Implementation of modules for wet and dry deposition into the ECMWF Integrated Forecast System Johannes Flemming (ECMWF), Vincent Huijnen (KNMI) and Luke Jones (ECMWF) Deliverable D G-RG 4.6 1
Abstract A description and initial evaluation of the dry and wet deposition schemes as implemented in ECMWF s Integrated Forecast System (IFS) is presented. Dry deposition is based on prescribed monthly deposition velocities. Wet deposition is based on the GMI Harvard scheme. 2
Contents Abstract... 2 1. Introduction... 4 2. Dry deposition simulation... 4 2.1. Dry deposition velocity fields input... 6 3. Wet deposition... 8 4. Evaluation of dry and wet deposition budgets... 9 5. Integration in the IFS code... 10 6. Conclusions and Outlook... 10 3
1. Introduction A major new development in the MACC G-RG subproject concerns the development of inline chemistry within IFS, which is referred to as C-IFS. The C-IFS system could resolve limitations of the current coupled CTM-IFS system, for instance tracer concentration displacement issues between the CTM and IFS. Furthermore, it runs more efficiently on the ECMWF supercomputers and would therefore allow for finer resolution or the set-up of a model ensemble in the operational forecasts of the future GMES atmospheric service. Building blocks of the C-IFS system are the chemistry and photolysis module (see deliverable 4.4), as well as modules for dry and wet deposition (described in this document) and emissions (deliverable 4.5). Dry deposition describes the process of the removal of atmospheric traces gases and aerosol to the earth surface without the interaction with precipitation or clouds. The dry deposition intensity depends on the species and the properties of the earth surface including vegetation to absorb the trace components. Dry deposition depends on the vertical turbulent transport (diffusion) because diffusion determines the concentration close to the surface. Wet deposition is the process of removal or transport of atmospheric traces gases and aerosol by cloud water or precipitation. It acts on species that dissolve in water. Rain-out is the removal of trace gases dissolved in cloud droplets by precipitation. Wash-out is caused by the transfer of traces gas in falling rain below the cloud. Cloud droplets with dissolved traces gases can be moved upward in convective clouds and can evaporate, which leads to the vertical transport of trace gases. 2. Dry deposition simulation The dry deposition flux Φ D at the model surface is calculated in C-IFS by a simple approach based on pre-scribed dry deposition velocity. The dry deposition velocity fields are a model input, which is described with more detail in section 3. Φ D is calculated from the mass mixing ratio MMR, the dry deposition velocity v D and the density at the lowest model layer in the following way: Φ = D vd MMRρ Given the long time step (3600 s) of C-IFS, the calculation of the concentration decrease by dry deposition has to account for the implicit character of the dry deposition flux since it depends on the mass mixing ratio itself. If not taken into account, the loss by dry deposition is overestimated which could result in negative concentration values. For example, a drydeposition velocity of 4.2 mm/s applied over a time step of 3600 s would empty a grid box of 15m vertical extent if the respective flux was simulated using only the mass mixing ratio at the start of the time step. In C-IFS, there are currently three alternative ways to calculate dry deposition: 4
1. Add dry deposition flux to surface flux (together with emissions) used in vertical turbulent transport scheme 2. Separately calculate the following MMR time derivative because of dry deposition in the lowest model layer using a implicit formulation: d MMR v = D MMR dt Z 3. Including dry deposition velocity as a first-order reaction for the surface as part of the CBM5 chemical solver (TM5 only) Approach 1 links the process of dry deposition with vertical diffusion, which is more favourable than the more separate treatment of dry deposition in approaches 2 and 3. However, it does currently not account for the implicit character of the dry deposition flux. It is therefore necessary to constrain the dry deposition flux to avoid negative concentrations. The current ad-hoc approach is to limit the deposition flux equivalent to mass loss smaller than 70% of the tracer mass at the start of the time step in the surface layer over the given time step. Approach 2 and 3 are applied after emissions injection and diffusion. Approach 2 allows the application of different numerical solutions of the change of the mass mixing ratio because of dry deposition. Approach 3 applies the implicit approach on a sub-time step and is supposed to produce a solution close to the analytical solution. Approach 2 allows the choice between the following numerical schemes: Explicit (Euler-Forward): Implicit (Euler-Backward): 1 v D 0 = 0 MMR MMR t MMR Z Centred implicit approach: 1 v D 0 = 1 MMR MMR t MMR Z v D MMR + MMR MMR1 MMR0 = t Z 2 Analytical solution: exp vd MMR1 = MMR0 t Z 1 0 The differences in the rate of mass mixing ratio change among the different approaches are shown in Figure 1. Larger differences occur for higher deposition velocities or time steps. The values of the deposition velocities in relation to the time step can be found in a regime were the differences between the numerical approaches become large. The centred implicit approach is closest to the true solution which is more costly to calculate. 5
Figure 1 Change in normalised mass mixing ration MMR/MMR 0 because of dry deposition (y-axis, MMR 0 is initial value) vs. a normalised dry-deposition velocity (xaxis) for different numerical solvers (see approach 2 above). The x-axis unit are scaled by the ratio of model time step and surface layer box height t/ Z. x = 1 corresponds to a total tracer mass loss if a constant dry deposition flux (explicit) is assumed. 3. Dry deposition velocity input The dry deposition fields are currently assumed constant for the model forecast, which is typically 24 to 72 hours. The dry deposition velocities have to be prepared as IFS input file before the model run. Currently the input file is interpolated in time from monthly global fields which had been prepared by averaging 3-hourly dry deposition velocity field output from the TM5 model. The monthly data set are available at the resolutions T159 and T255. The dry deposition velocities were calculated using the approach from Weseley et al. (1989) and Ganzeveld and Lelieveld (1995). It uses meteorological and surface input data including 2 meter temperature, wind speed, surface roughness and soil wetness. At the surface the model makes a distinction between uptake resistances for vegetation, soil, water, snow and ice (cf. Table 1). The vegetation resistance is calculated using the in-canopy aerodynamic, soil, and leaf resistance. The stomatal resistance is calculated online, depending on e.g. the soil wetness at the uppermost surface layer, where together with the cuticle and mesophyl resistances this is combined into the leaf resistance. The resulting annual means based on TM5 version TM5-chem-v3.0 (Huijnen et al. 2010), for the year 2006, using ECMWF operational meteorology are shown in Figure 2 to Figure 15 in the appendix. Table 2 contains the all the species, which are subject to dry deposition and the respective average, minimum and maximum dry deposition velocities. It is foreseen to use the SUMO model to calculate the deposition velocity to be used in C-IFS. 6
Trace gas r soil r wat r snow/ice r mes O 3 400 2000 2000 1 CO 5000 10 5 10 5 5000 NO 10 5 10 5 10 5 500 NO 2 / NO 3 600 3000 3000 1 HNO 3 / N 2O 5 1 1 1 1 H 2O 2 80 72 80 1 SO 2 100 1 1 1 PAN / ORGNTR 3994 295 3394 1 ALD2 10 5 300 10 5 200 CH 2O / CH 3 COCHO 1666 254 1666 1 CH 3OOH / ROOH 3650 293 3650 1 NH 3 100 1 10 5 1 Table 1 Selected soil, water, snow/ice and mesophyl resistances according to Ganzeveld and Lelieveld (1995) and Ganzeveld et al. (1998), in s m -1. The cuticle resistance is 10 5 s m -1, for all trace gases except for HNO 3 and N 2 O 5, where a value of 1 s m -1 is adopted. Species Min vd Max vd Ave vd HCOOH 0.008 59.781 14.471 HNO3 0.008 59.781 14.471 MCOOH 0.008 59.781 14.471 N2O5 0.008 59.781 14.471 NH3 0.008 72.741 14.331 SO2 0.323 30.318 9.885 H2O2 0.841 14.894 6.454 HNO4 0.494 12.562 3.633 CH2O 0.228 13.944 2.681 MGLY 0.228 13.944 2.681 CH3OH 0.139 8.869 2.202 CH3OOH 0.139 8.869 2.202 ETHOOH 0.139 8.869 2.202 ORGNTR 0.139 8.869 2.202 PAN 0.139 8.869 2.202 ROOH 0.139 8.869 2.202 ISPD 0.114 6.972 1.340 O3 0.193 13.891 1.124 ETHOH 0.069 4.434 1.101 NO2 0.147 10.763 0.812 NO3 0.147 10.763 0.812 ALD2 0.008 3.571 0.127 NO 0.008 3.571 0.127 CO 0.008 1.151 0.074 Table 2 The chemical trace species which are subject to dry deposition in the extended CBM4 mechanism and minimum, maximum and area-averaged dry deposition velocities in mm/s. 7
4. Wet deposition The module for wet deposition in C-IFS is currently a very simple scheme which is based on the Harvard wet deposition scheme (Jacob et al., 2000). The calculation of the wet deposition includes the following processes: 1. In-cloud gas scavening in liquid water based on Henrys-law-equilibrium concentrations in rain and snow 2. Gas release by evaporation of rain and snow 3. Wash out of gases by rain and snow falling through air outside clouds assuming a fall speed of 5m/s The processes 1, 2 and 3 are calculated separately and after each other for large-scale and convective precipitation. Wet removal processes are not considered for ice clouds. The effective Henry-coefficients k H and the reaction enthalpies H / R describing the temperature dependcies are provided as model input (see table 1). All gases are assumed to be in Henry-equilibrium. The mass transfer between gas phase and aqueous phase is not considered to be a limiting factor for species. The loss of dissolved gas by 1) in clouds and the gain by 2) are determined by the change in the precipitation flux between the model levels and the level above. The precipitation flux is scaled to be valid over the cloud cover fraction of the respective grid-box. The retention coefficient, describing the retension of dissolved gas in the liquid cloud condensate as it is converted to precipitation (< 1 accounts for volatilization during riming), is one for all species in warm clouds (T>268); it is 0.02 for all species in mixed-clouds except for HNO3 (1.0) and H2O2 (0.05). The effective available liquid water content for 3) is determined by the precipitation flux assuming a constant fall speed of 5 m/s. Name Molar Weight Henry coefficient kh H / R HNO3 63 3.20E+11 8700 SO4 96.1 3.20E+11 8700 NH4 18 3.20E+11 8700 MSA 96.1 3.20E+11 8700 Pb 210 3.20E+11 8700 H2O2 34 8.30E+04 7400 MGLY 62 3.20E+04 8700 HCOOH 46.01 8.90E+03 6100 MCOOH 62.02 4.10E+03 6300 CH2O 30 3.20E+03 6800 SO2 64.1 2.40E+03 5000 ROOH 47 340 6000 ETHOOH 62.02 340 6000 CH3OOH 48 3.10E+02 5200 CH3OH 31.01 220 5200 ETHOH 46.02 190 6600 8
NH3 17 75 3400 ALD2 24 17 5000 ORGNTR 14 1 6000 Table 3 The chemical trace species which are subject to wet deposition and the effective Henry-coefficients at 298 K and the reaction enthalpies. 5. Evaluation of dry and wet deposition budgets An evaluation of the deposition budgets is performed to assess the current schemes. These annual budgets are compared against the offline TM5 model using basically the same chemistry scheme and emissions. The deposition budgets are a result of the model concentration fields and the input deposition velocity fields and wet scavenging parameterization and therefore depend implicitly on other model parameterizations. Although the order of magnitude of the dry deposition budgets is in general similar, the absolute values of the budgets between TM5 and C-IFS can be different by 50%. For a few tracers like SO2, SO4, MGLY and ALD2 the deposition budgets show even larger differences between TM5 and C-IFS, suggesting discrepancies in the model input settings. It should be noted that the impact of the various fluxes on the chemical composition depends on its relative contribution to all production and loss terms. Name TM5 dry dep CIFS dry dep TM5 wet dep CIFS wet dep O3 803 801 - - CO 182 179-13 H2O2 50 59 293 202 SO2 195 68 178 60 SO4 30.5-259 69.5 NH3 36.2 20.6 5.9 11.1 NH4 - - 27.9 23 NO (Tg N) 0.33 2.6 - - NO2 (Tg N) 5.3 2.6 - - NO3 (Tg N) 0.001 0.004 - - N2O5 (Tg N) 0.073 0.06 - - HNO3 (TG N) 8.3 6.9 22.7 28.1 HNO4 (Tg N) 0.050 0.046 - - PAN (Tg N) 2.4 2.6 - - ORGNTR (TgN) 2.4 3.3 5.6 5.9 MSA - - 5.8 5.4 Pb - - 115e-10 109e-10 MGLY 2.3 1.9 10.4 1.8 HCOOH 32.4 4.5 9.5 3.9 MCOOH 34.4 43.7 143.6 118 9
CH2O 32.2 35 193 41 ALD2 7.0 2.5 0.45 5.9 ROOH 59.7 37 63.2 77 ETHOOH 1.5 0.9 3.0 3.4 CH3OOH 48.7 43.1 54.8 137 CH3OH 120 57 57.6 58 ETHOH 1.38 1.0 0.66 1.0 Table 4 Comparison of dry and wet deposition budgets for a one year run for 2007 for CIFS and TM5 (Tg species / yr). 6. Integration in the IFS code The mass mixing ratios at the beginning of the time step are used for the calculation of the chemical tendencies. The tendencies of the chemical conversion are added to the tendencies describing emission advection, injection, dry deposition and diffusion as well as wet deposition. Routine name Called from Purpose Output Chem_scav.F90 Chem_main.F90 Wet deposition for either large-scale or convective precip Chemical tendencies Chem_drydep.F90 Chem_main.F90 dry deposition Chemical tendencies Gems_init.F90 Callpar.F90 Add dry-deposition to emissions Table 5 Fortran routines added to or changed in the IFS code. 7. Conclusions and Outlook We have presented a first implementation of a very basic wet and dry deposition scheme into ECMWF s Integrated Forecast System. The global budgets are in-line with the number for the TM5 model. More testing is needed to investigate the spatial and temporal plausibility of the deposition fields. The following improvements should be implemented for the dry deposition scheme: 1. Implicit treatment of dry deposition flux as part of tracer diffusion (approach 1) 2. Investigate in more detail the difference of the different numerical approaches 3. Use of input data from the SUMO model 4. Update of deposition velocities during model run 10
The following improvements should be implemented for the wet deposition scheme: 1. Transfer limitation of scavenging for highly soluble species such as HNO3 2. Scavenging in convective updrafts as part of the convection code 3. Overall checking and refinement 11
References Jacob, D.J. H. Liu, C.Mari, and R.M. Yantosca, Harvard wet deposition scheme for GMI, Harvard University Atmospheric Chemistry Modeling Group, revised March 2000. Ganzeveld, L. and Lelieveld, J.: Dry deposition parameterization in a chemistry general circulation model and its influence on the distribution of reactive trace gases, J. Geophys. Res., 100(D10), 20999 21012, 1995. Ganzeveld, L., Lelieveld, J., and Roelofs, G.-J.: A dry deposition parameterization for sulfur oxides in a chemistry and general circulation model, J. Geophys. Res., 103(D5), 5679 5694, doi:10.1029/97jd03077, 1998. Lathiere, J., Hauglustaine, D. A., Friend, A. D., De Noblet-Ducoudre, N., Viovy, N., and Folberth, G. A.: Impact of climate variability and land use changes on global biogenic volatile organic compound emissions, Atmos. Chem. Phys., 6, 2129 2146, doi:10.5194/acp-6-2129- 2006, 2006. Weseley, M. L.: Parameterization of surface resistance to gaseous dry deposition in regional numerical models, Atmos. Environ., 16, 1293 1304, 1989. 12
Appendix: Annual averaged dry deposition velocities used in C-IFS 13
Figure 2 Annual SO2 dry deposition velocity in mm/s Figure 3 Annual ROOH dry deposition velocity in mm/s, also used for PAN, OGRNTR, ETHOOH, CH3OH and CH3OOH 14
Figure 4 Annual O3 dry deposition velocity in mm/s. Figure 5 Annual NO3 dry deposition velocity in mm/s, also used for NO2 15
Figure 6 Annual NO dry deposition velocity in mm/s Figure 7 Annual NH3 dry deposition velocity in mm/s 16
Figure 8 Annual N2O5 dry deposition velocity in mm/s, also used for MCOOH, HNO3, HCOOH Figure 9 Annual MGLY dry deposition velocity in mm/s, also used for CH2O 17
Figure 10 Annual ISPD dry deposition velocity in mm/s Figure 11 Annual HNO4 dry deposition velocity in mm/s 18
Figure 12 Annual H2O2 dry deposition velocity in mm/s Figure 13 Annual ETHOH dry deposition velocity in mm/s 19
Figure 14 Annual CO dry deposition velocity in mm/s Figure 15 Annual ALD2 dry deposition velocity in mm/s 20