Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
ECOLMAS Training Course – Introduction to climate modelling Bremen, 1-4 April 2008 Part 3: The Atmosphere Ute Merkel ECOLMAS Course Bremen 2008 – Part 3 Atmospheric modelling Storm "Emma", March 2008 (British Weather Service, 2008) Storm "Emma", March 2008 (British Weather Service, 2008) Weather versus Climate X Initial conditions at t=0 (almost identical, not perfectly known) Exponential growth of the error (i.e. the difference between the trajectories starting from almost identical initial conditions) The weather prediction does not make sense beyond a time certain time range (~ 15 days for the atmosphere) because the deviations between the trajectories are of the same order of magnitude as the change in X). (adapted from V. Moron, 2003) Atmospheric modelling - General atmospheric circulation Weather versus climate Climate is what we expect, weather is what we get. Larry Riddle Atmospheric modelling - General atmospheric circulation Introduction to atmospheric circulation Averaging with respect to longitude (zonal mean) zonal mean meridional circulation zonally asymmetric circulation stationary waves Averaging with respect to time time mean circulation transient circulation / eddies λ = longitude φ = latitude t = time Atmospheric modelling - General atmospheric circulation Introduction to atmospheric circulation thermally direct cells (Univ. Wales, Bangor) Atmospheric modelling - General atmospheric circulation Introduction to atmospheric circulation pressure (hPa) Annual mean Hadley Cells (NCEP reanalysis 1948-2001) from the meridional stream function (1010 kg/s) latitude (Liu and Alexander, 2007) ECOLMAS Course Bremen 2008 – Part 3 Atmospheric modelling Introduction to atmospheric circulation Zonal Walker circulation along the equator ⇒ low-level convergence and associated upward motion (convection) (Fig. 6.22, Hartmann, 1994) Atmospheric modelling - General atmospheric circulation July, boreal summer Introduction to atmospheric circulation Near surface wind fields and pressure systems Position of the ITCZ January, boreal winter (The Dynamic Earth, Fig. 2.18) General circulation: Monsoons January, boreal winter July, boreal summer (The Dynamic Earth, Fig. 2.19) Atmospheric modelling - General atmospheric circulation Introduction to atmospheric circulation Tropics: large-scale overturning by mean meridional circulation Extratropics: baroclinic eddies (cyclones and anticyclones with associated warm and cold fronts) and stationary waves Atmospheric modelling - General atmospheric circulation General circulation: Extratropics Jetstream at 10 km height • Meandering jet • wave-like structure at a maximum • polar air reaches lower latitudes and tropical air is advected to higher latitudes Atmospheric modelling - General atmospheric circulation General circulation: Zonally asymmetric circulation ⇒ Jets show clear deviations from zonal symmetry * separation of air masses * definition of tropics ? Atmospheric modelling - General atmospheric circulation General circulation: Heat and momentum transports • Atmospheric circulation provides import contributions to meridional heat and momentum transports to compensate for latitudinal gradients between tropics and polar regions. • Atmospheric circulation imposes momentum forcing to the ocean. Atmospheric modelling - General atmospheric circulation General circulation: Summary • Meridional gradients in solar insolation are the main driver for atmospheric circulation. • Hadley, Ferrel and polar cells (zonal mean meridional structure) • Important contribution to the structure of the atmosphere: Coriolis force (large-scale, trop. storms) • Important role of the Earth's rotation and the land sea contrasts Atmospheric modelling - General atmospheric circulation Why modelling? ⇒ Provide hypotheses on how mechanisms in the climate system are operating ⇒ better understand large-scale relationships in the atmosphere (e.g. teleconnections) and its interaction with other climate subsystems (ocean, land, ice sheets...) ⇒ better understand the different time and space scales, how they are interacting and superposed as shown by observations Atmospheric modelling - Motivation Time and spatial scales of the African monsoon (AMMA) Atmospheric modelling - Motivation Hierarchy of models From simple to comprehensive models • Energy Balance Models • Earth System Models of Intermediate Complexity (quasi-geostrophic approach, no humidity,...) • Atmospheric General Circulation Model (full dynamics and physical processes represented) • Coupled Atmosphere-Ocean Circulation Models Atmospheric modelling - Complexity of models From simple to comprehensive models • Include all fundamental processes • Resolve all spatial dimensions (Ruddiman, 2001) Atmospheric modelling - Complexity of models From simple to comprehensive models: Processes included in ECHAM3 model (DKRZ Report, 1993) Atmospheric modelling - Complexity of models Which processes are taken into account? Example: Radiation (UW Atmospheric Sciences) Atmospheric modelling - Complexity of models Which processes are taken into account? Example: cloud feedbacks Cloud-Albedo Feedback Cloud-Greenhouse Feedback – + (http://www.worc.ac.uk/LTMain/Rowland/mec/climate/Feedback/Cloud.html) What are atmospheric models based on? - Primitive equations • Conservation of energy (1st law of thermodynamics) – temperature • Conservation of momentum – horizontal velocity (wind, circulation) • Conservation of mass (continuity equation) – vertical velocity • Equation of state – ideal gas law Atmospheric modelling - Complexity of models Complex atmosphere models Example: The ECHAM model (MPI for Meteorology in Hamburg) • global general circulation model of the atmosphere • based on the ECMWF model for medium-range weather forecast → modifications and improvements for applications in climate research • prognostic variables: vorticity, divergence, temperature, logarithm of air pressure, specific humidity, mixing ratio of total water content in clouds • 19 levels • horizontal resolutions T21, T31,T42, T63, T106, T159,.. Atmospheric modelling - Complexity of models (McGuffie and Henderson-Sellers, 1997) Vertical coordinate system • Assumption of hydrostatic balance (i.e., ∆p = -ρg∆z) height (z) expressed in terms of pressure (p) • Pressure normalized to surface pressure (σ; terrain-following) • Troposphere and lower stratosphere (<20 km) usually represented (Hartmann, 1994) Atmospheric modelling - Spectral method and resolution Spectral method • Global atmospheric fields can be represented in terms of spherical basis functions • Similar to the use of trigonometric functions such as sines or cosines See Washington and Parkinson (1986), Chapter 4, pp. 18. Atmospheric modelling - Spectral method and resolution Spectral method and model resolution ● X : divergence, temperature, vorticity,...: represented in the model by a truncated series of spherical harmonics m = zonal wave number n = meridional index ● in ECHAM5 only triangular truncation can be done (implied by the parallelization of the model's spectral part) ● truncation done at a certain wave number (typically T21, 31, 42, 63, 85, 106, 159,...) Atmospheric modelling - Spectral method and resolution Spectral representation ADVANTAGES • Easy and exact spatial differentiation • Natural description of planetary waves in unbounded domain • Homogenous resolution on a sphere DISADVANTAGES • Transformations become computationally inefficient at high resolution • For any truncated basis function expansion, there is overshooting and undershooting (Gibbs phenomenon) • Gibbs phenomenon occurs near steep gradients – yields negative values of mass and humidity – makes representation of mountain ranges or ice sheets difficult Atmospheric modelling - Spectral method and resolution Fourier theorem • The actual shape of a vibrating string can always be represented as an infinite series of eigenvector basis functions: ∞ f ( x) = ∑ Ψ n ( x ) , n =! where 2 nπ x Ψn = sin . l l Atmospheric modelling - Spectral method and resolution Spectral representation: Gibbs phenomenon 4 1 1 f ( x ) = sin x + sin 3 x + sin 5 x + K 3 5 π Example of the Gibbs phenomenon (overshooting and undershooting) for a step function (Figure 4.6 from Washington and Parkinson,1986) Atmospheric modelling - Spectral method and resolution Land sea masks for the ECHAM model ~ 600 km ~ 300 km ~ 200 km ~ 120 km T31 simulations about 25 times faster than T106 Atmospheric modelling - Spectral method and resolution Horizontal model resolution ECHAM4 model topography over Europe [m] T42 T106 (Merkel, 2003) Atmospheric modelling - Spectral method and resolution Model resolution • Grid representation leads to the fact that some sub-grid scale processes cannot be fully simulated in their overall complexity => transfer of radiation => phase changes of water vapour => turbulent transports • Parameterizations (based on theoretical and observational considerations) take into account the impact of these processes onto model variables (via simplified functions of fully resolved model variables). Atmospheric modelling - Spectral method and resolution Role of model resolution: Stormtrack activity T42 control simulation [gpm] T106 control simulation [gpm] Root-mean-square (RMS) of bandpass-filtered (2.5-6 d) 500 hPa geopotential height data ⇒ Role of horizontal resolution (Merkel, 2003) Role of vertical model resolution Normalized RMS errors relative to T21 error (Ratio of RMS error of Txx to RMS error of T21) ERA40 w.r.t. ERA15 19 vertical levels 31 vertical levels Atmospheric modelling - Spectral method and resolution (Roeckner et al., 2006) Computational scheme of a spectral AGCM (McGuffie and Henderson-Sellers, 2005) Transformation to grid space samples field around zones of latitude and longitude Each atmospheric field held and moved in spectral space (“wave functions”) Spectral truncation restricts information Vertical exchange in grid space Each surface field held in grid space Surface fields are computed in grid (McGuffie and Henderson-Sellers, space 1997) Initialization • climatological values • previous model runs (restart run) • Spin-up: How long does it take for the atmosphere model to reach equilibrium? Typical climatological model runs with AGCM have a length of ~ 30-50 years - analysis of the last decades only Atmospheric modelling - Running the model Boundary conditions and forcings for an AGCM Orbital parameters Aerosols Greenhouse gas concentrations CH4 CO2 N2O Coupling to models of ocean and/or vegetation and/or sea ice.... Source: Montana State Univ. Sea level changes Continental ice sheets and albedo SST, Sea ice Source: SOEST, Hawaii Source: Scott Rutherford Atmosphere general circulation model Computing requirements (TerraFlops, 2002) Atmospheric modelling - Running the model Output of an AGCM • 2-d or 3-d distributions of state (“prognostic”) variables: – temperature – vorticity – divergence – .... • Many diagnostic variables, e.g.: – vertical velocity – clouds – SW radiation at top of atmosphere – LW radiation – snow depth – ... Atmospheric modelling - Running the model Output of an AGCM Annual mean precipitation [cm/yr] CCSM2 (~3.7°, 26 L) (M. Prange) Atmospheric modelling - Running the model Model performance How good are the models? Model results have to be compared to observations and paleo data ⇒ evaluate the model performance in reproducing - mean climate - climate variability Atmospheric modelling - Model performance Model performance Output of an AGCM as part of a CGCM Annual mean precipitation [cm/yr] CCSM2 (~3.7°, 26 L) Modern obs. (M. Prange) Atmospheric modelling - Model performance Model performance Surface air temperature [K] (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance Surface air temperature [K] (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance Precipitation [mm/day] (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance Precipitation [mm/day] (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance Temperature of troposphere and stratosphere [K] (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance Temperature of stratosphere [K] (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance - Sea ice North. hemisphere DJF extent South. hemisphere JJA extent (IPCC TAR, ch. 8) Atmospheric modelling - Model performance Model performance Tropics vs. extratropics Eastern tropical Pacific precipitation DJF (1979-1992) individ. experiments observations (Bengtsson et al., 1996) Atmospheric modelling - Model performance Model performance Tropics vs. extratropics Western Canada surface temperature DJF (1979-1992) (Bengtsson et al., 1996) Atmospheric modelling - Model performance