Development of a New Parameterization for BelowCloud Scavenging
Development of a New Parameterization for Below-Cloud Scavenging of Size. Resolved Particles by Rain and Snow Xihong Wang 1, Leiming Zhang 2, and Michael D. Moran 2 , 1 Kellys Environmental Services Toronto, Ontario, Canada 2013 CMAS Conference 2 Air Quality Research Division Environment Canada Toronto, Ontario, Canada Chapel Hill, North Carolina 28 -30 Oct. 2013
Outline • Motivation • Background and Definitions • Results from Recent Uncertainty Analyses • Methodology for Development of New Below. Cloud Scavenging Parameterization • Results for New Parameterization • Summary DRAFT – Page 2 – 10 September 2020
Motivation for this Work • Wet deposition (= in-cloud + below-cloud scavenging) is often the dominant removal pathway for atmospheric particles at the global scale (e. g. , Stier et al. , 2005) • A significant fraction of global precipitation is solid or frozen; one global modelling study (Croft et al. , 2009) estimated that ~30% of below-cloud scavenging of sulphate particles by precipitation is due to snow • Model intercomparisons have shown that precipitation scavenging parameterizations have large uncertainties (e. g. , Rasch et al. , 2000; Textor et al. , 2006) DRAFT – Page 3 – 10 September 2020
Contributing Studies Wang, X. , L. Zhang, and M. D. Moran, 2010: Uncertainty assessment of current size-resolved parameterizations for below-cloud particle scavenging by rain. Atmos. Chem. Phys. , 10, 5685 -5705, doi: 10. 5194/acp -10 -5685 -2010. Wang, X. , L. Zhang, and M. D. Moran, 2011. On the discrepancies between theoretical and measured below-cloud particle scavenging coefficients for rain - a numerical investigation using a detailed one-dimensional cloud microphysics model. Atmos. Chem. Phys. , 11859 -11866, doi: 10. 5194/acp-11 -11859 -2011. Zhang, L. , X. Wang, M. D. Moran, and J. Feng, 2013. Review and uncertainty assessment of size-resolved scavenging coefficient formulations for snow scavenging of atmospheric aerosols. Atmos. Chem. Phys. , 13, 10005 -10025, doi: 10. 5194/acp-13 -10005 -2013. Wang, X. , L. Zhang, and M. D. Moran, 2013. Development of a new semiempirical parameterization for below-cloud scavenging of sizeresolved aerosol particles by both rain and snow. Geoscientific Model Development Discuss. , 6 (Submitted). DRAFT – Page 4 – 10 September 2020
Types of Hydrometeors Rain (rain is “simple”) • Drops of different diameters Snow (Dp) (snow is more complicated) • [Dm < 5 mm] snow crystals: shapes or habits include plates, columns, stars, needles, dendrites, spheres, and bullets • [Dm > 5 mm] snowflakes: aggregates of snow crystals • Rimed snow crystals, ice pellets, graupel, hail DRAFT – Page 5 – 10 September 2020
Scavenging Mechanisms • • Brownian diffusion interception (size dependent) inertial impaction (mass dependent) thermophoresis diffusiophoresis electric charges wake effects Aerosol particle-hydrometeor collection efficiency E(dp, Dp) is a dimensionless parameter defined as the rate of collection of aerosol particles of diameter dp by the falling hydrometeor normalized by the number of upstream particles of diameter dp swept per unit time across an area equal to the effective cross-sectional area DRAFT – Page 6 – 10 September 2020 [Figure from Slinn (1984)] of the hydrometeor
Definition of Scavenging Coefficient Λ The scavenging coefficent Λ (s-1) describes the scavenging rate or the fractional reduction per unit time of either bulk or size-resolved aerosol particle number concentration (below) or mass concentration Bulk: Size-resolved: DRAFT – Page 7 – 10 September 2020
Theoretical Parameterization of Λ(dp) where Aeff is the hydrometeor effective cross-sectional area, Dp and dp are the hydrometeor and aerosol-particle diameter, V and v are the hydrometeor and aerosol-particle fall velocity, E(dp, Dp) is hydrometeor-aerosol particle collection efficiency, N(Dp) is the hydrometeor number size distribution. Most theoretical parameterizations of E(dp, Dp) consider only 3 collection mechanisms: Brownian diffusion, interception, and inertial impaction DRAFT – Page 8 – 10 September 2020
Finding 1 (Wang et al. , 2010): Different formulations for E(dp, Dp), N(Dp), and V(Dp) for scavenging by rain cause uncertainties in values of Λ(dp) of up to 2 orders of magnitude. Finding 2 (Wang et al. , 2010): Theoretical values for Λ(dp) for scavenging by rain are at least one order of magnitude smaller than estimates from field studies. DRAFT – Page 9 – 10 September 2020
Finding 3 (Wang et al. , 2011): Discrepancies between theoretical and field-derived values of Λ(dp) for scavenging by rain can largely be explained by the contribution of vertical diffusion, which is not considered in theoretical parameterizations for Λ(dp) but which does contribute to field-derived estimates. DRAFT – Page 10 – 10 September 2020
Finding 4 (Zhang et al. , 2013): Different formulations for E(dp, Dp), N(Dp), V(Dm), and A(Dm) for scavenging by snow can cause uncertainties in values of Λ(dp) of up to 3 orders of magnitude. Snow particle shape must also be known or assumed. DRAFT – Page 11 – 10 September 2020
Starting Point for a New Parameterization for Below-Cloud Scavenging of Size-Resolved Particles Recognizing Current Limitations • Few field experiments have investigated size-resolved scavenging by rain and even fewer by snow (2? ) • Field-derived values of Λ(dp) implicitly account for more processes than theoretical treatments and hence can only be expected to provide upper bounds on Λ(dp) • Theoretical parameterizations for Λ(dp) have large uncertainties but can be used; empirical schemes should not be used (to avoid process double-counting) • • Scavenging by both rain and snow should be considered • Focus should be on the higher values predicted by the available ensemble of theoretical parameterizations DRAFT – Page 12 – 10 September 2020 based on comparisons to field data (Wang et al. , 2010; Zhang et al. , 2013) A wide range of precipitation conditions and aerosol particle sizes should be considered for widest applicability
Methodology: Part 1 • Develop an ensemble of theoretical parameterizations for Λ(dp) based on consideration of all available formulas for E(dp, Dp), N(Dp), V(Dm), and A(Dm) for rain and for snow • For each value of dp calculate 50 th, 70 th, 80 th, 95 th, and maximum percentile values of Λ(dp) for selected precipitation intensities R from 0. 01 to 100 mm h-1 for rain and from 0. 001 to 10 mm h-1 for snow (as liquid water equivalent) DRAFT – Page 13 – 10 September 2020
Ensemble of Λ(dp) Formulations Considered Physical Quantity E(d, Dp) N(Dp) VD Λrain Number of Formulas 5 10 8 400 E(d, Dp) 3 N(Dp) 4 VD 8 A* 4 (habits) Λsnow 168 *Some VD formulas for snow are specific to one snow crystal habit, so – Page 14 – 10 September 2020 only 14 possible VD–ADRAFT combinations
Λ(dp) Ensemble for Scavenging by Rain DRAFT – Page 15 – 10 September 2020
Methodology: Part 2 • Starting from a selected percentile value (90 th) of Λ(dp), for each value of dp calculate a log-log best-fit line between Λ 90(dp) and R for both rain and snow or • Use a least-squares polynomial curve-fitting tool to calculate 6 th-order polynomial functions for A(d) and B(d) for both rain and snow; note that the accuracy is improved if fitting is performed for 2 contiguous segments • The result is 2 formulas for Λ 90(dp) as a function of DRAFT – Page 16 – 10 September 2020 precipitation intensity R, one for rain and one for snow
Calculation and Fitting of A(d), B(d) for Snow DRAFT – Page 17 – 10 September 2020
Goodness of Fit for A(d), B(d) for Snow Original A(d) vs. Polynomial-Fit A(d) Relative Error as a Function of d DRAFT – Page 18 – 10 September 2020
Summary • A new semi-empirical parameterization for below-cloud scavenging of size-resolved aerosol particles by both rain and snow has been developed for use in regional and global air-quality models • The results of recent uncertainty analyses of existing theoretical parameterizations for Λ(d) have been considered as well as available field-derived values ( known representativeness) • The new parameterization has a simple power-law form and only requires precipitation intensity and precipitation type to be provided as inputs ( easy to implement) DRAFT – Page 19 – 10 September 2020
- Slides: 19