44th Annual CMOS Congress May 31 June 4, 21, Ottawa Interactive lakes in the Canadian Regional Climate Model (CRCM): present state and perspectives A. Martynov, L. Sushama, R. Laprise Centre ESCER Université du Québec à Montréal
Motivation: Lakes and regional climate 9% of Canada s surface is covered by lakes. In total: more than 2 million lakes of all kinds and sizes. Lakes influence the regional climate in many ways; Some examples: - Thermal moderation - Enhanced evaporation - Local wind and participation patterns: breeze, lake-effect snow, etc Lakes are, in their turn, influenced by climate change. Tundra ponds Glacial lakes and reservoirs (Canadian shield) Large lakes are most important for the regional climate The influence of small lakes can be important in areas, where such lakes are abundant because of their cumulative effects. Mountain lakes (Rocky Mountains) Great Lakes Wetlands (swamps, peatlands) Lakes form an important element of the climate system and it is essential to include them in climate models, particularly for lake abundant regions, such as Canada, Scandinavia and northern Russia. This can be achieved by interactive coupling of the regional/global climate models with the lake models.
Regional climate models and lake models Regional climate models: - Horizontal resolution: 1- km - Simulation periods: -2 years and more coupling with complex 3D lake models would be computationally expensive and would require better horizontal resolution. 1D lake models: - Horizontal fluxes are ignored - Few input data are required - Different kinds of lakes can be simulated - Relatively simple - Computationally cheap Two 1D lake models with ice component were chosen for coupling with CRCM (other models might be added later): - The model of S.W. Hostetler (Hostetler et al. 1993) - The model FLake (Mironov et al. 21) A number of off-line tests and validation studies have been carried out, prior to coupling. They were presented on CMOS 29 in Halifax and recently published: Martynov A., L. Sushama, R. Laprise: "Simulation of temperate freezing lakes by one-dimensional lake models: performance assessment for interactive coupling with regional climate models" Boreal Env. Res. 1: 143 164 Both lake models are good in reproduction of small and medium-sized temperate lakes, but problems arise in large and deep lakes, such as Laurentian Great Lakes. Advanced off-line studies: the Lake Model Intercomparison Project (LakeMIP) was initiated in 28 in order to compare the performance of different 1D lake modes in various conditions. For more details, see: www.lakemip.net and: Stepanenko, V. M., Goyette, S., Martynov, A., Perroud, M., Fang, X. & Mironov, D. 21: First steps of a Lake Model Intercomparison Project: LakeMIP. Boreal Env. Res. 1: 191-22.
Canadian Regional Climate Model (CRCM) The first generation of the Canadian Regional Climate Model was developed more than 2 years ago at the University of Quebec at Montreal and many versions have been developed since that time (Laprise 28). The 4th generation model is used, for example, by Consortium Ouranos (www.ouranos.ca) for climate change simulations. The current production version, CRCM4, uses the CLASS 2.7 (Verseghy et al. 1993 ) land surface scheme. Lakes can be treated in CRCM4 in two different ways: external SST data (AMIP II) or a simple mixed layer model (Goyette 2) can be applied. This lake model is inconvenient in use and requires additional simulations, compared with "no-lakes" configurations. This models has limitations, particularly when used for future climate simulations. A new version of the Canadian Regional Climate Model is being developed: CRCM, based on GEM, the global NWP model of Environment Canada. In GEM, lake surface temperature and ice fraction are prescribed, using AMIP II data. This work presents results from interactive coupling of 1D lake models with CRCM in order to improve the representation of surface processes in the climate model. - Two 1D lake models were coupled with CRCM4/CLASS2.7 for resolved lakes only and applied to the Great Lakes region. - The Hostetler lake model was coupled with CRCM for both resolved and subgrid lakes and applied to the Great Lakes region.
CRCM: coupling with 1D lakes Two different lake model coupling schemes were developed: 1. CRCM4 + CLASS 2.7 2. CRCM (GEM) Atmospheric module: Forcing variables at the lowest atmospheric level Atmospheric module: Forcing variables at the lowest atmospheric level Lake models are called by the surface scheme launcher Only resolved lakes can be treated by CLASS 2.7 lake models are applied only to resolved lakes (the mosaic approach is possible In newer CLASS versions) CLASS launcher: Atmospheric forcing at the screen height For resolved lakes Lake coupler: choice of lake model Surface heat and momentum fluxes, surface energy balance Lake model Feedback to the atmosphere: SST, ice thickness, etc. Lake model launcher Is independent from surface schemes. The mosaic approach is realised before the surface couplers are called The resulting surface variables are aggregated, based on their fractions on the grid tiles. GEM SURFACE module Mosaic approach Lake fractions are calculated for grid tiles For tiles with lake fractions Lake coupler: choice of lake model Surface heat and momentum fluxes, surface energy balance Lake model Aggregation of calculated surface parameters for different surface types
Coupled CRCM4 simulations: resolved lakes First coupled simulations : 26 year-long period, 1979-. Five different CRCM4 model configurations were compared Simulation domain: 8x9 grid, centered on the Great Lakes Boundary forcing: NCEP/NCAR reanalysis. Observation data: NDBC buoy water surface temperature Simulation domain and lake tiles
Comparison of coupled and uncoupled simulations: - AMIP II SST - Goyette model - Hostetler model (native surface fluxes) - Flake model (native surface fluxes), 6 m max. depth 3 2 1 1 Coupled CRCM4 simulations: resolved lakes Buoy observations AMIP Goyette Hostetler FLake Buoy 41, depth: 261.6 m 1986 1987 1988 1989 As in off-line tests, best results are obtained in shallow lakes (Erie) Problems arise in deeper lakes (Michigan, Superior). 3 2 1 1 Buoy 47, depth: 164.6 m 1986 1987 1988 1989 Hostetler model provides too rapid and too strong spring heating, due to lacking under-ice mixing. FLake produces too few ice in southern lake Michigan. Both lake models have difficulties in describing the spring temperature patterns in deep lakes. 3 2 1 1 Buoy 4, depth: 12.6 m 1986 1987 1988 1989
Coupled CRCM4 simulations: resolved lakes The importance of surface heat flux parameterization. - Flake model (native surface fluxes): Flake I - Flake model (Hostetler model surface fluxes, BATS-based ) : Flake II 3 2 1 1 Buoy observations FLake I FLake II Buoy 41, depth: 261.6 m 1986 1987 1988 1989 1986 1987 1988 1989 FLake II with BATS-based surface fluxes is generally too cold, compared with FLake II and buoy observations. Not only the correct lake model, but also correct surface heat flux parameterizations have to be used: a task for LakeMIP? 3 2 1 1 Buoy 47, depth: 164.6 m 3 2 1 1 Buoy 4, depth: 12.6 m 1986 1987 1988 1989
Coupled CRCM4 simulations: resolved lakes Simulation averages and climatological means Simulation: 26-year-long averaged SST values Climatology: from Irbe 1992, Goyette et al. 2 Among all tested CRCM versions, best results are obtained by the FLake I configuration FLake I is comparable with the Goyette model, which is the standard lake model in CRCM4, but is simpler in use, than the Goyette model. In the shallow lake Erie, most CRCM versions are close to climatological means (except Flake II) - good hint for shallow subgrid lakes. Average water surface temperature, o C 2 Superior 1 2 1 Huron Ontario 3 2 Erie Climatic monthly average AMIP II Goyette Hostetler FLake I FLake II Michigan 1 2 4 6 8 Months 1 12 2 1 2 1
Coupled CRCM simulations: resolved/subgrid lakes First multiannual simulation, carried out by CRCM (GEM) climate model, interactively coupled with the Hostetler lake model, is presented. Both resolved and sub-grid lakes are taken into account (mosaic approach). Simulation domain: x grid, centred on the Great Lakes, horizontal resolution :.. Simulation period: 1977-1988. Coupled lake model: Hostetler, lake depth: 4 m everywhere. Simulation domain and lake fraction map (wetlands not included).1.2.3.4..6.7.8.9 1.
Coupled CRCM simulations: resolved/subgrid lakes CRCM/Hostetler model, compared with CRCM4/Hostetler and with NDBC buoy observations Buoy observations Hostetler + CRCM4 Hostetler + CRCM 3 2 1 Lake Superior, buoy 41 198 1982 1984 1986 Lake Michigan, buoy 47, depth: 164.6 m 3 2 1 1 198 1982 1984 1986.1.2.3.4..6.7.8.9 1. 3 2 1 1 Lake Mistassini (subgrid, lake fraction =.26) 3 2 1 1 198 1982 1984 1986 Lake Erie, buoy 4, depth: 12.6 m 198 1982 1984 1986 Preliminary results demonstrate the physically correct work of coupled model. For better results, surface heat flux parameterization tuning and using realistic Lake depth and water transparency data are required.
Future work Adjustment of the surface flux parameterization for coupled lake models. Participation in surface flux parameterization comparison activity of LakeMIP. Adding realistic lake depth and water transparency data to the coupled CRCM model. Surface heat flux parameterization tuning. Validation runs for CRCM, coupled with Hostetler model. Adding other lake models to CRCM: FLake and, possibly, other models. Modification of the Hostetler and FLake lake models in order to improve the performance in large deep lakes, possibly using 3D lake simulations as a base for the enhanced mixing parameterization. Comparison of coupled CRCM model with other lake-coupled climate models (RCA, etc.) Climate change simulations, using lake-coupled climate models.
References Dickinson R.E., Henderson-Sellers A. & Kennedy P. J. 1993. Biosphere Atmosphere Transfer Scheme (BATS). Version 1e as coupled to the NCAR Community Climate Model. NCAR Tech. Note NCAR/TN- 387+STR, 72 pp. Goyette S., N.A. McFarlane, & G. Flato, 2. Application of the Canadian Regional Climate Model to the Laurentian Great Lakes Regions. Implementation of a Lake Model. Atmos. Ocean., 38: 481-3. Hostetler S.W., Bates, G.T., & Giorgi, F, 1993. Interactive coupling of a lake thermal model with a regional climate model, J. Geophys. Res., 98: 4-7. Laprise R. 28. Regional climate modelling J. Comp. Phys. 227: 3641--3666. Irbe, G.J. 1992. Great Lakes surface water climatology. Environment Canada, AES, Climatological Studies 43, 21 pp. Martynov A., L. Sushama, R. Laprise: "Simulation of temperate freezing lakes by one-dimensional lake models: performance assessment for interactive coupling with regional climate models" Boreal Env. Res. 1: 143 164 Mironov, D., Heise, E., Kourzeneva, E., Ritter, B., Schneider, N. & Terzhevik, A. 21: Implementation of the lake parameterisation scheme FLake into the numerical weather prediction model COSMO. Bor. Env. Res. 1: 218-23 Stepanenko, V. M., Goyette, S., Martynov, A., Perroud, M., Fang, X. & Mironov, D. 21: First steps of a Lake Model Intercomparison Project: LakeMIP. Boreal Env. Res. 1: 191-22. VerseghyD. L., McFarlaneN. A., and LazareM., "CLASS-A Canadian land surface scheme for GCMS, II. Vegetation model and coupled runs," Int. J. Climatol., vol. 13, no. 4, pp. 347-37, 1993. Thank you! Questions?