Transcript Document

Scaling and scenarios:
the ghosts of data
past, present and future
Tim Mitchell
(CRU-past, TYN-present, ???-future)
?? Monthly or Extremes ??
The ‘data’ (past, present, future) here is exclusively
monthly.
Is it relevant to extremes?
Yes!
• monthly extremes are extremes too!
• monthly data may sometimes be used to estimate
daily extremes
• experiences with monthly data help to plan our daily
work
• many techniques are similar for monthly and daily
data
Of course, there are many contrasts between monthly
Overview
data
past
work
past
work
present
work
future
data
present
data
future
Work: past … Data: past
Monthly grids of observations (1901-2000)
•
•
•
•
work previously led by Mark New (now in Oxford)
climatology at 10 minutes, time-series at 0.5 degrees
global land only (excluding Antarctica)
tmp, pre, dtr, cld,
vap, (wet, frs)
• space-optimised
Work: past … Data: future
The ATEAM scenarios (2001-2100)
• ATEAM = EU 5th project on
ecosystems
• Europe (10 minute), global land (0.5
deg)
• 16 scenarios: 4 GCMs * 4 SRES
• tmp, pre, dtr, cld, vap
Overview
data
past
work
past
work
present
work
future
Monthly grids of
observations
(1901-2000)
data
present
data
future
The ATEAM
scenarios
(2001-2100)
Work: present … Data: past
Supplying users: sub-sample from March 2003
Bulgaria
verification of regional climate model outputs
USA
“sweeping predictions of climate change based on whimsical reconstructions from tree-rings here and
there around the world to upset our colleagues and neighbors”
Taiwan
quantifying climatic variations around East Asia
Slovenia
statistical downscaling and climate change scenarios for Slovenia
China
using our ecological model to investigate carbon cycle in China
Russia
quantifying changes in carbon stocks and emissions sinks from arable soils in European ex-USSR
Italy
a vegetation model’s carbon balance of mediterranean forests
Spain
the potential distribution of some tree species in Iberia
Denmark
distribution of forest herbs in Denmark and neighbouring countries
UK
a range of vector borne diseases worldwide, particularly malaria and dengue
Australia
to examine relationships between dengue fever outbreaks and temperature in Indonesia
UK
phenology of European butterflies
Portugal
the global space-time distribution of active fires
Philippines
the baseline climate of the Philippine region and its impact on the agricultural life of our people
USA
impact of climate shocks on African economic growth
Germany
development of sustainable pastoral and agropastoral livelihood systems in West Africa
Netherlands
adaptation to climate change in 7 basins
Work: present … Data:
present
Six-monthly updates to monthly grids
• Many users are working with non-climate data-sets that
are very short and very recent.
• Many users also need up-to-date information to feed into
the policy-making process.
• Therefore we are currently implementing a regular
(probably 6-monthly) update to the monthly grids.
• Flow of information:
CLIMAT bulletins (Met Office)  CRU databases 
grids
Work: present … Data: future
Developing the pattern scaling technique
devised ~1990, tested ~2000, extended ~now
• from linear change to stabilisation
• from changes in the mean to changes in variability
• from changes in the ‘usual’ to changes in ‘extremes’
Overview
data
past
data
present
data
future
work
past
Monthly grids of
observations
(1901-2000)
The ATEAM
scenarios
(2001-2100)
work
present
Supplying users Six-monthly
updates to
monthly grids
Developing the
pattern scaling
technique
work
future
Work: future … Data: past
Improving the grids
This is an endless task, but an important one, carried out
largely in the background and without explicit funding.
• Augment the station databases to improve coverage
outside the West and early in the 20th century.
• Check, re-check, and check again the temporal
homogeneity of the station time-series.
• A proposal to the EU (Clare G, Phil J) hopes to develop
similar daily databases, and high-resolution grids for
Europe.
Work: future … Data: future
Dreaming…
• Develop facility to produce user-specified scenarios ondemand [successor to Scengen, likely in 2003]
• Eliminate the mismatch in spatial scales between past
and future, by using RCMs. [awaits a large RCM sample]
• Introduce scenarios with changes in variability statistics.
[awaits the development of a robust theory]
• Introduce scenarios with credible month-to-month
variations derived from models, rather than observations.
[requires RCMs, and errors not to exceed ‘no change’
hypothesis]
Overview
data
past
data
present
data
future
work
past
Monthly grids of
observations
(1901-2000)
The ATEAM
scenarios
(2001-2100)
work
present
Supplying users Six-monthly
updates to
monthly grids
Developing the
pattern scaling
technique
work
future
Improving the
grids
Dreaming…
Overview
data
past
data
present
data
future
work
past
Monthly grids of
observations
(1901-2000)
The ATEAM
scenarios
(2001-2100)
work
present
Supplying users Six-monthly
updates to
monthly grids
Developing the
pattern scaling
technique
work
future
Improving the
grids
Dreaming…