 5 10 50 Tage (years) Modified from Manga, 1999 © Oregon State University 62 McGuire, OSU Tracers: RT Chlorofluorocarbons (CFCs) CFC-11 (CFCL3), CFC-12 (CF2Cl2), & CFC-13 (C2F3Cl3) long atm residence time (44, 180, 85 yrs) Concentrations are uniform over large areas and atm concentration are steadily increasing Apparent age = CFC conc in GW to equivalent atm conc at recharge time using solubility relationships © Oregon State University 63 McGuire, OSU Tracers: RT 85Kr Radioactive inert gas, present is atm from fission reaction (reactors) Concentrations are increasing worldwide Half-life = 10.76; useful for young dating too Groundwater ages are obtained by correcting the measured 85Kr activity in GW for radioactive decay until a point on the atm input curve is reached © Oregon State University 64 McGuire, OSU Tracers: RT 85Kr (con’t) Independent of recharge temp and trapped air Little source/sink in subsurface Requires large volumes of water sampled by vacuum extraction (~100 L) © Oregon State University 65 McGuire, OSU Tracers: RT Model 3… Uhlenbrook et al., 2002 © Oregon State University 66 McGuire, OSU Tracers: RT © Oregon State University Large-scale Basins 67 McGuire, OSU Tracers: RT Notes on Residence Time Estimation • 18O and 2H variations show mean residence times up to ~4 years only; older waters dated through other tracers (CFC, 85Kr, 4He/3H, etc.) • Need at least 1 year sampling record of isotopes in the input (precip) and output (stream, borehole, lysimeter, etc.) • Isotope record in precipitation must be adjusted to groundwater recharge if groundwater age is estimated © Oregon State University 68 McGuire, OSU Tracers: RT Class exercise ftp://ftp.fsl.orst.edu/pub/mcguirek/rt_lecture Hydrograph separation Convolution FLOWPC Show your results graphically (one or several models) and provide a short write-up that includes: – Parameter identifiability/uncertainty – Interpretation of your residence time distribution in terms of the flow system © Oregon State University 69 McGuire, OSU Tracers: RT References Cook, P.G. and Solomon, D.K., 1997. Recent advances in dating young groundwater: chlorofluorocarbons, 3H/3He and 85Kr. Journal of Hydrology, 191:245-265. Duffy, C.J. and Gelhar, L.W., 1985. Frequency Domain Approach to Water Quality Modeling in Groundwater: Theory. Water Resources Research, 21(8): 11751184. Kirchner, J.W., Feng, X. and Neal, C., 2000. Fractal stream chemistry and its implications for contaminant transport in catchments. Nature, 403(6769): 524527. Maloszewski, P. and Zuber, A., 1982. Determining the turnover time of groundwater systems with the aid of environmental tracers. 1. models and their applicability. Journal of Hydrology, 57: 207-231. Maloszewski, P. and Zuber, A., 1993. Principles and practice of calibration and validation of mathematical models for the interpretation of environmental tracer data. Advances in Water Resources, 16: 173-190. Turner, J.V. and Barnes, C.J., 1998. Modeling of isotopes and hydrochemical responses in catchment hydrology. In: C. Kendall and J.J. McDonnell (Editors), Isotope tracers in catchment hydrology. Elsevier, Amsterdam, pp. 723-760. Zuber, A. and Maloszewski, P., 2000. Lumped parameter models. In: W.G. Mook (Editor), Environmental Isotopes in the Hydrological Cycle Principles and Applications. IAEA and UNESCO, Vienna, pp. 5-35. Available: http://www.iaea.or.at/programmes/ripc/ih/volumes/vol_six/chvi_02.pdf © Oregon State University 70 McGuire, OSU Tracers: RT Outline Day 1 Morning: Introduction, Isotope Geochemistry Basics Afternoon: Isotope Geochemistry Basics ‘cont, Examples Day 2 Morning: Groundwater Surface Water Interaction, Hydrograph separation basics, time source separations, geographic source separations, practical issues Afternoon: Processes explaining isotope evidence, groundwater ridging, transmissivity feedback, subsurface stormflow, saturation overland flow Day 3 Morning: Mean residence time computation Afternoon: Stable isotopes in watershed models, mean residence time and model strcutures, two-box models with isotope time series, 3-box models and use of isotope tracers as soft data Day 4 Field Trip to Hydrohill or nearby research site © Oregon State University 71 McGuire, OSU