What is a climate model?

• Substitutes for reality• Closely mimics some

essential elements• Omits or poorly

mimics non-essential elements

What is a Model?

What is a Model?

• Quantitative and/or qualitative representation of natural processes (may be physical or mathematical)

• Based on theory• Suitable for testing “What if…?” hypotheses• Capable of making predictions

1. Energy from the Sun(energy from the interior)

2. Planetary Albedo

3. Speed of Planet’s Rotation

4. Mass of the Planet

5. Radius of the Planet

6. Atmospheric Composition

7. Ocean-Land, Topography

S (depends on Sun itself and distance from Sun)

M

a

H2O, CO2, O3, clouds

h*

The Climate of a Planet Depends On …

Example: Energy BalanceModel

Solar Radiation

S = 1361 Wm-2

(plane, parallel)

In equilibrium,

INCOMING ENERGY = OUTGOING ENERGY

(1 - ) S a2 = E (4 a2)

E = 1/4 (1 - ) S

Measured albedo () = 0.30Measured planetary E = 238 Wm-2

Implied TE = 255 K

(Note: Water freezes at 273 K)

Planetary Emission

This is a VSCM: Very Simple Climate ModelExperts prefer a GCM:Global Climate Model(General Circulation Model)

Earth’s Energy Balance

Solar Radiation

S = 1361 Wm-2

(plane, parallel)

Assume radiative equilibrium, so that

INCOMING ENERGY = OUTGOING ENERGY

(1 - ) S a2 = E (4 a2)

E = 1/4 (1 - ) S

Measured albedo () = 0.30Measured planetary E = 238 Wm-2

Implied TE = 255 K

Planetary Emission

Measured surface Es = 390 Wm-2

Atmosphere absorbs 152 Wm-2

Measured Ts = 288 K

WHY??The Greenhouse Effect

… But it’s a little more complicated than that …

CLIMATE DYNAMICS OF THE PLANET EARTH

S

Ω

a

g

T4

WEATHER

CLIMATE .

hydrodynamic instabilities of shear flows; stratification & rotation; moist thermodynamics

day-to-day weather fluctuations; wavelike motions: wavelength, period, amplitude

T_

y,U_

y

T_

z,U_

z

S, , a, g, ΩO3

H2OCO2

stationary waves (Q, h*), monsoons

h*: mountains, oceans (SST)w*: forest, desert (soil wetness)

(albedo)

Climate System

Theory

Discretization

Climate System

Theory

Discretization

(approximation)

Mass conservation

Energy conservation

Newton’s law

= p / ps

Climate System

Theory

Discretization

Equations of motions and laws of thermodynamics predict rate of change of:

T, P, V, q, etc. (A, O, L, CO2, etc.)

Climate System

Theory

Discretization

Equations of motions and laws of thermodynamics predict rate of change of:

T, P, V, q, etc. (A, O, L, CO2, etc.)

Discretization

Atmosphere and ocean are continuous fluids … but computers can only represent discrete objects

Discretization

Atmosphere and ocean are continuous fluids … but computers can only represent discrete objects

• Equations of motions and laws of thermodynamics to predict rate of change of:

T, P, V, q, etc. (A, O, L, CO2, etc.)

• 10 Million Equations: 100,000 Points × 100 Levels × 10 Variables

• With Time Steps of: ~ 10 Minutes

• Use Supercomputers

What is a Climate Model?

Moore’s “Law”

IPCC-1

IPCC-2

IPCC-3

IPCC-4

103-fold jump since 1st IPCC106-fold jump in last 30 years

Latest advance due to dual-core chipsNear-term advance w/quad-core chips

Johnvon Neumann

Seymour Cray& Cray-1

ENIAC

IBM 360

Cray-2

ColumbiaNASA

Climate Models circa early 1990s Global coupled climate models in 2007 and new ESMs

New decadal prediction modelsGlobal coupled models in 5 yrs post-AR5

~500 km ~100 – 200 km

~50 km ~10 km

We ran 7-km grid on 640 nodes (2560 cores), because constrained by memory per core on Athena … more grid bisections means more “ghost rows” means more memory demand

NICAM Domain Decomposition

1990 1996

2001 2007

The complexity of global climate models has increased enormously over the last 20 years, as shown in this flow chart. Beneath each time period is a list of the components included in state-of-the-art models such as the NCAR-based Community Climate System Model (Warren Washington, NCAR)

Ultimate: all physico-biogeochemical Earth System

Balancing future demands on computing power

Duration and/or Ensemble size

Re

so

luti

on

ComputingResources

Complexity

1/120

EO, Data Assimilation

Model Grid Size (km) & Computing Capability

Peak Rate: 10 TFLOPS 100 TFLOPS 1 PFLOPS 10 PFLOPS

100 PFLOPS

Cores1,400(2005)

12,000(2007)

80-100,000(2009)

300-800,000(2011)

6,000,000?(20xx?)

Global NWP0: 5-10 days/hr

18 - 29 8.5 - 14 4.0 - 6.3 1.8 - 2.9 0.85 - 1.4

Seasonal1: 50-100

days/day17 - 28 8.0 - 13 3.7 - 5.9 1.7 - 2.8 0.80 - 1.3

Decadal1: 5-10 yrs/day

57 - 91 27 - 42 12 - 20 5.7 - 9.1 2.7 - 4.2

Climate Change2:

20-50 yrs/day120 - 200 57 - 91 27 - 42 12 - 20 5.7 - 9.1

Range: Assumed efficiency of 10-40%0 - Atmospheric General Circulation Model (AGCM; 100 vertical levels)1 - Coupled Ocean-Atmosphere-Land Model (CGCM; ~ 2X AGCM)2 - Earth System Model (with biogeochemical cycles) (ESM; ~ 2X CGCM)

* Core counts above O(104) are unprecedented for weather or climate codes, so the last 3 columns require

getting 3 orders of magnitude in scalable parallelization

Thanks to Jim Abeles (IBM)