Transcript On the way to Ensemble Hydrological Forecasts: Lessons Learned from MAP D-PHASE Mathias W Rotach1, Marco Arpagaus1, Manfred Dorninger2, Christoph Hegg3, Andrea Montani4, Roberto Ranzi5 1Federal Office for.

On the way to Ensemble
Hydrological Forecasts:
Lessons Learned from
MAP D-PHASE
Mathias W Rotach1, Marco Arpagaus1, Manfred
Dorninger2, Christoph Hegg3, Andrea Montani4,
Roberto Ranzi5
1Federal
Office for Meteorology and Climatology, MeteoSwiss, Zurich,
of Vienna, Vienna, Austria, 3WSL, Swiss Federal Institute for Forest
Snow and Landscape Research Birmensdorf, Switzerland, 4ARPA-SIMC,
Bologna, Italy, 5University of Brescia, Brescia, Italy
2University
Outline
• What is D-PHASE?
• End users
 Their involvement & participation
 Feedback (questionnaires…)
• Outreach to real applications
 a business case
 an operational case
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
D-PHASE essentials
•
•
•
•
Forecast Demonstration Project for MAP
2nd WWRP Forecast Demonstration Project (FDP)
Focus on heavy precipitation and flood forecasting
D-PHASE: Demonstration of Probabilistic
Hydrological and Atmospheric Simulation of
flood Events in the Alpine region
• D-PHASE Operations Period (DOP):
June to November 2007 (COPS & “MAP season”)
• 9 countries involved
• 30 atmospheric models/ 7 hydrological models in
over 40 catchments
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
BAMS Paper, Sept 2009
Key elements of D-PHASE
• Centralised Visualisation Platform (forecasts &
alerts; in real-time);
• Data archiving ( research / analysis);
• Nowcasting tools;
• Systematic integration of end users;
• Evaluation, objective and subjective.
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Users
• WWRP JSC asked for ‘at least one’
 over fourty....
 forecasters (atmospheric & hydrological)
 COPS (WWRP RDP) mission planning
 end users: civil protection, water management, ...
• Quality
 feedback
 transfer to operations
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Forecaster / end user feedback
• Forecasters (MeteoSwiss):
(Fill in feedback forms every day)
 ‘everything’ to Visualisation Platform?
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
~
Forecaster / end user feedback
• Forecasters (MeteoSwiss):
(Fill in feedback forms every day)
• ‘everything’ to Visualisation Platform?
• ensemble prediction systems
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
~

Question III.3: Additional benefit of ensemble
models as compared to established det. models?
additional benefit of the ensemble prediction systems
(wet days)
100
+
0
–
Frequency [%]
75
CLEPS (n = 125)
50
PEPS (n = 48)
25
0
added value
no added value
poorer guidance
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
empty
Forecaster / end user feedback
• Forecasters (MeteoSwiss):
 everything to Visualisation Platform?
 ensemble prediction systems
 high-resolution deterministic models
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
~


Question III.2: Additional benefit of hires models as
compared to coarser ones?
additional benefit of the higher resolution models
(wet days)
100
+
0
–
Frequency [%]
75
COSMOCH2 (n = 93)
COSMOCH7 (n = 129)
50
CLEPS (n = 125)
PEPS (n = 48)
25
0
added value
no added value
poorer guidance
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
empty
Forecaster / end user feedback
• Forecasters (MeteoSwiss):
 everything to Visualisation Platform?
 ensemble prediction systems
 high-resolution deterministic models
~


 appreciate multitude of similar systems
But NOT too many!
 Interaction with hydrological forecasts
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)

Forecaster / end user feedback
• End users:
• workshops before / after DOP
• O(30) participants
• 16 answered both questionnaires
• no ‘significant’ results (n too small)
Frick and Hegg, in prep
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Judgments about information
manageability
comprehens. progress diagr. level 3
comprehens. model plots level 3
model information can be integrated
background of model info clear
Understand information
possible to draw conclusions
Need guidance
interpretability
model information trustworthy
amount of information (too much)
relevant information detectable
adequate area size
-2
-1
negative
0
neutral
1
2
positive
N=24
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Frick and Hegg, in prep
Appraisal of benefit
The Visualisation platform ....
supports evaluation of situations
Supports situation
analysis
supports decision making
was incorporated
into decisions / measures
is important
for decisions / measures
availability of ensemble
improves decisions
under time pressure proceeded as usual
Action: business as usual
in concrete situation proceeded as usual
-2
-1
0
disagreement
1
2
agreement
N=24
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Frick and Hegg, in prep
End user feedback
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
End user feedback
• Take end users early on board
• Resolution needs to further increase (and will)
• EPS’s need careful support (for interpretation)
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
• ‘Ensemble thinking’ for air pollution modelling, heat wave
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
warnings, health impact (e.g., pollen), …
Outreach to applications
Business case:
Construction of a new train station in Zurich
 requires river duct partially be closed
Sihl at Zürich HB, June 16th 2009
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Photo: A. Ludwig
Outreach to applications
• Cost of opening the gates: 1 Mio CHF
• Damage if not opened: many billion CHF
• Flood wave to station: 2-6 hours
• Evacuation construction site: 2-4 hours
• Opening one gate: 1-2 hours
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Relative value
Relative Value
Total cost
Economic value of forecast
cost / loss ratio
param. conv.
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
resolved conv.
Outreach to applications
D-PHASE inspired information platform in Switzerland:
 combines all natural hazards
 D-PHASE platform up and running on user request
to bridge time to completion
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Conclusions
For an FDP to be successful….
• Involve users
 early!
 feedback
• Commitment ‘pays back’
 most efforts / most rewards
• society profits
 warning systems
 business enabling …
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
Thanks!
ARPA
Piemonte
LUBW
ARPASIMC
AEMet
Uni
Wien
ZAMG
APAT
WWA
Kempten
DLR
MeteoSwiss
CNMCA
ETHZ
Uni
Brescia
ARPA
Liguria
Uni
Hohenheim
ARPA
Veneto
IMK-IFU
Uni Paul
Sabatier
POLIMI
ARPA
Lombardia
Env
Canada
WSL
Met
Office
BAFU
ISACCNR
Météo
France
SRNWP
Steering Committee & WG chairmen
Operational
Service
University
Res. Inst.
D-PHASE lessons learned | TTISS Monterey CA, 14-18 September 2009
Mathias Rotach (mathias.rotach [at] meteoswiss.ch)
DWD
Data
Provider