A [simple] land cover change intercomparison

1
A simple land cover change intercomparison

• A. Pitman, R. Betts, R. Pielke Sr. et al.

2
Background

• LCC affects 45 of the terrestrial surface
  (Vitousek et al., 1997)
• likely an underestimate (Williams, 2003)
• Globally distributed but regionally centred

3
Background

• Deforestation experiments demonstrate an impact
  on regional climates
• But some are now attributing large changes in
  climate remote from LCC to LCC via
  teleconnections
• Mechanisms include Walker and Hadley cell changes
  and Rossby wave propagation

4
Chase et al, 2000
5
Betts, 2000
6
Status

• The IPCC (2001) notes possible regional impact of
  LCC
• some are interpreting GCM results as evidence of
  the global scale impact of LCC
• Others see LCC only in terms of radiative impacts
• Some see any remote effects of LCC as model
  variability.
• either might be true - but it is something that
  we need to know more confidently.

7
Status

• There are problems with the design of all
  attempts to explore the climate impact of LCC
  using GCMs
• Many use short (lt20-year) simulations for natural
  and current vegetation
• Most perform single realizations
• Many perform standard t-tests that do not account
  for the autocorrelation in the data
• Spatial resolution tends to be quite coarse.

8
Proposal

• A LCC intercomparison involving 10-15 groups
  with
• a common land cover perturbation (historical land
  cover to current). We might do a future scenario
  too
• AMIP-2 length simulations, using the AMIP-2
  design
• multiple realizations with each model (5-10)
• use appropriate statistics to determine whether
  there are regional impacts of LCC.

9
Proposal

• a common land cover perturbation (historical land
  cover to current, but we might do a future
  scenario too)
• Crops other Betts/de Noblet
• 1900 and 2000 snap-shots
• Static vegetation
• Modellers free to translate changes into pfts
• Future scenario not decided

10
Proposal

• AMIP-2 length simulations, using the AMIP-2
  design
• Fixed SSTs
• Limiting relevance but cheap and easy inclusive
• Easy for most groups
• AMIP-2 standard output format (easy)
• We need to recognise that the set of people who
  are pushing LCC as a major climate driver have
  limited overlap with core climate modelling
  groups limits the level of experimental
  complexity that is possible.
• It is more politically important to include these
  groups that have a larger sample of core climate
  modelling groups.

11
Proposal

• multiple realizations with each model (5-10)
• Advice from GLASS appreciate on the number
  required
• Advice welcomed on best way to perturb the sample

12
Proposal

• use appropriate statistics to determine whether
  there are regional impacts of LCC
• Again, advice encouraged.

13
Timeline

• We wanted to mesh with IPCC not possible
• Review paper from the community
• Data sets available by November/December 2004
• Simulations performed by October 2005
• Analysis over the subsequent six months.
• data will be made available to individual groups

14
Objectives

• We do not aim to answer the LCC question
• We aim to start a process if the LCC community
  conduct these experiments and the answers are
  interesting, we have a common foundation to build
  from
• Our experiments are limiting but we have to
  balance what is achievable by the specific
  community we are trying to involve
• If GLASS thinks the experiments are too limiting
  then we would prefer to know now !

15
Questions

• Is this worth doing ?
• relatively cheap, but it is limited in scope
• too slow for IPCC 4th assessment
• would force some to confront model variability
  cf. teleconnection issue
• Is AMIP-2 ok as a framework ?
• Advice on the LCC data ?
• Realizations ?
• Statistics ?