13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Mauro del Longo, Andrea Montani, Tiziana Paccagnella, Silvano Pecora and Giuseppe Ricciardi
ARPA Emilia Romagna - Italy
Use of the COSMO-LEPS ensemble for hydrologic forecasts in the Warning Operational Center of Emilia
Romagna (Italy)
www.arpa.emr.it
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
A combination of probabilistic meteorological inputs and scheduling of hydrologic/hydraulic models within the system give us a lot of information needing a more complex OPERATIONAL interpretation.It can be useful to fulfill the recent ITALIAN CIVIL PROTECTION law linking forecast activity and probable risk scenarios. A conservative approach can consist in choosing the heaviest rain, the highest peak or the earliest one, or the maximum discharge volumes forecast; however, i.e. due, in some cases, to underdispersivity of probabilistic meteorological forcasts, it can not eliminate all missed alarms while it can generate too many false alarms.
An alternative approach can consist in combining all informations in real time, to discern among different forecast scenarios and to give each scenario a “Subjective weight” (subjective forecast), expecially by mean of COSMO LEPS.
Aim of this work
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Introduction
Meteorological products (Cosmo suites - Leps, I7, I2 N2-RUC)
Hydrological and hydraulic products (Mike-NAM/HD, HEC-HMS/RAS, Topkapi/Sobek)
Scheduling and operational use
Operational System
Activities
Methodology
Case studies
Conclusions
Spatial Resolution
Temporal Resolution
Runs per day Lead time
Cosmo -LEPS/LM-
DET7 km 3 h 2 120 h
CosmoI7 7 km 1 h 2 72 h
CosmoI2 2.8 km 1 h 2 48 h
RUC 2.8 km 1 h 8 18 h
Hydrological model Hydraulic model
Mike11 HD-NAM
Lumped modelAverage basin area about
100km2
11.000 sections100- 500 m
1D, “Quasi” 2D
HEC HMS-RASLumped model
Average basin area about 100km2
11.000 sections100- 500 m
1D, “Quasi” 2D
Topkapi-Sobek Distributed modelGrid size 250m or 500m
11.000 sections100- 500 m
1D, “Quasi” 2D
Mike11 NAM-
HDHEC HMS-
RASTopkapi-Sobek
Cosmo -LEPS/LM-DET
Every 6 hours17 ensemble
members runs in parallel
Every 12 hours17 ensemble
members runs in parallel
Every 12 hours17 ensemble
members runs in parallel
CosmoI7 Every 3 hours Every 3 hours Every 3 hours
CosmoI2 Every hourNot yet
implementedNot yet
implemented
RUC Every hourNot yet
implementedNot yet
implemented
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
COSMO-LEPS (run at ECMWF) COSMO Consortium
d-1d-1 dd d+5d+5d+1d+1 d+2d+2 d+4d+4d+3d+3
older EPSolder EPS
younger EPSyounger EPS
clustering clustering periodperiod
0000
1212
Cluster Analysis and RM identificationCluster Analysis and RM identification
4 variables4 variables
Z U V QZ U V Q
3 levels3 levels
500 700 850 hPa500 700 850 hPa
2 2 time time stepssteps
Cluster Analysis and RM identificationCluster Analysis and RM identification
European European areaarea
Complete Complete LinkageLinkage
16 Representative Members driving the 16 COSMO-model
integrations (weighted according to the cluster
populations)
Using either Tiedtke or Kain-Fristch convection scheme
(members 1-8 T, members 9-16 KF)
+Perturbations in turbulence
scheme and in physical parameterisations
COSMO-LEPS
clustering area
• suite runs twice a day (00 and 12UTC) as a “time-critical application” managed by ARPA-SIMC on behalf of COSMO consortium;
• Δx ~ 7 km; 40 ML; fc+132h;• COSM0 v4.26 since January 2013;• computer time (30 million BUs for
2013) provided by the ECMWF member states in COSMO.
COSMO-LEPS
Integration Domain
Montani AndreaNWP4 Room 102Friday, 13 Sep
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Hydrological/hydraulic modellingMike NAM-HD11 Hec HMS-RAS Topkapi - Sobek
Elements of modeling system FEWS Observation network
Modeling tools and suites
Discharge field measurements
Monitoring network :water level gauges (blue triangle)
raingauges (green dots)thermometers (green dots)
dams (violet)
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
From:• time of forecast• comparison among models• consistency• scales• season • meteorological phenomenaWe define the variability and reliability of inputs to hydrological-hydraulic chains
The reliability judgment on models and the use of ensembles give informations about the uncertainty
Meteorological
Hydrological From:• reliability of hydrological-hydraulic chains• comparison among outputs • catchment, river network and hydraulic devices conditions • hydrological phenomena• post correction, manual forecast• judgment on outputs • warning level
Probabilistic Bulletin
Activities
We define the variability and reliability of hydrologic flood forecast
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
We present the activity of the Flood Warning Center of Emilia Romagna on some events occurred in the last year and we propose a methodology for a deeper analysis.
Methodology
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction
Members exceeding thresholds
Time at threshold exceeding / Peak time Threshold exceeding duration
All members median of maximum levels
Maximum level range
Present activityWe look at 5-day forecast to identify:
• the probability of exceedance of a hydrometric threshold
• the number of members exceeding the threshold
• the forecast consistency in the prediction at the ranges +5, +4, …. +1day
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Time (hours) 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 126 132 138 144 150 156
Large Catchment
Small Catchment Ensemble informations (Leps)
Coupled Forecast (Leps/I7)Observed simulation
forecast/I2/I7Observed Simulation/Upstream discharges/RUC
Ensemble Forecast (Leps)
Observed Simulation/RUC
Coupled Forecast
(Leps/I2)Coupled followup (Leps/I7)
Date 10-11/11/2012624 and 694 km2 Lead time 120 h
Date 10-11/03/2013Catchment: 694 km2 Lead time 120 h
Date 14-20/05/2013Catchment: 41.000 km2 Lead time 144 h
LARGE Lead time: 48 - 120 hSpatial scale: 30.000 - 5.000 km2
REGIONAL Lead time: 24 - 48 hSpatial scale: 5.000 -1.000 km2
MULTICATCHMENT Lead time: 12 - 24 hSpatial scale: 1.000 - 500 km2
SINGLE CATCHMENT Lead time: 6 - 12 hSpatial scale: 500 - 200 km2
Case studies
CASE3
CASE2
CASE1
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Case 1: Po at Piacenza
Case 2/3: Parma at Ponte Verdi and Secchia at Lugo
GEOGRAPHICAL LOCALIZATIONS OF CASE STUDIES
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
0
1
2
3
4
5
6
7
8
9
10
wat
er le
vel (
m)
Observed level since 16th to 22th at Piacenza
Total precipitation 15 may 00:00 - 20 may 00.00: 135 mm
Partial precipitation 15 may 00:00 17 may 12.00 80 mm18 may 12:00 19 may 12:00 45 mm
Forecast Discharge 5.500- 6.000 m3/s
Peak time 18th may from 00:00 to 12:00Observed
Discharge 5.300 m3/s Peak time 18th may at 10.00
Case 1: Po at Piacenza
Meteosat image on European-Atlantic area on 17th May 2013 at 08:00 LTC
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Piacenza 2013 05 13
0.1
0.5
0.8
0.90.95
0.0 2.0 4.0 6.0 8.0 10.0
1- P
h [m]
2013 05 13 PiacenzaMaximum level frequency analysis
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Piacenza 2013 05 14
0.1
0.5
0.8
0.90.95
0.0 2.0 4.0 6.0 8.0 10.0
1- P
H [m]
2013 05 14 PiacenzaMaximum level frequency analysis
Case 1: Po at Piacenza
Consistency of the ensemble forecast (red dashed spaghetti plot) from 13th to 16th May show:-the consolidation of the meteorological dynamic system -the rotation and reduction of thickness of the band of threshold exceedings (see left diagrams)
Observed flood (continuous blue line) peak on 18th May h max = 5.96 m Second observed peak on 21st May h max = 6.50 m
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Piacenza 2013 05 15
2013 05 15 Piacenza Maximum level frequency analysis
0.1
0.5
0.8
0.9
0.95
0,0 2,0 4,0 6,0 8,0 10,0
h [m]
1- P
2013 05 16 Piacenza Maximum level frequency analysis
0.1
0.5
0.8
0.90.95
0,0 2,0 4,0 6,0 8,0 10,0
h [m]
1- P
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Piacenza 2013 05 16
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Case 2: Parma at Ponte Verdi and Secchia at Lugo
Total site precipitation from 10th to 11th Nov 2012Parma basin: Bosco di Corniglio station 180 mmSecchia basin: Ospitaletto station 183 mm
Reflectivity maps from operational ARPA-SIMC radars on 11th Nov 2012 05.30 UTC (up) 06.00 UTC (down)
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
Case 2: Parma at Ponte Verdi and Secchia at Lugo
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Lugo 2012 11 08
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Ponte Verdi 2012 11 08
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Ponte Verdi 2012 11 09
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Lugo 2012 11 09
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Ponte Verdi 2012 11 09
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Lugo 2012 11 09
Deterministic forecast driven by Cosmo I7 0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Ponte Verdi 2012 11 10
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Lugo 2012 11 10
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Ponte Verdi 2012 11 10
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
Hydrological Ensamble Prediction at Lugo 2012 11 10
Deterministic forecast driven by Cosmo I7
From 8th to 10th Nov:• the variability of the signal decreases • the peak values first increase and than decrease• peak timing is consistent
Observed flood peak on 11th Nov 11:00 LTCParma h max = 3.18 m Q max = 500 mc/sLugo h max = 2.40 m Q max = 535 mc/s
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 08 Ponte VerdiMaximum level frequency analysis
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 08 Lugo Maximum level frequency analysis
From 8th to 10th Nov shift and slight reduction of thickness of the band. The number of threshold exceedings after a first increase falls down0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 09 Ponte VerdiMaximum level frequency analysis
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 09 Lugo - Maximum level frequency analysis
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 10 Lugo - Maximum level frequency analysis
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 10 Ponte Verdi Maximum level frequency analysis
Case 2: Parma at Ponte Verdi and Secchia at Lugo - FREQUENCY ANALYSIS
10th Nov: Ponte Verdi 0 members exceed L3
10th Nov: Lugo 1 members exceed L2
Deterministic and probabilistic meteorological analisys are not sufficient.What if we add a probabilistic hydrological analisys?
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
wat
er le
vel (
m)
Sobek -Hydrological Ensamble Prediction at Ponte Verdi 2012 11 10
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 10 Ponte Verdi Sobek Maximum level frequency analysis
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
wat
er le
vel (
m)
Sobek Hydrological Ensamble Prediction at Lugo 2012 11 10
0.1
0.5
0.8
0.90.95
0.0 1.0 2.0 3.0 4.0 5.0
1- P
h [m]
2012 11 10 Lugo Sobek Maximum level frequency analysis
10th Nov: Ponte Verdi 9 members exceed L3
Cosmo I7 (green line) driven run still underestimates
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
wat
er le
vel (
m)
Sobek -Hydrological Ensamble Prediction at Ponte Verdi 2012 11 10
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
wat
er le
vel (
m)
Sobek Hydrological Ensamble Prediction at Lugo 2012 11 10
Parma at Ponte Verdi and Secchia at Lugo - SOBEK ANALYSIS at + 48 hours
10th Nov: Lugo 7 members exceed L2
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
HEP at Ponte Verdi 2013 03 13
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
HEP at Ponte Verdi 2013 03 12
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
wat
er le
vel (
m)
HEP at Ponte Verdi 2013 03 11F
orec
ast
con
sist
ency
b
etw
een
+72
an
d +
96 h
ours
11th March
12th March
13th March
Case 3: Parma at Ponte Verdi – FALSE ALARM
13th March abrupt dump of the signal without threshold exceedings
11th March forecast gives threshold exceedings between 14th and 15th March
12th March forecast consistency with fair fading
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
CASE1Deterministic
Ensemble m1,m2….
Observed
CASE2: Deterministic
Ensemble m1,m2….
Observed
CASE3
Deterministic
Ensemble m1,m2…. Observed
Conclusions1_Joint use of deterministic and ensemble forecast helps to reduce MISSED ALARM compared to the deterministic only case (e.g. Case 2).
2_Ensemble predictions providing thresholds exceeding several days in advance, even for small basins (e.g. Case 2 and 3), gives forecasters MORE TIME TO UNDERTAKE ANALYSIS, useful for false alarm case (e.g. Case 3).
3_Coupling the meteorological ensemble with a hydrological/hydraulic multi model gives forecasters more information too. Only the PROFESSIONAL SKILL can give a “subjective weight” to the forecast.
4_It is therefore necessary to know the limits of forecasting systems. Verification through statistical analysis and comparisons between forecast from skilled operator and observations are essential to test and improve the quality of forecasts, allowing a “DIAGNOSTIC” on hydrometeorological models and a self assessment on forecast products.
13th EMS Annual Meeting & 11th European Conference on Applications of Meteorology (ECAM) | 09 – 13 Sep.2013 | Reading, United Kingdom
ASI – AS19 Interfacing hydrological and meteorological models in forecasting systems (Monday, 9 September 2013)
To test and improve forecast quality we are going to extend the presented work:
• building the forcasted hydrogram setting a set of constraints and separately
analizing each variable (e.g. peak time, peak discharge …)
•using other events
• implementing statistical analysis
• analyzing QPF, discharges, time at threshold exceeding and duration
• comparing subjective forecasts and observations
• considering performance indicators
Future step
Thank you for attention
100th Anniversary of Italian Hydrographic Istitute
www.arpa.emr.it