Folie 1

1
Regionalization of climateprojections
andclimate impact modelingin GLOWA-IMPETUS COSM
OS General Assembly 2009, Berlin, 16. June 2009
Andreas H. Fink, Institute of Geophysics und
Meteorology, University of Cologne, Cologne,
Germany
2

• GLOWA-IMPETUS
• An Integrated Approach to the Efficient
  Management of Scarce Water Resources in West
  Africa
• Case studies for selected river catchments in
  different climate zones
• www.impetus.uni-koeln.de
• Joint project of the Universities of Cologne and
  Bonn
• 11 institutes, 17 PIs, 15 Post Docs, 30 Ph.
  Ds., 30 Diploma students,
• about 27 Mill. for three project phases
  (2000-2009), ca. 90 funding from the federal
  Ministry for Education and Research,
• GLOWA Global Change and the Hydrological Cycle

3
Choice of Catchments
Wadi Drâa
Morocco
Ouémé
Benin
Source IMPETUS
4
Physical map and annual rainfall(Upper and
Middle Drâa, 28428 km2)
Source Schulz, 2008, IMPETUS Atlas
Source Fritzsche, 2008, IMPETUS Atlas
5
Natural Environment (Upper and Middle Drâa)
A. Fink
A. Fink
Upper Drâa Assif n-Ait Ahmed, river oasis,view
west towards the Mgoun massif (3996 m)
Middle Drâa Saharan Hamada Semi-desert in
theDjebel Bani foreland, Acacia trees in the
middle ground

• arid to sub-humid
• headwaters only one perennial tributary
• heterogonous geologic setting (High Atlas,
  sedimentary basins, Anti-Atlas)
• sparse vegetation cover (acacia, shrubs,
  juniperus trees), only dense in the oasis

A. Fink
Upper Drâa River Oasis, Ait Benhaddou, Basin of
Ouarzazate
6
Motivation

• The Maghreb hot spot of Global Change
• Present situation
• Moderate population growth (1.5)
• Intensive irrigation and inefficient use of
  groundwater
• Water scarcity 934 (443) m3/a/capita in Morocco
  (Algeria)
• Sinking groundwater table salt water intrusion
  in river mouths (e.g. Oued Souss near Agadir)
• In the Atlas Mountains Overgrazing and erosion
• Strong natural rainfall variability
• Future climate
• Warming trend very likely, drying likely
  (IPCC 4AR)

7
Rainfall Variability in Northwest Africa
NAO-G explains about 50 of the annual rainfall
variability (1900-2002)
Cut-off and sfc cyclones over the western
Mediterranean Sea, no correlation to NAO
Cut-off and sfc cyclones near the Canary Islands,
VE by NAO 10
Source Fink and Born 2009, IMPETUS
Source after Knippertz et al. 2003, IMPETUS
8
Rainfall Variability in Northwest Africa
IMPETUS finding Up to 40 of the annual rainfall
south of the Atlas Mountains is associated with
Tropical-Extratropical Interactions
Tropical Plume, 22.10.2003 12 UTC abundant
rains from Senegal to the Maghreb set the stage
for the locust outbreak in 2004
Further reading on TEIKnippertz (2003,
MWR)Knippertz and Martin (2005, QJRMS) Knippertz
and Fink (2006, PROMET)
Source Fink and Born 2009, IMPETUS
Source after Knippertz et al. 2003, IMPETUS

• Investigating historical data and identifying
  (regional) processes that drive the regional
  climate change are two pillars of regionalization

9
Third Pillar Global Climate Models
IPCC-AR4 21 global GCMs with radiative forcing
(2080-2099) minus (1980-1999)
warming all over Africa by 2.5-4C, largest
warming inthe deserts
ECHAM5 with radiative forcing
overall drying in North Africa, wetterin the
inner Tropics

large model uncertainty over Sub-Saharan
Africa,consensus dryingin North Africa
ECHAM5 is consistent with AR4 multi-model ensemble
mean
Source IPCC 4th AR, Christensen et al. 2007
10
Fourth Pillar Methods of Regionalisation

• Global Scale
• 200 km, century
• ECHAM5/MPI-OM1
• Continental Scale
• 50 km, several
• decades
• REMO
• Regional Scale
• 7-28 km, single years
• LM
• Local Scale
• 1-7km, 24-48h
• FOOT3DK

Impact Analysis ?
Source IMPETUS
11
REMO Forcings
REMO 0,5
Increase in greenhouse gases
A1B B1
Changes in land use
Source Paeth et al. 2009
12
REMO Climate Projections until 2050

• prominent warming and drying in sub-Saharan Africa

• land degradation is primarily responsible for
  the drying

Source Paeth et al. 2009
13
REMO validation
Source after Heiko Paeth, IMPETUS

• Realistic simulation of key characteristics of
  the North African climate, however
• Precipitation biases MOS-corrected for all impact
  assessments
• Temperature bias (in East Africa) MOS-corrected
  for Malaria modeling

14
Statistical-dynamical regionalisationWeather-typ
e recombination
after K. Piecha, IMPETUS
15
Statistical-dynamical regionalisationWeather-typ
e recombination
Pressure data forCWT classification
Input data for FOOT3DK
after K. Piecha, IMPETUS
16
Statistical-dynamical regionalisationWeather-typ
e recombination
Re-combined FOOT3DK annual rainfall 2002
COSMO LM annual rainfall 2002
after K. Piecha, IMPETUS

• In a highly structured terrain with few
  observations dynamic downscaling with
  weather-type recombination yields acceptable
  results

17
Circulation Weather Types Present-day (1986-2000)
and future (2036-2050)
Source Piecha, IGM Cologne
An increase in the Circulation Weather Types in
the NE through to S sector is projected is
projected for the Northwestern Maghreb.
18
Statistical-dynamical regionalisationWeather-typ
e recombination
FOOT3DK Forcing Control, A1B scenario
Source K. Piecha, IGM Cologne

• An increase of the circulation weather types in
  the sector NE to S causes an increase in rainfall
  south of the Atlas Mountains and in the
  Anti-Atlas
• A model confirmation of the process understanding
  and observational hints Out-of-phase rainfall
  signal north and south of the Atlas due to TEI

19
The IMPETUS Approach of regional downscaling
20
The IMPETUS Approach of regional
downscalingThe Information System SMGHydraa
REMO
Information System Statistical Model for the
Generation of meteorological data for
hydrological modeling in the Drâa region
Author K. Born, IGM Cologne
21
The concept of zones
Nine aridity zones

• Considerations for the applicability suggest data
  reduction
• Weather/climate data are
• aggregated in (arbitrary) zones
• Choice of zones problem-oriented or as climatic
  similar
• regions (aridity index)
• Selectable temporal resolution
• (days, months, years)
• Daily values of
• rainfall,
• near-surface temperature,
• Humidity and
• Wind
• are preprocessed for the integration into the
  IMPETUS framework

Catchments
Source K. Born, IGM Cologne
22
The IMPETUS Approach of regional
downscalingThe Information System SMGHydraa
REMOless rain on more rainy days in the High
Mountain zone
REMOcorrection using multiple regression of the
height dependencies of annual rainfall, annual
rainy days and precipitation per wet day
Author K. Born, IGM Cologne
23
The IMPETUS Approach of regional
downscalingThe Information System SMGHydraa
Only few reporting stations at lower altitudes
were used (blue) in the regression. Using
6-years data from the GLOWA-Impetus stations, we
could just assess if the regression is meaningful
at higher altitudes
Author K. Born, IGM Cologne
24
The IMPETUS Approach of regional
downscalingThe Information System SMGHydraa
Author K. Born, IGM Cologne
25
SPI time series extreme value analysis from
kernel estimation
Tendency extreme dry years occur more often in
the future
Source Born et al. 2008 Cologne
Climate Change!
26
The IMPETUS Approach of regional
downscalingThe Information System SMGHydraa
Born et al. 2008
27
Climate model data for impact studies
impact studies
climate models
statistical calibration
model output statistics (MOS)
hydrology
food agriculture
weather-type recombination
land use
weather generator
health
I M P E T U S
Source IMPETUS
28
The Impetus Climate Scenarios
Scenario X Process understanding Scenario Y
Transient climate model predictions including
statistical-dynamical downscaling Scenario Z
Persistence of recently observed trends
? State-of-the art regionalization of climate
change and post-processing of climate model
data for use in impact modeling
29
(No Transcript)
30
High precipitation (Schulz 2006)
Baseflow Cappy 2006
Steep slopes Shallow soils (A. Klose 2008)
Snow DynamicsSchulz 2006
Stream-Aquifer InteractionsCappy 2006
31
Little but highly variable precipitation Born et
al. 2008 Schulz et al 2008
Gentle slopes Variable soil properties
Irrigation
Routing of mountain discharge Transmission losses
32
SWAT

• 378 HRUs
• Unique combinations of
• soil, land cover and slope

Reservoir
Irrigation Cropwat
After Arnold et al. 1993
33
Calibration and Validation
Source Busche Diekkrüger, IMPETUS
34
Water balanceUpper Drâa 1978-2007
Source Busche Diekkrüger, IMPETUS
35
Source Busche Diekkrüger, IMPETUS
36

• Baseline Period
• 1978-2007
• Scenario
• 2001-2030
• 0.8C
• -10 Precipitation

Source K. Born, IGM Cologne
37
Scenario setup
2 climate scenarios (A1B,B1) X 3 ensembles X 4
downscaling approaches 24 scenarios
Source Busche Diekkrüger, IMPETUS
38
Exceedance probability for discharge into
reservoir larger than 300 Mm³/year (250 Mm³
irrigation requirement 50 Mm³ evaporation
losses) Baseline 48 Scenario 31
Source Busche Diekkrüger, IMPETUS
39
Large uncertainty in boundary condition
Strongest signal Highest certainty
Source Buche Diekkrüger, IMPETUS
Slight shift in discharge composition
Small relative increase in Evapotranspiration
40
Conclusions

• The importance of conducting climate measurements
  in elevated regions has been highlighted.
• SWAT has been adapted to local conditions
• The present system state is modeled successfully
• Climate change scenario analysis indicates a
  decrease in snowfall and runoff
• The threshold of 300 Mm³ discharge to the
  reservoir is likely to be exceeded in only 31 of
  the years under climate change conditions in
  contrast to 48 in the reference period

41
Conclusions

• An urgent demand for integrated modeling of the
  impacts of global change
• GLOWA-IMPETUS has developed about 30 computerized
  Decision Support or Information Systems in
  various sectors
• The impact models are loosely coupled (data
  exchange) to the post-processed output from
  regional climate models
• Within GLOWA-IMPETUS the present four pillars
  (IPCC AR4) of regionalization have been used
• The full possibility of uncertainty estimates of
  climate projections could not be explored (e.g.
  multi-model approach)
• Lending more credibility on the climate
  projections always requires validation data and
  process studies

42
Thank you for your attention
Shower cloud over the Atlas, seen from
Ouarzazate, 26.10.2006, photo courtesy of K.
Born, IGM Cologne