Page 1
27 Jul 2011 IGARSS 2011 - session WE2.T10 1
Benefits of satellite altimetry for Benefits of satellite altimetry for transboundary basinstransboundary basins
S. Biancamaria 1,2, F. Hossain 3, D. P. Lettenmaier 4, N. Pourthié 2 and C. Lion 1,2
1 LEGOS, Toulouse, France2 CNES, Toulouse, France
3 CEE, Tennessee Tech University, Cookeville, TN, USA4 CEE, University of Washington, Seattle, WA, USA
Page 2
Transboundary basinsTransboundary basins
• 256 river basins are shared among 2 or more countries (Wolf et al., 1999) = 45% land surfaces
27 Jul 2011 IGARSS 2011 - session WE2.T10 2
Page 3
OutlineOutline
27 Jul 2011 IGARSS 2011 - session WE2.T10 3
1. Forecasting Brahmaputra/Ganges water elevations using satellite altimetry
2. Monitoring Indus reservoirs with SWOT
Page 4
Brahmaputra and Ganges basins Brahmaputra and Ganges basins • Brahmaputra: drainage area=574,000km2;
population=30 Millions; unmanaged.• Ganges: drainage area=1,065,000km2;
population=500 Millions; 34 dams/diversions.
27 Jul 2011 IGARSS 2011 - session WE2.T10 4
Bangladesh
Page 5
IssueIssue• 90% of water flowing in Bangladesh comes from
India.• No India/Bangladesh real time data sharing.• Using in-situ measurements at its border ->
forecast in Bangladesh only with 2 or 3 days lead time.
• Study purpose: Use satellite-based water elevation upstream in India to forecast water elevation at the gauge locations (India/Bangladesh border).
27 Jul 2011 IGARSS 2011 - session WE2.T10 5
Page 6
Data used: in-situ measurementsData used: in-situ measurements
27 Jul 2011 IGARSS 2011 - session WE2.T10 6
Page 7
Data used: satellite altimetryData used: satellite altimetry• Topex/Poseidon (T/P) satellite altimeter.• Overlap with in-situ: January 2000/August
2002.• Data downloaded from HYDROWEB:
http://www.legos.obs-mip.fr/en/soa/hydrologie/hydroweb/
27 Jul 2011 IGARSS 2011 - session WE2.T10 7
242_1166_1
014_1
116_2
T/P Virtual station
Distance from gage
Mean time between obs.
242_1 550 km 14 days
166_1 250 km 16 days
Mean time between obs.
Distance from gage
T/P Virtual station
12 days 1560 km 116_2
22 days 530 km 014_1
Page 8
Methodology 1/2Methodology 1/2
• Compute the cross-correlation between upstream T/P and in-situ measurements:
27 Jul 2011 IGARSS 2011 - session WE2.T10 8
Water level Correlation
Time Lead time00
0.6
k)(th(t)h
k)(th(t),h=n(k)Correlatio
altiinsitu
altiinsitu
stdevstdev
covwith k=lead time
k
0.8
k
Upstream:halti(t) Downstream;
hinsitu(t)
Page 9
Methodology 2/2Methodology 2/2
• Compute scatter plot in-situ measurements & T/P measurements k days earlier.
• Use linear fit to forecast water level at gauge location from T/P measurements.
27 Jul 2011 IGARSS 2011 - session WE2.T10 9
hin-situ(t)
0 halti(t-k)
Linear fit of hinsitu(t)=f[halti(t-k)]
Water level
0 Time
hinsitu (downstream)
k day lead time forecast
Page 10
Results on the BrahmaputraResults on the Brahmaputra
• 5-day lead time Forecasts:
27 Jul 2011 IGARSS 2011 - session WE2.T10 10
T/P virtual station 250 km upstream: T/P virtual station 550 km upstream:
5-day forecast RMSE ~ 0.5 m
5-day forecast RMSE ~ 0.5 m
Brahmaputra water elevation543210
-1-2-32000 2001 2002
Wat
er e
leva
tion
(m)
Brahmaputra water elevation543210
-1-2-32000 2001 2002
Wat
er e
leva
tion
(m)
In-situ
T/P forecast
Legend:
Page 11
Results on the GangesResults on the Ganges
27 Jul 2011 IGARSS 2011 - session WE2.T10 11
• 10-day lead time forecast:T/P virtual station 530 km upstream:
5-day forecast RMSE ~ 0.6 m
Ganges water elevation
6
4
2
0
-2
-42001 2001.4 2001.8
Wat
er e
leva
tion
(m)
10-day forecast RMSE ~ 0.9 m
Ganges water elevation
6
4
2
0
-2
-42001 2001.4 2001.8
Wat
er e
leva
tion
(m)
T/P virtual station 1560 km upstream:
• 5-day lead time forecast:
In-situ
T/P forecast
Legend:
Page 12
SWOT and the Brahmaputra/GangesSWOT and the Brahmaputra/Ganges
27 Jul 2011 IGARSS 2011 - session WE2.T10 12
• SWOT = Water mask + water elevation (and river slope) with 2 or more observations per 22 days
Page 13
Expected benefits from SWOTExpected benefits from SWOT• Higher precision on measurements -> better
forecasts.• More observations on the basin -> better time
sampling.• Water extent will improve inundation forecast:
27 Jul 2011 IGARSS 2011 - session WE2.T10 13
Brahmaputra water elevation543210
-1-2-32000 2001 2002
Wat
er e
leva
tion
(m)
Brahmaputra water elevation
Dis
char
ge (
104 m
3 .S
-1) 7
6
5
4
3
2
1
02000 2001 2002
Page 14
Conclusion for Brahmaputra/GangesConclusion for Brahmaputra/Ganges
• Forecasting water elevation from nadir altimeters with lead time between 5 day and 10 day.
• Expected improvement from SWOT due to water elev. + extent, better accuracy, global observation.
• Fore more details: Biancamaria et al., GRL, 38, L11401, “Forecasting transboundary river water elevations from space” (June 2011).
27 Jul 2011 IGARSS 2011 - session WE2.T10 14
Page 15
OutlineOutline
27 Jul 2011 IGARSS 2011 - session WE2.T10 15
1. Forecasting Brahmaputra/Ganges water elevations using satellite altimetry
2. Monitoring Indus reservoirs with SWOT
Page 16
SWOT and world lakes/reservoirsSWOT and world lakes/reservoirs
27 Jul 2011 IGARSS 2011 - session WE2.T10 16
2 4 6 8 10SWOT visits per repeat cycle
3
2.5
2
1.5
1
0.5
0Su
rfac
e ar
ea s
een
(10
6 km
2 )
Page 17
Indus reservoirsIndus reservoirs
27 Jul 2011 IGARSS 2011 - session WE2.T10 17
• Indus basin=1.14x106km2, 53% to Pakistan, 34% to India.
• 2008 filling of Baglihar reservoir by India.
• 2009 construction of Kishenganga dam.
-> Lack of information = difficulties in downstream water management
Page 18
SWOT and Indus reservoirsSWOT and Indus reservoirs
27 Jul 2011 IGARSS 2011 - session WE2.T10 18
• Reservoirs are seen between 2 to 3 times per 22 days
Page 19
Baglihar dam Baglihar dam
27 Jul 2011 IGARSS 2011 - session WE2.T10 19
• Baglihar dam: 450 MW Run-of-river type– Pondage volume = 37.5 x 106 m3
– Full pondage level – dead storage level = 5 m-> Pondage area > 1 km2
5 m
37.5x106 m3
• SWOT requirements on lakes and reservoirs = 10 cm error on 1 km2 area.
-> SWOT should be able to observe Baglihar dam
Page 20
Baglihar damBaglihar dam
27 Jul 2011 IGARSS 2011 - session WE2.T10 20
• Reservoir in mountainous region = SWOT might be affected by layover.
• Layover=geometric distortion when radar beam reaches top of a tall feature before it reaches the base.
• Layover modeled by SARVisor for ALOS/PalSAR, 7° incidence angle (yellow=layover):
BagliharBaglihar
Ascending track: Descending track:
Page 21
Kishenganga projectKishenganga project
• 330 MW hydro-power plant.• Layover modeled by SARVisor for ALOS/PalSAR,
7° incidence angle (yellow=layover):
27 Jul 2011 IGARSS 2011 - session WE2.T10 21
Kishenganga
Ascending track: Descending track:
Kishenganga
Page 22
Conclusion for Indus reservoirsConclusion for Indus reservoirs
27 Jul 2011 IGARSS 2011 - session WE2.T10 22
• Hydro-electric reservoirs are needed to respond to the growing demand on electricity.
• Water management for downstream country is more difficult.
• Huge potential of SWOT to provide reservoir water volume changes
• Ongoing study to characterize layover on SWOT data and better quantifying SWOT accuracy and time sampling.
Page 23
Thank you for your attentionThank you for your attention
27 Jul 2011 IGARSS 2011 - session WE2.T10 23