Transcript BALTEX-KALME_16112009_Arheimer.PPT

Balt-HYPE:
a tool for high resolution hydrological
modelling of the Baltic basin
Assoc. Prof., Dr. Berit Arheimer
Head of Hydrological Research
Swedish Meteorological and Hydrological Institute
(SMHI)
Outline:
SMHI and the WFD in Sweden
Models for unmonitored waterbodies
The Balt-HYPE model
SMHI support
to water authorities
for implementation of WFD
Berit Arheimer
SMHIs role in the Swedish Water
Administration
 SMHI
shall play an active role in the WFD implementation.
 SMHI shall supply hydrometeorological information to meet the general needs
of the Swedish society. Such information must encompass the entire area of the
country, in a cost-effective manner, to pre-agreed quality assurance targets.
 The goal of SMHI in the WFD implementation is to create, manage and make
accessible information, data, modelling tools and their results at the individual
waterbody scale.
 Everything produced by SMHI for the Swedish water administration is free of
charge for all non-commercial users.
Berit Arheimer
SMHI support to water authorities / WFD
SVAR – Swedish Water Archive of hydromorphology,
physiography and statistics
PBD – National database of pressures and loads
New observation technology – 50 mobile hydrological
stations
WRAP - Reporting of local (campaign) measurements to
SMHI via the Internet
WQweb and SHARK – National monitoring data on
hydrology and status in coastal zone, respectively
Modelled time-series and status – Database including
modelled data of 20 000 waterbodies and 600 coastal zones
HOME Vatten – a management tool for water quality planning
Berit Arheimer
Theme side
www-beta.smhi.se
Models for predictions in
unmonitored basins
Not possible to measure everywhere!
Berit Arheimer
Models for predictions in unmonitored basins
20 000 fresh-water bodies and 600 coastal zones in Sweden
300 Water discharge
Forcing data:
300 Temperature, 800 Precipitation
Sweden = 450 000 km2
900 Nutrient conc.
Berit Arheimer
Models for predictions in ungauged
basins
5 water districts
20 000 fresh-water bodies and 600 coastal zones in Sweden
Fresh water
HYPE model:
Coastal zone
Probe-Scobi model:
• dynamic (daily)
• dynamic (15 min.)
• integrated water systems
• 1 m vertical resolution
• process-based
• mechanistic
• semi-distributed (HRU)
• 45 state variables
• water & chemistry
• water, chemistry, biology
Berit Arheimer
Models for predictions in ungauged
basins
Spatial fit: Long-term average (10 yrs)
Tot-N [ug/L]
Q [mm/år]
12000
1400
180
10000
Water discharge (m3/s)
500
1000
60
observation
1996
4000
2000
40
9.2
8.8
0
1500
0
Obs
20
2000
0
2000
4000
6000
8000
obs
observation
10000
8.4
1997 1998 1999 2000 2001 2002 2003 2004 2005
10
9.6
9.2
80
60
20
0
0
8.8
50
100
150
200
obs
observation
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
8.4
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Water discharge
(mm)
2000
600
Total Nitrogen
Total
Nitrogen
(μg/L)
Correlation: 0.96
Correlation: 0.94
16000
16002
Total Nitrogen
(μg/L)
NSE R : 0.92
NSE
R2: 0.88400
Tot-P
8000
600
Total Phosphorus
Total Phosphorus
(μg/L)
Correlation:
0.79
Total
Phosphorus
NSE
R2: 0.59 (μg/L)
400
200
Tot-P
12000
1200
Tot-N
100
40
12000
0
800
Lake water level (m)
120
mod
6000
model
mod
200
0
0
Water discharge (m3/s)
10.4
9.6
Sjövattenstånd (m)
200
140
Sjövattenstånd (m)
400
400
160
8000
model
600
80
600
Vattenföring (m3/s)
800
Tot-N
Mod
Vattenföring (m3/s)
1000
model
200
10
Temporal fit: Best model
performance
(S-HYPE)
Lake water level
(m)
800
1200
Tot-P [ug/L]
Temporal fit: Median model performance (S-HYPE)
200
4000
400
0
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
0
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
0
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Berit Arheimer
Models for predictions in ungauged basins methods for uncertainty reduction: Last decade
Atmospheric
deposition
Fertilizers,
Manure,
Plant residues
Plant
uptake
Denitrification
Rainfall,
Snowmelt
S1
Surface
runoff
S2
N&P pools
Macropore
flow
Groundwater
outflow, conc. of
IN, ON, SP & PP
N&P pools
Groundwater
N&P pools
Stream depth
S3
Tile drain
Regional groundwater flow
 Multi-basin approach (large domain for internal PUB)
 Linkage of model parameters to physical conditions (HRU)
 Step-wise interactive calibration
 New criterions for evaluation: spatial NSE (R2), median of NSE (R2)
 Assimilation of observed runoff and internal state variables, e.g.
snow and water level
Evapotranspiration
Lindström et al., 2009 (in press), Hydrology Research
Berit Arheimer
HYPE model performance (when calibrated)
120
S (cm)
6
0
E (mm/d)
80
4
-1
40
2
-2
0
0
-3
2000
2001
2002
2001
138
F (cm)
0
137
-20
136
-40
135
-60
134
1986
-8
1987
1988
2003
W (m)
1972
12
O18 (‰)
2002
1986
200
160
120
80
40
0
1973
1974
TN (mg/l)
1200
8
800
-12
4
400
-14
0
0
1995
1996
1990
1991
1992
1987
1988
1977
1978
1998
1999
Q (m3/s)
1976
-10
1994
G (m)
TP (µg/l)
1997
Berit Arheimer
Models for predictions in ungauged basins - methods
for uncertainty reduction: Future challenges
 New methods for data assimilation
(state variables and output):
• parameter tuning?
• up-dating?
• Kalman filtering?
 Methods for validation of state-variables (point to grid? Isotops?)
 New evaluation criteria (NSE, R2, MSE- no good!) + spatial pattern
 Validating pollution sources and sinks, i.e. nutrient isotops
 Campaign monitoring for validation of PUB
 What are the limits for PUB using multi-basin modelling?
The Balt-HYPE model
All models are wrong – but some may be useful!
Berit Arheimer
Balt-HYPE: Baltic basin –
HYdrological Predictions for the Environment
WHAT?
High resolution (250 km2), daily model of
water variables (e.g. flow rates, soil moisture)
and water quality
(N, P, TOC) over the entire basin
WHY?
•Homogenous model (impartial platform),
•Systematically implemented (easily run for
new scenarios),
• Linked to oceanographic model (RCO-scobi)
• Ensemble member (compared with local and basin scale models, or as a
harmonised reference model)
Berit Arheimer
Input data for the Baltic region: Readily Available
Global Databases

Topography: HydroSHEDS

Land use + soil: ECOCLIMAP
Lakes
Agriculture
 Forcing data (P & T): A combination of ERAInterim (ECMWF) and ERAMESAN data from
hindcasting
Conifers
 Major Dams: ICOLD
 Agricultural Data: Eurostat (as used in
CAPRIS model inputs)
 Point Sources: Population data from HYDE
database, treatment level and standard
values for emissions
 Atmospheric Deposition: Long term averages
taken from an atmospheric chemistry model,
the MATCH model (SMHI)
Peat soils
Fine soils
Berit Arheimer
Data for model calibration
and evaluation
Baltex stations for measuring daily Q
Catchment area (km2)
no info
0 - 10000
10001 - 50000
50001 - 100000
100001 - 281000
Finland
Norway
Sweden
Russia
 Observed river discharge: GRDC,
BALTEX (daily and monthly)
 Observed yearly river discharge: EEA
(yearly volumes)
 Observed nutrients: EEA, seasonal and
yearly totals and averages
 Possibility for additional data through
regional and local collaboration and
partnership
Estonia
Estonia
Latvia
Denmark
DenmarkDenmark
Lithuania
Belarus
Poland
Germany
Czech Republic
France
Ukraine
Slovakia
-
Berit Arheimer
Preliminary model results
Temperature
Precepitation
Acc. Discharge
Water balance
Berit Arheimer
Preliminary model results
Nitrogen
Phosphorus
Berit Arheimer
On-going: Evaluation, correction, calibration
Berit Arheimer
Next: Climate model data on relevant scale
GCM data
Global
RCM
Regional
application
Pungwe
Nhazonia
Frontiera
Vunduzi
Pungwe Falls
Katiyo
Pungue Sul
Honde Mavonde
Tacuraminga
Bue Maria
Hydrological
modelling
Hydrological
Future
runoff,variables,
water resources
and status
Matching
Information
Levels
(Streamlining)
Berit Arheimer
Next: Modelling the effect of measure
programmes and climate change
BSAP - Do we need new targets?
Transport [tonnes P yr-1]
An example from Rönneå River, Southern Sweden
60
Load with present land use&emissions in a future climate
50
40
30
20
Load after implemented measure-plan in a changed climate
10
0
1960 1980 2000 2020 2040 2060 2080 2100
Berit Arheimer
Possibilities
 Improvements using local high
resolution data-sources.
 Comparing results to other
models – ensemble modelling.
 Free result distribution to various
stake-holders on different levels,
using the SMHI production system
and web tools.
 A common platform (open source
code) for evaluation of ideas
regarding nutrient reducing
measures and climate change
impact.
Conclusions
Berit Arheimer
Balt-HYPE is:
 An homogenous, high resolution, open
source model of the entire Baltic Sea
catchment area, run operationally at
SMHI, with daily runoff and nutrients
among possible outputs.
 Intended to supplement local and other
regional modelling approaches.
 Can be used to examine the effects
(and evaluate concepts) of climate
change on river runoff and nutrient
inflows to the Baltic Sea.
 Can be used to examine the effects
(and evaluate concepts) of both local
and large scale remedial measures on
nutrient inflows to the Baltic Sea.
Thank you!
Berit Arheimer
HOME Water
a management tool
for water quality planning
 Status of coastal
zones
 Status of inland
waterbodies
 Source
apportionment of
pollutants
Berit Arheimer
HOME Water
Planning measures
to reduce eutrophication
 Treatment
plants
 Industries
 Agriculture
 Constructed
wetlands
 Buffer strips
 Rural
households
 Forest clear-cut
 Urban drainage
Berit Arheimer
HOME Water
 Compaire results
from different
simulations
 Economical analysis
– cost effectivity and
total cost
Berit Arheimer
Not just a model: A production system
Meteorological data
Climate projections
Policy scenarios
Local data
H
Y
P
E
Continental & river specific
Maps, Time series
and Statistics
www.smhi.se
Global & free databases
SMHI
Operational
Production
WHIST
GMES satellite products
• Present conditions
• Forecasts
• Climate change impact
• Measure effects / scenarios
Source apportionment
Model evaluation
User interface