Transcript Analysis of Streamflow Characteristics over Northeastern Canada in

Analysis of Streamflow
Characteristics over Northeastern
Canada in a Changing Climate
O. Huziy, L. Sushama, M.N. Khaliq,
R. Laprise, B. Lehner, R. Roy
1
Outline
•
•
•
•
•
Motivation
Model and experiment description
Analysis methodology
Validation
Climate change results
– Ensemble mean approach
– Merged (long sample)
2
Motivation
• Better understanding of processes involved in
winter and spring flow regimes
• Compare the approaches of merged samples
and of ensemble mean for analysing climate
change signal in return levels of extreme
events
• Validate river model with observations
3
Experiment setup
CRCM4 – simulation domain
Driving data:
- CGCM3: 5 current members
(1970-1999) and 5
corresponding future members
(2041-2070)
- ERA40: reanalysis
CGCMv3,
ERA40
CRCM4 – configuration:
1) resolution – 45 km
2) land surface scheme – CLASS2.7 with 3 soil layers
* CRCM4 simulation results were provided by Ouranos team
CRCM4
Runoff*
Routing Scheme
WATROUTEmodified
4
Experiment setup
Driving data:
- CGCM3: 5 current members
(1970-1999) and 5
corresponding future members
(2041-2070)
- ERA40: reanalysis
CGCMv3,
ERA40
CRCM4
Runoff*
* CRCM4 simulation results were provided by Ouranos team
Routing Scheme
WATROUTEmodified
5
Methodology for calculating return
levels and assessing uncertainties
• GEV distribution is used to calculate return levels of
extreme events (PDF):
• Parameters of the distribution obtained by maximizing
the probability of the extreme values simulated by the
model (GML-method):
6
Validation with observations
Temperature compared to CRU dataset
SWE compared to the dataset from
Brown et al (2003)
Validation of streamflow
Observed
Modelled
Spring high flow
The observed data is provided by CEHQ.
Winter low flow
Streamflow (m3/s)
Modelled
Observed
8
CC: Mean values
Grid points with non-significant
changes at 95% confidence
level are shown in grey
9
CC: Timing of the high flow events
Change
(2041-2070)
(1970-1999)
Northern basins
Generally high flow events
tend to occur earlier
in future period
Southern basins
10
Modelled timings of high flow events
(mean over the ensemble) March-July
11
Changes to return levels of 15-day low flow
Changes to return levels of 1-day high flow
CC: Return levels of extreme events
(separate ensemble members)
12
CC: Return levels of extreme events
(merged samples)
High flow
Low flow
• There appear some significant
changes to the high-flow return
levels in the case of the merged
sample analysis.
• Since the changes are the same as
in the mean sample analysis, this
means that the uncertainties related
to the estimation of distribution
parameters have decreased.
• Changes to the return levels
corresponding to smaller return
periods (10 years) are more
significant than for the longer
return periods (30 years)
13
Conclusions
• The models reproduce reasonably mean
hydrographs and values of spring peaks.
• The model has difficulties to reproduce winter
flow, probably due to the absence of drainage in
the regions with near surface bedrock.
• It is shown that the longer samples can be more
reliable when assessing climate change signal in
high flow return levels.
• The studies with different GCMs and RCMs are
required to increase the confidence in the
obtained climate change results.
14
Thank you!
15