The use of one-dimensional and bulk lake models in studies of lake – atmosphere interaction Parameterization of Lakes in Numerical Weather Prediction and. - презентация

1 The use of one-dimensional and bulk lake models in studies of lake – atmosphere interaction Parameterization of Lakes in Numerical Weather Prediction and Climate Modelling September 2008, St. Petersburg (Zelenogorsk), Russia V. M. Stepanenko (1), E. Dutra (2) (1) Moscow State University, Scientific Research Computing Center (2) University of Lisbon, Center of Geophysics The work is supported by SKIF project and RFBR grant N

2 The effects of water reservoirs on atmosphere breezes and associated tracer transport breezes and associated tracer transport severe snowfalls over large lakes in winter severe snowfalls over large lakes in winter Weather Climate the change of regional hydrological system due to global warming emission of methane by Siberian lakes emission of methane by Siberian lakes

3 Numerical water reservoir models in coupled lake – atmosphere studies 1) 3-dimensional (~oceanic) 2) 2-dimensional vertically averaged (Shlychkov)vertically averaged (Shlychkov) averaged in one lateral direction (CE-QUAL x.x model)averaged in one lateral direction (CE-QUAL x.x model) 3) 1-dimensional single-column (GOTM model (Burchard et al.), Lake model, V. M. Stepanenko & V. N. Lykosov, 2005);single-column (GOTM model (Burchard et al.), Lake model, V. M. Stepanenko & V. N. Lykosov, 2005); laterally averaged models (O. F. Vasiliev et al., 2007) – applicable in many applicationslaterally averaged models (O. F. Vasiliev et al., 2007) – applicable in many applications 4) ½ - dimensional models – the vertical profiles of temperature, salinity etc. are parameterized (Flake model, D. V. Mironov et al., 2006) – high computational efficiency application in operational forecast 5) 0 – dimensional (mixed models)

4 Lake model (SRCC MSU) the equation for the horizontally averaged temperature: the equation for the horizontally averaged temperature: the equations for horizontal velocities: the equations for horizontal velocities: the salinity/hydrosol transport the salinity/hydrosol transport turbulent dissipation Coriolis force horizontal pressure gradient force the friction of flow on vegetation the advection by tributaries gravitational sedimantation Coordinate transformation:

5 Snow and soil models Ice model Ice model Snow model (Volodina et al., 2000) Snow model (Volodina et al., 2000) Soil model (Volodin and Lykosov, 1998) Soil model (Volodin and Lykosov, 1998) diffusion terms freezing terms gravitational infiltration Snow Ice Water Soil (sediments) U H,LE EsEs EaEa S

6 Turbulent mixing parameterization Kolmogorov formula (1942) M – friction frequency, N – Brunt-Vaisala frequency stability functions k-ε parameterization Boundary conditions - counter-gradient effects missing

7 Willis-Deardorff experiment (1974) Setup for numerical experiment: horizontally homogeneous water layer of infinite depth; horizontally homogeneous water layer of infinite depth; linear initial temperature profile with the lapse rate -1ºС/10 m; linear initial temperature profile with the lapse rate -1ºС/10 m; the constant sensible heat flux at the surface 100 W/m 2 (cooling); the constant sensible heat flux at the surface 100 W/m 2 (cooling); the horizontal velocities 0 m/s; the horizontal velocities 0 m/s; Coriolis force is neglected Coriolis force is neglected

8 The terms of turbulent energy budget Lake model 1) E-ε model by Canuto et al., ) LES results (Mironov et al., 2000) The k-closure The shortcoming: does not take into account non-local effects of convective thermals does not reproduce uniform temperature profile

9 The counter-gradient heat transport by convective thermals Temperature flux (Lake model) Temperature flux (other models and LES) (Soares et al. 2004)

10 Implementation issues Fortran 90 code Fortran 90 code MPI libraries MPI libraries Netcdf libraries Netcdf libraries Lake driver implementation for N points (lakes) at P processors, NP Lake driver implementation for N points (lakes) at P processors, NP 1kP …… P+1P+k2P … … … 2P+k … Number of lake MPI- process rank 1kP1kPk Number of netcdf output file 1 k P

11 Surface temperature from Lake model and from observations Kossenblatter lake, Germany, June, 1998 Tiksi lake, Siberia, July, 2002 Monte-Novo lake, Portugal,

12 The temperature in Mogaiskoe reservoir Surface temperature time series, – Vertical temperature profile, 01:00, Important features not taken into account: Lengmuir circulations; Lengmuir circulations; Seiches; Seiches; advection by tributaries. advection by tributaries.

13 Snow surface temperature, (Kolpashevo, Western Siberia, February, 1961) Temperature, C Time, days

14 Lake Alqueva (Portugal)

15 Sensible heat flux ( Flake and Lake) lake Alqueva, summer 2007

16 Sensible heat fluxes (Flake and Lake) Averaging interval TESSEL+FlakeFlakeLake Observati ons 15 days 29.8 W/m W/m 2 22 W/m W/m 2 55 days 30 W/m W/m W/m 2 Source code of Lake model and data for verification

17 The role of lake depth (Baikal) 1)The real depth m 2) The depth 100 m

18 The role of radiation extinction coefficient

19 The role of radiation extinction coefficients (Baikal and Caspian sea)

20 Mesoscale atmospheric model The code of Nh3d model (Miranda & James, 1992) 3-dimensional 3-dimensional σ-coordinates σ-coordinates non-hydrostatic equation set non-hydrostatic equation set warm cloud microphysics warm cloud microphysics ISBA soil model ISBA soil model New features: shortwave (Clirad-SW) and longwave (Clirad-LW) radiation parameterization lake model lake model aerosol transport scheme aerosol transport scheme

21 Aerosol transport scheme - turbulent diffusion (1-st order closure), - Raileigh damping term Boundary conditions: at all boundaries Numerical scheme: Smolarkiewich monotonous scheme spatial discretization – 2-d order spatial discretization – 2-d order temporal discretization – 2-d order temporal discretization – 2-d order - aerosol source, - sedimentation speed,

22 Aerosol distribution in test case Resolution: x = y = 3.7 km x = y = 3.7 km 21 σ – levels 21 σ – levels Hanty-Mansiiskregion Time integration t = 5 sec t = 5 sec 8 days 8 days Breezes develop over water bodies and transport the tracer far from source even in calm synoptic conditions Near surface wind Aerosol cloud

23 The work underway: aerosol emission and sink at water bodies Lake Yarato Norilsk: Emission of Zn, Cu, Hg,… lakes Aral Sea

24 Future development of Lake model Insertion the model of methane generation, transport and sink in the soil (lake sediments) – B. Walter and M. Heimann, 2000; Insertion the model of methane generation, transport and sink in the soil (lake sediments) – B. Walter and M. Heimann, 2000; Introduction of the methane ebullition and bubble convection parameterization in the water body; Introduction of the methane ebullition and bubble convection parameterization in the water body; Incorporation of the computationally efficient version of the model into climate model of the Institute for Numerical Mathematics, Moscow; Incorporation of the computationally efficient version of the model into climate model of the Institute for Numerical Mathematics, Moscow; …

25 Acknowledgements Dmitrii Mikushin implemented aerosol transport scheme in mesoscale model; Dmitrii Mikushin implemented aerosol transport scheme in mesoscale model; Vasilii Lykosov has initiated this research and supports it; Vasilii Lykosov has initiated this research and supports it; Rui Salgado, Maria Grechushnikova provided the observational data Rui Salgado, Maria Grechushnikova provided the observational data Pedro Soares provided the code of counter- gradient convection parameterization Pedro Soares provided the code of counter- gradient convection parameterization Dmitrii Mironov, Pedro Viterbo, Pedro Miranda, Gianpaolo Balsamo initiated useful discussions Dmitrii Mironov, Pedro Viterbo, Pedro Miranda, Gianpaolo Balsamo initiated useful discussions

26 Thank you! Your questions are welcome!

27 Conclusions

28 Parallel implementation aspects Explicit scheme! FFT + Eulerian elimination Cycles with independent iterations % - the solution of elliptical equation for geopotential (via horizontal FFT) % - the integration of thermodynamic equation (including radiation model) % - the solution of continuity equation % - computation of turbulent transport % - the integration of momentum equation 6. 6 % - the calculation of momentum fluxes 7. 5 % - the integration of water vapour and aerosol transport 8. 5 % - the soil, vegetation and lake models

29 Application in education