health - Potsdam Institute for Climate Impact Research

ISI-MIP: the health sector – 1st
intercomparison of malaria models
Franziska Piontek
Teams involved
• LMM_RO (Liverpool Malaria Model) Monthly
Model – Andrew Morse, Cyril Caminade (U
• MARA – Cyril Caminade (U Liverpool)
• VECTRI – Adrian Tompkins, Felipe Colón-González
(Abdus Salam International Centre for Theoretical
Physics, Trieste, Italy)
• UMEA Statistical Model – Joachim Rocklov, Hans
Stenlund (Umea University, Sweden)
• MIASMA Model – Pim Martens (U Maastricht),
Cyril Caminade (U Liverpool)
• 5 GCMs, 4 RCPs
• SSP2 population for people at risk metric
• present constant climate to isolate additional
effect of climate change
Available data in the data archive
• 0.5x0.5° spatial resolution, annual and
decadal data (Umea model: only decadal)
• cs = climate suitability  0 or 1 for each grid
• Length of transmission season  # of months
(VECTRI: # of days, Umea did not provide this
Paper also estimates additional population
at risk & additional person months at risk
• SSP2 population projections, gridded (first version)
• PAR: population present in area where climate is
suitable for malaria transmission (cs > 0.5)
• PMAR: length of transmission season x population in a
given grid cell
• PAR and PMAR attributable to climate change:
difference between sceanrio with climate change and
scenario with current climate (GCM-specific baseline
• 3 time periods: 2020s (2005-2035), 2050s (2035-2065),
2080s (2069-2099)
• Large spread between malaria model results
• Also affected by lack of agreement between
climate models
• Historical validation against observations – mixed
results between models, depends on assumption
of world with/without intervention on malaria
• Increase in climate suitability has limited meaning
for prediction of actual spread of malaria due to
missing socio-economic factors + questions of
actual spread of vector/pathogen
• Overall climate change effects likely to be small
Historical validation
Kovats at al.,
submitted to PNAS
Results: effect of climate scenarios on future malaria
distribution – changes in lts between 1980-2010 and
2069-2099 periods, mean of all GCMs
Kovats at al.,
submitted to PNAS
> 60% of model
Additional population at risk due to CC
impacts – spread over impact models
Kovats at
LMM_RO Monthly Model
• Input: monthly rainfall and temperature
• # of adult mosquitos in one month ~ to rainfall
in previous month
• Combine population with data on biting rate,
sporogonic cycle length (development of
infecting forms of malaria parasite in
mosquito), survival probability based on
monthly temperature  reproduction ratio R0
• Malaria transmission in that month if R0>1
MARA model
• Very simplified seasonal model of malaria
transmission, originally developed for Africa only
• Conditions for malaria occurrence:
3 month rainfall > minimum value
Catalyst month with rainfall > another minimum value
Temperature > threshold value
Seasonality index based on standard deviation of
monthly rainfall
 Malaria on or off
VECTRI model
• Dynamical model, daily timestep, accounting for subseasonal variations in climate – most sophisticated model
in the study
• Accounts for impact of temperature and rainfall variability
on mosquito in larval and adult stage as well as of parasite
• Rainfall effects on transmission represented by physicallybased model of surface pool hydrology – low rainfall 
increase in available breeding sites
• Accounts for human population density in calculation of
biting rates – dilution of parasite ratio for higher population
• Requires population migration to transport malaria to new
regions becoming suitable for transmission
UMEA Statistical Model
• Regression models empirically estimate
relation between endemic malaria
transmission and climatic factors on global
• Also takes into account socio-economic factors
• Models temperature effects on survival
probability and biting frequency of mosquitos
 calculate transmission potential as critical
vector density required for sustainable disease
• Additional assumption: minimum level of
monthly rainfall of 80 mm
• No constraint by current distribution of
malaria vectors

similar documents