Tailored scenarios for streamflow
climate change impacts based on
the perturbation of precipitation
and evapotranspiration
Victor Ntegeka, Patrick Willems
KU Leuven, Hydraulics section
Pierre Baguis, Emmanuel Roulin
Royal Meteorological Institute of Belgium
Research problem
When studying the hydrological impacts of climate change:
The number of available climate change simulations grows
Ensemble of climate model outputs
GCMs: CMIP3
(SRES based)
Belgium, 2005 ‐ 2010:
CMIP3 (SRES) based:
• 44 runs with 21 global climate models
• 57 runs with 10 regional climate models
(PRUDENCE, ENSEMBLES)
European RCMs:
PRUDENCE, ENSEMBLES
Results for
Belgium
Ensemble of climate model outputs
GCMs: CMIP5
(RCP based)
Belgium, 2015:
CMIP5 (RCP) based:
• >200 runs with 34 global climate models
• 22 runs with regional climate models
(EURO‐CORDEX)
• Few runs with local area models (ALARO, CCLM, 3‐4km resolution)
European RCMs:
CORDEX
Belgian
LAMs
Research problem
The number of available climate change simulations grows
 Need to develop few synthesized scenarios
 Research hypothesis: these synthesized scenarios depend
on the specific (type of) impact application
Tailored climate scenarios
(impact-centric scenarios instead of
climate-centric scenarios)
Proposed tailoring process
Step 1: Obtain climate change signals from all available
climate model runs
Climate change signals
>200 GCMs CMIP5 (RCP based) for Belgium: change for 100 years (2000 -> 2100):
Change in mean monthly rainfall:
%
Climate change signals
>200 GCMs CMIP5 (RCP based) for Belgium: change for 100 years (2000 -> 2100):
Change in mean monthly rainfall:
%
Climate change signals
>200 GCMs CMIP5 (RCP based) for Belgium: change for 100 years (2000 -> 2100):
Change in wet day frequency (per month):
%
Climate change signals
>200 GCMs CMIP5 (RCP based) for Belgium: change for 100 years (2000 -> 2100):
Change in extreme rainfall quantiles:
Climate change signals
30 RCMs for Belgium: change for 100 years (2000 -> 2100):
Change in extreme rainfall quantiles:
GPD extreme value distribution
calibration:
Direct forcing approach
Bias correction
Historical
series
GCM
today
RCM
today
Historical
series
Correction factors
Rainfall-runoff model
today
Climate
change
impact
= control run
scenario
scenario
climate change
scenario
= scenario run
Climate system
Hydrological system
“Delta approach”
or perturbation approach
or MOS approach
GCM
RCM
Rainfall-runoff model
today
today
= control run
scenario
scenario
Perturbation
factors
Historical
series
today
Historical
series
+ perturbation
climate change
scenario
= scenario run
Climate system
Hydrological system
Model Output Statistics (MOS)
Perturbation = Change
e.g. Perturbation factors / Change factors / Climate factors
Climate
change
impact
Quantile perturbation method
Combined spatial and temporal downscaling and bias correction
Water system
response forcing
Water system
Change factors depend on:
 Temporal & spatial scale
 Return period
 Time of the year (month,
season)
 …
for:
 Rainfall intensity
 Rain storm frequency
 Dry spell duration
 …
Ntegeka, V., Baguis, P., Roulin, E., Willems, P. (2014), ‘Developing tailored climate change scenarios for
hydrological impact assessments’, Journal of Hydrology, 508C, 307-321
Climate perturbation tool
•
•
•
•
Perturbs historical series to high/mean/low climate scenarios
Time scales: daily, hourly, 10-minutes
Based on quantile perturbations:
o
change in rain storm frequency and rain storm intensity
o
dependent on return period and season
Time horizons till 2030, 2050, …, 2100
Wet day frequency
perturbation
Wet day intensity
perturbation
Combined perturbation
Daily
Hourly
10min
Time
series
Month i
Month i
Month i
Time
series
Ntegeka, V., Baguis, P., Roulin, E., Willems, P. (2014), ‘Developing tailored climate change scenarios for
hydrological impact assessments’, Journal of Hydrology, 508C, 307-321
Result for selected catchment
Molenbeek catchment (Erpe-Mere, Belgium):
Example:
Impacts on hourly river peak flows vs. return period:
high
Winter
Perturbation [-]
1.8
Individual
RCM based
scenarios
1.4
1
low
0.6
EM
WH initial
WH tailored
WL initial
WL tailored
RCMs
0.2
1
10
Return period [years]
Ntegeka, V., Baguis, P., Roulin, E., Willems, P. (2014), ‘Developing tailored climate change scenarios for
hydrological impact assessments’, Journal of Hydrology, 508C, 307-321
100
Proposed tailoring process
Step 2: Check the hydrological impacts of the climate change
signals from selected climate model simulations
(based on a simplified (computationally less expensive)
impact model)
“Back-tracking” of extreme flows from driving scenarios
: which meteorological input combinations give the highestmean-lowest hydrological impacts ?
Proposed tailoring process
Step 3: Inter-seasonal tracing of rainfall changes, but also of
rainfall – T/ETo changes
High – river flood scenario?
Winter
Summer
High R and High ETo
14
1.4
Rainfall perturbation [-]
Rainfall perturbation [-]
High R and Low ETo
15
6
1.2
10
13
1
18
5
22
20
74
1
21
Low R and Low ETo
0.8
1.2
ETo perturbation [-]
High R and Low ETo
High R and High ETo
25
6
4 7
0.8
13
10
22 5
15
1
21
20
0.6
14
18
25
Low R and Low ETo
Low R and High ETo
1
1
1.4
1.
Low R and High ETo
0.4
0.8
1
1.2
ETo perturbation [-]
1.4
Proposed tailoring process
Step 3: Inter-seasonal tracing of rainfall changes, but also of
rainfall – T/ETo changes
High – sewer flood scenario?
Winter
High R and High ETo
Rainfall perturbation [-]
Rainfall perturbation [-]
High R and Low ETo
Summer
1.4
1.2
11
10
12
7
9
23
8
1
High R and Low ETo
High R and High ETo
23
11
79
0.8
10
8 12
0.6
Low R and Low ETo
1
Low R and Low ETo
0.8
Low R and High ETo
Low R and High ETo
1
1.2
ETo perturbation [-]
1.4
0.4
1.6 0.8
1
1.2
ETo perturbation [-]
1.4
Proposed tailoring process
Step 4: Obtain tailored scenarios for studying impacts
Our focus: impacts on hydrological extremes (high & low
flows rivers & urban drainage):
4 tailored scenarios considered:
• HW (High/wet Winter): high river flood scenario
• HS (High/conv. Summer): high sewer flood scenario
• LS (Low/dry Summer): extreme low flow scenario
• EM: Ensemble Mean/mild scenario
Result for selected catchment
Molenbeek catchment (Erpe-Mere, Belgium):
Example:
Impacts on hourly river peak flows vs. return period:
Winter
high
Perturbation [-]
1.8
Individual
RCM based
scenarios
1.4
1
low
0.6
EM
WH initial
WH tailored
WL initial
WL tailored
RCMs
0.2
1
10
Return period [years]
Ntegeka, V., Baguis, P., Roulin, E., Willems, P. (2014), ‘Developing tailored climate change scenarios for
hydrological impact assessments’, Journal of Hydrology, 508C, 307-321
100
Statistical downscaling
GCMs
Large ensemble: CMIP5
RCMs
Reduced ensemble:
EURO-CORDEX
LAMs
Statistical
downscaling
incl. bias
correction
local impact
models
Climate change signals
: fine vs. coarse resolution climate models?
CCLM model runs: Van Lipzig et al. (2015) KU Leuven
Conclusions
We moved from climate-centric scenarios to impactcentric scenarios:
• development of few synthesized scenarios that reveals full
uncertainty range (available climate model runs)
• based on careful examination of inter-seasonal P-T/ETo
changes and “back-tracing” of impact results: design
climate scenarios that facilitates specific type of impact
studies (hydrological extremes in our case)
More info…
Ntegeka, V., Baguis, P., Roulin, E., Willems, P. (2014), ‘Developing tailored climate
change scenarios for hydrological impact assessments’, Journal of Hydrology, 508C,
307-321
Willems, P., Arnbjerg-Nielsen, K., Olsson, J., Nguyen, V.T.V. (2012), ‘Climate change
impact assessment on urban rainfall extremes and urban drainage: methods and
shortcomings’, Atmospheric Research, 103, 106-118
Willems, P. (2013). ‘Revision of urban drainage design rules after assessment of climate
change impacts on precipitation extremes at Uccle, Belgium’, Journal of Hydrology, 496,
166–177
Willems P., Vrac M. (2011), ‘Statistical precipitation downscaling for small-scale
hydrological impact investigations of climate change’, Journal of Hydrology, 402, 193–
205
Tabari, H., Taye, M.T., De Troch, R., Termonia, P., Saeed, S., Brisson, E., Van Lipzig, N.,
Willems, P. (2015), ‘Local impact analysis of climate change on precipitation extremes:
Are high-resolution climate models needed for realistic simulations?’, In preparation
[email protected]