Transcript Providing seamless seasonal to centennial projections for

Providing seamless seasonal to
centennial projections for health
impacts of climate change
Andy Morse
School of Environmental Sciences,
University of Liverpool, Liverpool, U.K.
[email protected]
Earth System Science: Global Change, Climate and People,
AIMES Open Science Conference, Edinburgh May, 2010
Thanks to: Cyril Caminade and Anne Jones, School of Environmental Sciences,
University of Liverpool, Liverpool, U.K.; Matthew Baylis, School of Veterinary
Science, University of Liverpool; Helene Guis, CIRAD, Montpellier, France.
Background, Methods and Results, Discussion
Themes
• Reflections on a decade+ of end-to-end (-to end) modelling
• Simple thoughts on climate, disease and model integration
• Ensemble prediction, malaria models and towards being
seamless from seasons to decades
•
• Future integration of ESM dynamic surface fields
Background, Methods and Results, Discussion
Over A Decade of End-to-End Research
•
•
•
•
•
•
•
•
•
•
•
Integrated climate impacts with model outputs major European climate
modelling centres, ECMWF, The Met. Office and Metéo France
range of applications at number research institutes, many in Africa
FP4 PROVOST – data used
FP5 DEMETER – led impacts groups – seasonal EPS
FP6 ENSEMBLES – co-led impacts section - EPS and RCM
FP6 AMMA; NERC AMMA linked to FP6 ENSEMBLES
NERC EQUIP decadal prediction
FP7 NERC ERA-NET ENHanCE co-lead and FP7 QWeCI coordinator
Initial condition multi-model ensemble predictions (probabilistic):
days to decades Climate Variability. Seasonal scales.
Climate projections - GHG driven global climate models & RCMs
multi-model Climate Change.
Developed skill base/team/network to extract useful information and integrate
with impacts models (and society) for health (food security and water).
Climate Services agenda.
Background, Methods and Results, Discussion
Introduction
• Climate variability important component determining incidence number of diseases
(vector-born especially) with significant human and animal health impacts.
• Observed and simulated climate datasets drive models and map the risk of key relevant
(animal) and human diseases for the recent past, to verify seasonal scale hindcasts and to
project them into the future based on climate simulations.
• Other important non-climatic factors also need to be considered in the disease modelling
approach.
• Current few if any disease model have realistic land surfaces and ESM gives an
opportunity to develop this aspect of the modelling for future projections
Background, Methods and Results, Discussion
Background, Methods and Results, Discussion
Background, Methods and Results, Discussion
Integrated Climate Model Impacts Verification Paradigm
from Morse et al. (2005)
Tellus A 57 (3) 464-475
Introduction, Methods and Results, Discussion
Seasonal Ensemble Prediction
Introduction, Methods and Results, Discussion
Malaria Prediction Plume
Malaria
Malaria
Prevalence
Prevalence
0.45
0.45
0.4
0.4
0.35
0.35
0.3
95
0.3
0.25
0.25
0.2
65
85
35
15
0.2
0.15
0.15
0.1
5
ERA
0.1
0.05
0.05
0
0 1
1
31
31
61
91
121
61 Forecast
91 Day 121
151
151
Forecast Day
Botswana malaria forecast
for February 1989
LMM (Hoshen and Morse, 2004) driven by DEMETER multimodel 63 members
(ERA-driven model shown in red)
Introduction, Methods and Results, Discussion
Seasonal Forecasts –decision making contexts
1.0
Event observed
H
Event forecast
0.5
Yes
No
Yes
Hit (a)
False alarm (b)
No
Miss (c)
Correct
rejection (d)
H
0.0
0.0
0.5
1.0
a
ac
F
b
bd
F
F orecas t probability
0.7
‘User defined’ Decision threshold, P
0.6
0.5
0.4
0.3
0.2
0.1
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
0
DEMETER multi-model-driven malaria forecasts for above upper tercile malaria, Botswana, November
forecast months 4-6 (FMA), compared to observed anomalies from 1982-2001 published index.
Introduction, Methods and Results, Discussion
Potential Seasonal Skill in Epidemic Zones for Malaria
Based on the Liverpool Malaria Model simulations driven by seasonal ensemble
multi-model outputs (Rainfall and Temperature)
ENSEMBLES Seasonal EPS May 4-6 (ASO) upper tercile
epidemic transmission zone ROCSS
Introduction, Methods and Results, Discussion
Mean Annual Malaria Modelled Incidence 1990-2007
Endemic areas >80%
“Endemic and seasonal”
areas between 20-80%
Epidemic Areas (<20%)
-> Northen fringe of the
Sahel
-> Strongly connected to
climate variability
Underestimation of the
Northern extension of
the malaria incidence
belt by LMM
ITCZ extends too far
north in the RCM world
Mean annual simulated malaria incidence (1990-2007) driven by
“Observed datasets” and the ENSEMBLES RCM ensemble
Introduction, Methods and Results, Discussion
Shift of the epidemic belt 2031-50 vs 1990-2010
Grey: Location of the
epidemic belt 1990-2010
Black dots: Future location of
the epidemic belt 2030-2050
The epidemic belt location is
defined by the coefficient of
variation, namely:
Mean Incidence > 1%
1stddev > 50% of the average
Southward shift of the
epidemic belt over WA
-> to more populated areas...
Introduction, Methods and Results, Discussion
Earth System Model integration with disease modelling
•
•
•
•
•
Disease models shown here have no realistic land surface
Dynamic vegetation for vector habitats?
Realistic vegetation to constrain temperature cycle/ ranges?
Realistic surface hydrology forvector and parasite life cycle.
Away from ESM – models of society and social systems – ESM
integrated with large agent based modelling???
• How do we build seamless systems FP7 QWeCI (months to
decades)
Introduction, Methods and Results, Discussion
Summary
• Demonstrated disease model – seasonal ensemble
prediction system integration and impact verification
• Will most diseases respond to climate change or just a
few?
• Is it possible that the diseases ‘that matter most’ are the
least likely to respond to climate change?
• Society will change +/- disease threat
• The use of ESM inputs to improve future disease
projections?
QWeCI
FP7 SEVENTH FRAMEWORK PROGRAMME THEME ENV.2009.1.2.1.2
Methods to quantify the impacts of climate and weather on health in
developing low income countries
Collaborative Project (small- or medium scale focused research project) for
specific cooperation actions (SICA) dedicated to international cooperation
partner countries
Quantifying Weather and Climate Conditions on health in developing
countries (QWeCI)
3.5 MEu EC contribution (~4.7MEu total) 1st Feb 2010 start
13 partners = 7 Africa, 6 EU, Liverpool coordinator, 42 months
UNILIV, CSE, CSIC, ECMWF, IC3, ICTP, ILRI, IPD, KNUST, UCAD,
UNIMA, UOC, UP