1. Isotopic Evolution of Snowmelt Vicky Roberts Paul Abood Watershed Biogeochemistry 2/20/06

2. Isotopes in Hydrograph Separation Used to separate stream discharge into a small number of sources Oxygen and hydrogen isotopes are widely used because they are components of water and are conservative over short time scales

3. Problem For hydrograph separations involving snowmelt runoff –Some studies assume snowmelt to have a constant  18 O value equal to the average  18 O of the snowpack –  18 O in snowmelt ≠  18 O snowpack

4. Snowmelt Isotopes Snowmelt –Depleted in  18 O early in melting season –Enriched in  18 O later in melting season Why? –Isotopic exchange between liquid water and solid ice as water percolates down the snow column

5. Physical Process At equilibrium, the  18 O of water is less than the  18 O of ice; initial snowmelt has lower  18 O than the snowpack Snowpack becomes enriched in  18 O ; melt from the enriched pack is itself enriched (  18 O )

6. Papers Theory –Feng, X., Taylor, S., and Renshaw, C.E. 2002. Lab –Taylor, S., Feng, X., and Renshaw, C.E. 2002. Field –Taylor, S., Feng, X., Williams, M., and McNamara, J. 2002.

7. Feng: Theoretical model quantitatively indicating isotope exchange Varied two parameters:Varied two parameters: –Effectiveness of isotopic exchange (Ψ) –Ice-liquid ratio (γ)

8. Isotopic exchange R liq controlled by advection, dispersion and ice-water isotopic exchange R ice controlled by ice-water exchange Rate of isotopic exchange dependent on: Fraction of ice involved in exchange, f –Dependent on size and surface roughness of ice grains –Accessibility of ice surface to infiltrating water –Extent of diffusion within ice –Amount of melting and refreezing at ice surface Ice-liquid ratio quantified by: γ = bf a + bf wherea = mass of water b = mass of ice per unit volume of snow i.e. ratio of liquid to ice

9. Effectiveness of exchange: Ψ= k r Z u * K r is a constant Z = snow depth U * = flow velocity Ψ and γ dependent on melt rate and snow properties e.g. grain size, permeability

10. Results: Effect of varying ψ (effectiveness of isotope exchange) Relative to original bulk snow (  18 O=0) Where Ψ is large = curved trend (a) –Base of snowpack is 18 O depleted as substantial exchange occurs –Low melt rate so slower percolation velocity Where Ψ is small = linear trend (e) –Constant 3‰ difference between liquid and ice

11. Effect of varying γ (and therefore f): Relative to original bulk snow (δ 18 O=0) Low γ = curved trend (e) –Slow melt rate –Lower liquid: ice ratio as lower water content High γ = linear trend (a) –Fast melt rate –Higher water content so more recrystallization Therefore constant difference in 18 O of snowmelt and bulk snow

12. Conclusions: High melt rate = effective exchange and high liquid: ice ratio. Higher percolation velocity so constant difference in 18 O. Increased water content triggers recrystallisation, a mechanism of isotope exchange. –linear trend Low melt rate = Large difference in 18 O initially due to substantial exchange –Only a small proportion of ice is involved in isotopic exchange therefore insignificant change in 18 O of bulk ice – 18 O of liquid and ice reach steady state resulting in curved trend as equilibrium is reached

13. Assumptions: Snow melted from the surface at constant rate Dispersion is insignificant 18 O exchange occurs between percolating water and ice

14. Implications: Variation in  18 O between snowmelt and bulk snow causes errors in hydrograph separation if bulk snow values are used

15. Taylor: Laboratory experiment to determine k r Determination of k r to allow implementation of model in the field Controlled melting experiments: –Melted 3 snow columns of different heights at different rates – 18 O content of snowmelt relative to snow column substituted into model equation to obtain k r K r = Ψu * Z

16. K r = Ψu * Z Range of ψ (effectiveness of isotopic exchange) values obtained by melting a short column rapidly (low ψ) and long column slowly (high ψ) Z = initial snow depth U * = percolation velocity

17. Model used to calculate k r as  18 O is used to infer Ψ (effectiveness of exchange) so equation K r = Ψu * Z can be solved

18. Results k r = 0.16  0.02 hr -1 Small range (0.14 – 0.17 hr -1 ) Small standard deviation (15%) Successful parameterization of k r indicates that the model captures the physical processes that control the isotopic composition of meltwater

19. Results Estimate of f is uncertain –Test 1:0.9 Tests 2-3: 0.2 –Uncertainties Snowpack heterogeneity Recrystallization

20. Snowpack Heterogeneity Real snowpacks are not homogeneous in terms of pore size If water content is low, water may only percolate in small pores Reduces surface area where isotopic exchange can occur

21. Recrystallization Snow metamorphism due to wetting of snow –Small ice grains melt completely No isotopic fractionation –Water refreezes onto larger ice crystals 18 O preferentially enters ice Liquid becomes depleted

22. Recrystallization Change to fraction of ice participating in isotope exchange (f) depends on two processes –Increase in f High mass of snow involved in melt – freeze –Decrease in f Larger mean particle size reduces surface area available for ice – liquid interaction

23. Taylor, S., Feng, X., Williams, M., and McNamara, J. 2002. How isotopic fractionation of snowmelt affects hydrograph separation

24. Locations Central Sierra Snow Laboratory (CA) –Warm, maritime snowpack Sleeper River Research Watershed (VT) –Temperate, continental snowpack Niwot Ridge (CO) –Cold, continental snowpack Imnavit Creek (AK) –Arctic snowpack

25. Methods Sample collection –Meltwater collected from a pipe draining a meltpan (CA, VT, CO) –Plastic tray inserted into the snowpack at the base of a snow pit (AK) Determination of  18 O for meltwater samples

26. Results

27. Results At all locations, meltwater had lower  18 O values at the beginning of the melt event and increasingly higher values throughout the event (3.5% to 5.6%) Trend holds despite widely different climate conditions

28. Why is this important? Using the average  18 O value of pre-melt snowpack leads to errors in the hydrograph separation Timingearlylate  18 O lowerhigher New water estimation overestimatedunderestimated

29. Error Equation  x = estimated error in x x = fraction of new water  18 O New -  18 O Old = isotopic difference between new and old water  18 O New = difference between  18 O in average snowpack and meltwater samples

30. Error Equation Error is proportional to: –Fraction of new water in discharge (x) –Difference in  18 O between snowpack and meltwater (  18 O New ) Error is inversely proportional to: –Isotopic difference between new and old water (  18 O New -  18 O Old )

31. Error Large error if meltwater dominates the hydrograph Expected in areas of low infiltration –Permafrost –Cities Underestimate new water –Assume more enriched water is a mixture of new and old water

32. Error Error magnitude depends on time frame of interest –Maximum error at a given instant in time –Error is lower if entire melt event is considered  18 O Melt ≈  18 O Pack during middle of melt season Negative error and positive error cancel out

33. Other Factors Additional precipitation events Varying melt rates Meltwater mixing Spatial isotopic heterogeneity

34. Additional Applications Incorporation into other models –Mass and energy snowmelt model SNTHERM Glaciers –Climate studies involving ice cores

35. Questions