An Introduction to Climate Modeling

1
An Introduction to Climate Modeling
Wednesday May 18, 2005
2
Overview

• What are climate models and why use them?
• Limitations of climate models
• Types of climate models and climate model
• components
• Sensitivity experiments with climate models
• Last Glacial Maximum simulations

3
What are climate models?

• First, what is a model ?
• A model is a simplified representation of a
  system designed to provide insight into how that
  system responds to change.

4
What types of systems?

• Can be traffic flow, fluid flow, factory
• productivity, economic trends, climate, etc.
• Some systems are inherently complex
• (e.g., atmosphere, climate), for instance
• - the atmosphere is a large (but thin!)
  compressible fluid on a non-uniformly heated,
  rotating sphere with varying surface
  characteristics (terrain and vegetation) and
  processes occurring on a wide range of temporal
  and spatial scales.

5

• Yet, atmospheric processes, like many others, can
  be described mathematically.
• Hence, climate models are mathematical
  descriptions of the climate system (atmosphere,
  ocean, sea ice, land surface, glacier).

6
Why use climate models?

• To make climate predictions.

7
Why use climate models?

• Much of the impetus comes from concern over
• rising greenhouse gas concentrations.
• Climate models provide predictions on how the
• climate system will respond to (for instance)
• increases in CO2 concentrations over time.
• Such investigations provide the basis for the
• Intergovernmental Project on Climate Change
• (IPCC) reports on the effects of increased
• greenhouse gases.

8
Limitations of climate models(How do we know
whether theyre right?)

• Since a climate model is a numerical
• representation of the climate system integrated
  on a computer, numerical errors (e.g.,
  truncation) will affect the solution.

9

• Since a climate model is a simplification of the
• climate system, it cant explicitly represent all
  
• processes that take place in the real system.
• We make some assumptions about processes that we
  dont yet understand. We parameterize some
  processes that are too complex (require too much
  computer time) or operate on relatively small
  time/space scales (e.g., thunderstorms).

10

• Parameterization - a simplified representation of
  an unresolved process as a function of other
  (resolved) variables. For instance, clouds are
  typically represented in terms of temperature and
  humidity.
• Other processes may be completely omitted from
  the model representation of the climate system.
• These and other factors contribute to error in
  the
• model solution.

11

• Confidence in a model prediction of future
• climate is largely built on the models ability
  to
• accurately simulate past climate. Hence, the
• importance of paleoclimate modeling.
• The interaction between those who make climate
• proxy observations (e.g., ice cores, tree rings)
  and those who make climate predictions (modelers)
  is
• mutually beneficial.

12
Types of climate models and climate model
components

• Climate models vary in complexity from very
• simple to highly complex. The choice of which
• climate model to use depends on multiple factors.

13

• From simple to most complex
• Energy Balance Models consider only surface
• temperature as a function of energy exchange.
• These can be zero-dimensional (energy balance of
• the whole earth), one-dimensional (consider the
• earth in terms of zonal bands), to
  two-dimensional
• (latitude vs. height, land vs. ocean). EBMs are
• computationally efficient, so long simulations
  can
• be run in a short time span.

14

• Statistical/dynamical use equations that
  describe
• changes in the climate system over a long period
• of time instead of explicitly resolving all of
  the
• processes that occur on shorter time scales.
• One example would be a simple atmosphere-land-ocea
  n model with dynamics, heat and moisture
• transport, etc., parameterized.
• SD models are also computationally efficient and
  could be coupled to an ice sheet model (for
  instance) for long simulations.

15

• Radiative Convective Models examine radiation
• processes in a vertical column of the model
• atmosphere. Temperature profile of the atmosphere
  is maintained by convective adjustments (air
  moving
• vertically).
• RCMs have been used to study the effects of
  aerosols, clouds, and changing greenhouse gases
  on temperature.

16

• General Circulation Models (atmosphere and
• ocean) most complex type of climate model.
• Explicitly represents many processes using
• dynamical (mathematical) equations for
  conservation
• of mass, energy, and momentum (motion).
• Earth is divided into grid boxes that extend
• vertically into the atmosphere, which is divided
• into multiple levels. Computations take place at
• each grid box.

17
(No Transcript)
18

• GCM resolution varies greatly, hence, so does
• computation time.
• Early GCMs had 8ox10o lat/lon resolution too
• coarse to represent Great Britain!
• Current GCMs are commonly 2.8o lat by 2.8o lon
  horizontal resolution (one grid box in Ohio) with
  18 to 26 vertical levels.

19

• Earth Simulator (Japan) -- global 10 km
  resolution, 96 vertical levels

20

• NCAR CCM3 (147456 grid points) coupled to a
• Land Surface Model running on the Cray SV1
  supercomputer takes 4 wall clock hours to
  complete one model year
• Regional climate model (788800 grid points) with
• North American domain running on a Linux
• cluster 50 days to complete one model year
• Higher resolution means higher computational
• cost!

21

• Even fine scale GCMs are too coarse to represent
  small scale processes like thunderstorms (10 km
  diameter), which are critical in the tropics for
  atmospheric circulation. Instead, these are
  parameterized.
• Climate model components of varying complexity
  can include atmosphere, ocean, biosphere,
  cryosphere (sea ice, snow, land ice). These
  simulate processes that
• work on different time scales. Often, the
  components are coupled asynchronously.

22
Coupling atmosphere and ocean multiple options
23

• Coupled models have been used to trace pathways
• of materials within the climate system (dust,
• isotopes, moisture) to determine the locations of
  
• source regions and how these change over time.

24
Sensitivity experiments with climate models

• Climate models are very useful for conducting
• sensitivity experiments (how does a change in one
  element of the system boundary condition affect
  the rest of the system?)
• Run a control for some number of years to
• equilibrium. Then change a boundary condition,
• re-run the model, and compare the experiment
• with the control.

25

• Boundary conditions for paleoclimate simulations
  include orbital forcing, trace gases, ice sheets,
  sea level, sea surface temperature, vegetation,
  and aerosols.

26

• Orbital variations since the last glacial period

27

• Orbital forcing and glaciation
• Proxy data indicate that sea level fell by 50 m
• from 115-105 kyr BP, consistent with the presence
  
• of continental ice sheets.
• Orbital configuration at 115 kBP leads to a 7
• reduction in summertime incoming solar radiation
• in the Northern Hemisphere - necessary to
  preserve snow cover and grow ice sheets.

28

• Rind et al. (1987) used a low resolution GCM.
  Results show that orbital forcing alone is
  insufficient to grow ice sheets. Only by
  specifying a 10 m thick ice sheet (higher
  albedo), lowering CO2, and cooling tropical SSTs
  could an ice sheet be maintained.

29

• Dong and Valdes (1995) -- able to get permanent
  snow cover only by lowering CO2 and predicting
  SSTs. The model predicted ice growth in northeast
  N. America and Fennoscandia, but also in Tibet,
  Siberia, and Alaska.

30

• Gallimore and Kutzbach (1996) -- Changes to
  orbital forcing, CO2, and albedo due to expanding
  tundra led to permanent snow cover at high
  latitudes. Their study demonstrates the
  importance of nonlinear feedbacks of
  vegetation/snow cover albedo on the climate
  system.

31

• Orbital forcing and monsoon
• Orbital changes in the early Holocene (9 kBP)
• led to increased seasonality and higher summer
• insolation in the NH relative to present day.

32

• Mitchell et al. (1988) -- simulated modern and 9
• kBP climates.
• Increased summer insolation in the
• Northern Hemisphere heats the continents
  generating low pressure and enhanced monsoon
  flow.
• The changes in monsoon precipitation are
  consistent with proxy evidence of elevated lake
  levels at low latitudes.

33
Note the warming over midlatitude North America
with largest values over land. Increased land/sea
temperature contrast enhances monsoon flow.
34

• Kutzbach et al. (1996) showed that orbital
• changes and vegetation/soil changes (run RVS
  below) produce more precipitation in tropical
  monsoon regions than orbital changes alone.

35

• Influence of continental ice sheets
• Elevated terrain, such as the Rockies and the
  Tibetan Plateau, alters atmospheric flow.

Mid-tropospheric trough downstream of elevated
terrain
Associated with lower-tropospheric temperature
anomalies (e.g. downstream cooling)
36

• Similarly, large ice sheets (e.g. the Laurentide
  Ice Sheet) at the LGM impacted the planetary
  albedo and atmospheric circulation patterns.
• Manabe and Broccoli (1985 a,b) -- used a low
• resolution GCM coupled to a mixed layer ocean
• model. Results show substantial atmospheric
• cooling downstream of the ice sheets. There is no
  
• SH cooling in the model until CO2 levels are
• lowered to appropriate values.

37
Note the coarse representation of land masses in
this early GCM.
38
Northern Hemisphere midlatitude jet stream
splits around the continental ice sheets.
39

• Rind (1987) -- presence of the ice sheets lowers
• the atmospheric temperature/moisture content.
• Topographic influence of the ice sheets results
  in
• a split midlatitude jet stream.
• Hall et al. (1996) -- show a uniform jet stream
• with the storm tracks along the ice sheet
  southern
• margins. Dramatic cooling occurs in the NH in
• the vicinity of the ice sheets resulting in
  positive
• mass balance.