Use of upstream satellite-derived flow signals for river discharge estimation:
application to major rivers in south Asia
To be submitted to: Remote Sensing of the Environment
1
Abstract
In this work we demonstrate the utility of satellite-based flow signals for river discharge
nowcasting and forecasting for two major rivers, Ganges and Brahmaputra, in south Asia.
Satellite-based daily flood signals estimated at more than twenty locations upstream of Hardinge
Bridge (for Ganges) and Bahadurabad (for Brahmaputra) gauging stations were used to: 1)
examine capability of remotely sensed flow signals to track the downstream propagation of flood
waves and 2) produce river flow nowcasts and forecasts at 1-15 days lead time using the flow
signals. We estimated the flow time from the correlation pattern of the data and applied a crossvalidation regression model to select a number of flow signals that produces a more accurate
river discharge. The results show that the well-correlated satellite-derived flow (SDF) signals
were able to capture a propagation of flood wave along both river channels. The daily river
discharge nowcast produced from the upstream SDFs has Nash-Sutcliffe coefficient of 0.8 for
both rivers; and the 15 day forecasts have Nash-Sutcliffe coefficient of 0.53 and 0.56 for Ganges
and Brahmaputra respectively. Overall, we conclude that satellite-based water level estimates are
a good source of surface water information in data scarce regions and they could be used for data
assimilation and model calibration purposes in near-time hydrologic forecast applications.
2
1. Introduction
River flow measurements are critical for hydrological data assimilation and model calibration
in flood forecasting and other water resource management issues. In most parts of the world,
however, in situ river discharge measurements are either totally unavailable or difficult to access
for timely use in operational flood forecasting and disaster mitigation. Over such areas, where insitu river discharge observations are missing, flood signals derived from microwave remote
sensors (e.g. Smith, 1997; Brakenridge et al., 1998; Brakenridge et al., 2005; 2007; Bjerklie et
al., 2005; Temimi et al., 2005; Smith and Pavelsky, 2008; De Groeve, 2010 and Birkinshaw et
al., 2010) could be used as alternative source of surface water information.
Brakenridge et al. (2007) demonstrated, through testing over different climatic regions of the
world, including rivers in the Unites States, Europe, Asia and Africa, that satellite passive
microwave data can be used to estimate river discharge changes, river ice status, and watershed
runoff. The method used the difference in brightness temperature between water and land
surfaces to estimate the area of land covered by water over a long period of time and then
translate the surface water change over time into river discharge signals. The brightness
temperature (related to the physical temperature through emissivity) was retrieved from
microwave sensors of Advanced Microwave Scanning Radiometer–Earth Observing System
(AMSR-E onboard on NASA’s Aqua satellite). Using the same data from AMSR-E, De Groeve
et al. (2006) developed a method to detect major global floods on a near-real time basis. More
recently, De Groeve (2010) showed in Namibia, southern Africa that the passive microwave
based flood signal was well correlated with the observed hydrograph data. It was also noted in
the study that the signal to noise ratio was highly affected by the local conditions on the ground
3
and, therefore, the quality of the signal was dependent on the number of pixels used in producing
the flood signal.
Upper-catchment satellite based flow monitoring may provide significant improvements to
flood forecast accuracy, primarily in the developing region where there is a limited availability
of ground based river discharge measurements. Bangladesh is one such case where river flooding
has historically been a very significant problem. Major flooding occurs in Bangladesh with a
return period of 4-5 years (Hopson and Webster, 2010) caused by the Ganges and Brahmaputra
Rivers, which enter into the country from India, and join in the Bangladeshi low lands. By virtue
of limited consistent river discharge data sharing between these two countries, the only reliable
river streamflow data come from what Bangladesh measures once the rivers cross the India–
Bangladesh border, traditionally limiting forecast lead-times to 2 to 3 days in the interior of the
country.
However, there is substantial utility in accurate and timely river flow forecast in Bangladesh;
for instance, according to estimates (CEGIS, 2006; Hopson and Webster, 2010) an accurate 7
day forecast has the potential of reducing post-flood costs by as much as 20% over a cost
reduction of 3% achieved with just a two-day forecast. Beginning in 2003, Hopson and Webster
(2010) developed and successfully implemented a real-time probabilistic forecast system of
severe flooding for both Ganges and Brahmaputra Rivers in Bangladesh, which triggered early
evacuation of people and livestock during the 2007 severe flooding of Brahmaputra. Although
the forecast system has shown useful skill out to 10-day lead-times by utilizing satellite-derived
TRMM (Huffman et al. 2005, 2007) and CMORPH (Joyce et al. 2004) precipitation estimates
and ensemble weather forecasts from the European Center for Medium Weather Forecasts,
Hopson and Webster (2010) also suggested the accuracy of the forecasts could significantly be
4
improved if the river flow measurements higher up in the catchments were available.
Satellite flood signals could be used for a purpose of updating soil moisture states in
hydrologic models, as well as for model calibration purposes in areas where the ground
discharge observations are not available. For Ganges and Brahmaputra rivers, Papa et al. (2010)
produced monthly discharges for the rivers from satellite altimeter. The monthly and seasonal
discharge estimates are very important for weather and climate applications but shorter time
scale information is also needed, such as daily or hourly, for operational short term flood
forecasting.
In current study we examine the utility of satellite based flood signal for near-real time river
flow estimation and forecasting for Ganges and Brahmaputra rivers in Bangladesh. The study has
two main parts. First, we investigated the capability of the satellite flow signal produced by the
Joint Research Center (JRC) of European Commission to track flood wave propagation along the
Ganges and Brahmaputra rivers. The second part of the study is using the satellite-derived flow
(SDF) signals for river flow simulation and forecasting in Bangladesh. The details of data used
are described in section 2. Section 3 presents the results of the flow signal analysis while the
variable selection method has been described in section 4 followed by results of discharge
nowcasting and forecasting in section 5. Finally in section 6 we summarize the conclusions
based on the results of the study.
2. Data sets
The Joint Research Center (JRC) of European commission in collaboration with Dartmouth
Flood Observatory (DFO) produces and provides daily near real-time flood signals, along with
small scale flood maps and animations, in more than 10 000 monitoring areas globally for major
rivers (GDACS, 2011). The detail methodology used to extract the daily signals from the passive
5
microwave remote sensing (AMSR-E and TRMM sensors) is described in De Groeve (2010) and
Brakenridge et al. (2007). In this study, we used the daily SDF signals along the Ganges and
Brahmaputra river cannels provided by the Global Flood Detection System (GFDS) of JRC. The
flood signals are available staring from December 8, 1997. A total of 22 data sets from locations
ranging from upstream distance of 63 to 1828 KM have been analyzed in case of Ganges river;
and similarly 23 data sets with a distance range of 53 to 2443 KM have been used for
Brahmaputra river. The details of the data sets used including the site ID, Lat/Lon of the sites
and the flow path length (FPL) have been presented in Table 1. We also used daily rating curvederived gauge discharge observations (from December 8, 1997 to December 31, 2010) of the
Ganges River at Hardinge Bridge and Brahmaputra River at Bahadurabad gauging stations (see
fig. 1) for model training and validation purposes.
3. Satellite-derived flow signals
3.1. Correlation with gauge observation
Figure 2.a and 2.b show correlations between three satellite-derived flow (SDF) estimates
and gauge discharge observations at Hardinge Bridge (Ganges) and Bahadurabad (Brahmaputra)
vs. lag time, respectively, with the correlation maxima shown by solid circles on the figures. The
in-channel distances between the locations where the upstream SDF were measured and the
outlet of the watershed have also been indicated in the figures. The correlations variation with
lag time has different characteristics depending on the distances (also known as flow path length
(FPL) - a hydrologic distance between the SDF detection and the outlet). Specifically, for shorter
FPL the correlation decreases monotonically with increasing lag time; however, for longer FPL
the correlation initially increases, reaches a maximum value, and then decreases with increasing
lag time. This change of correlation pattern, such as shifting of the maximum with upstream
6
distance, is an indicator of the utility if the SDF for capturing the flood wave propagation in the
river channel (see next section for detail).
3.2. Variation of flow time with flow path length
We estimated the travel time from the correlation pattern of the SDFs by assuming that the
lag time at which the maximum correlation occurred is a proxy measure of the flood wave
propagation time. The estimated flow time for each SDFs has been shown on Fig. 3.a. & 3.b. for
Ganges and Brahmaputra rivers respectively. In these figures the flow time estimated from the
flood signals were plotted against the flow path length, where the flow path length is the
hydrologic distance between the flood signal detection sites to the outlet of the watershed. We
estimated the flow path length from digital elevation map.
If the flow speed is assumed constant, then the flow time should increase with flow path
length but this is not strictly the case for both rivers in this study (see Fig’s 3.a.and 3.b.); a rather
inconsistent increase of flow time with distance has been found. The flow time is less than or
equal to 1 day for flow distances shorter than 750 KM
and 1000 KM for Ganges and
Brahmaputra respectively. We were not able to show for a time step less that 1 day due to the
time resolution (daily) of both the SFD signal and ground discharge measurement. The flow
time, calculated for available data points, varies going up to more than 10 days for Ganges and 7
days for Brahmaputra beyond the above mentioned flow path lengths. One of the factors
contributing to the inconsistent increase of the flow time with flow length could be the noise
introduced by the local ground conditions at locations where satellite observations were made, as
suggested by De Groeve (2010) that the local ground conditions affect the correlation.
There was a significant difference found when the ratios of the flow path to flow time
(calculated from the time at which maximum correlation) across the Ganges and Brahmaputra
7
rivers were compared especially for longer flow distances (see Fig’s 3.a.and 3.b.). For example,
the estimate shows that it takes 11 days to travel 1828 KM distance (the furthest upstream point,
11691) in case of Ganges, while for Brahmaputra the flow time is only 2 days for the comparable
path length of 1907 KM (site 11687). We fitted a liner line (as shown in the fig’s) by a weighting
the residuals according to their respective correlations, and also constraining the line in such a
way that it passes through the origin (zero distance and zero flow time). It was found that the
flow speed estimated from the slopes of the fitted line for Brahmaputra (9.85 m/s) is three times
more compared to that of Ganges (2.85 m/s). This difference could be due to many factors,
including the topography, special scale and temporal scale of precipitation, among others. The
Ganges river basin has a flat topography compared to the Brahmaputra which could contribute to
the higher residence time for water before it reaches the outlet. To investigate the influence, if
any, of the precipitation scale on the flow time, we conducted a simple synthetic experiment
where we varied both the distribution and location of rainfall over a “hypothetical watershed”
and route the excess rainfall down to the outlet using a liner reservoir unit hydrograph (Chow, et
al. 1998). In the synthetic experiment (not shown), we found that the spatial scale of precipitation
has an effect on the time at which maximum correlation occurred. However, to systematically
describe the influence of the precipitation scale on flood wave propagation time, we suggest that
a more realistic experiment with observed precipitation data over the river basin is necessary to
come to a tangible conclusion.
4. Selection of satellite flow signals for discharge estimation
As presented in the previous sections, the microwave based flow signals are well correlated
to the ground discharge measurement and they also capture the propagation of flood wave going
downstream as shown above for Ganges and Brahmaputra rivers. We used the microwave flow
8
signal available upstream of the Hardinge Bridge (Ganges) and Bahadurabad (Brahmaputra) to
produce daily discharge nowcast and forecasts at 1-15 day lead times at the gauging stations.
We applied cross-validation regression model in which the SDF signals were used as a
regression variable and the ground discharge observation at the outlet was used for training and
validation purposes. The nowcasting/forecasting steps for each lead time are as follows:
i. Calculate the correlation map. The correlation map is helpful for understanding the
linear relationship between the SDF signals and the ground discharge observation.
The variability of the correlation with lag time (as described in section 3) could also
be used to trace the flood wave propagation. Another useful aspect of the correlation
is that it could be used as a very handy indicator of the most relevant variables to be
used in the discharge estimation model. The correlation map for normalized data
(transformed to standard normal by subtracting the mean and then dividing by
standard deviation) is shown in Fig 4.a and 4.b for Ganges and Brahmaputra River
respectively. All data sets have different correlations depending on the location, flow
path length and lag time indicating that the local ground condition, besides the place
and time of observation, should be taken in to consideration before using the SDF for
any application. All data sets do not have strong linear relationship with the ground
observation and hence this step is useful for identifying the variables more related to
the river flow measurement for the discharge estimation model to be used in the next
steps.
ii. Sort the correlation in decreasing order. Variables which are more correlated with
ground discharge measurements will be used in the forecast model, thus to simplify
9
the selection process, we sorted the correlations calculated (see fig 4) in step i before
performing the selection task.
iii. Pick the variables to be used in the discharge estimation model and generate the river
discharge. We used a cross-validation approach to select variables, among the SDF
signals at multiple sites, to be used in the model. Identifying the most relevant
regression variables is very critical in order to prevent over fitting and consequently
reduce the error in the estimated discharge. We selected the most correlated flood
signals to the ground discharge observation as “the most relevant variables” to be
used in the model. To determine the optimal number, we applied a ten-percent leaveout cross-validation model, where 10% of the data is left out (to be used for
validation) at a time and a linear regression is fit to the remaining 90%. This is done
repeatedly until each data point is left out, but no data point is used more than once
for the validation purpose. This is followed by calculating the root mean square error
(RMSE) of the validation sets. Finally, the number of variables that produced the
smallest RMSE calculated over the whole out-of-sample data sets is considered as the
optimal number to be used in the regression model.
iv. Repeat the steps ii-iii for all lead times. We generated the river discharge nowcast and
forecast for each lead time (1 to 15 days) by repeating the regression variable
identification and discharge generation steps.
5. Results of discharge nowcast and forecast
5.1. Discharge nowcast
We used the cross-validation approach presented above to generate discharge nowcast from
the SDF signals detected at multiple points (see Table 1) upstream of the Hardinge Bridge
(Ganges) and Bahadurabad (Brahmaputra). The rating curve-derived gauge discharge
10
observations at the outlets were used for model training purpose. The time period of the data sets
ranges from December 8, 1997 to December 31, 2010. Figure 5.a. shows the time series plots of
the discharge simulation overlaid the observation for Ganges River at the Hardinge Bridge
during the monsoon flood of 2003. The discharge estimated from SDF correctly captured the
peak flow of September 20, 2003, and also matched (with little fluctuations) the falling limb of
the discharge for the summer period. However, it underestimated the flow during the early stage
of the summer. The Nash-Sutcliffe (NS) efficiency coefficient (see eq. 1) for the time series
(December 8, 1997 to December 31, 2010) is 0.80.
N
NS  1 
 (Q
i 1
N
oi
 (Q
i 1
 Qmi ) 2

oi
(1)
 Qo )
2
The 2007 Brahmaputra flooding (see Fig. 5.b.) is a different case: the discharge estimated from
the SDF signal did not fully capture the peak flows of the 2007 flood, it agreed with the
observation on the falling limb. The overall NS efficiency coefficient for the whole time (from
December 8, 1997 to December 31, 2010) series is 0.78.
5.2. Discharge forecast with satellite flood signal only
We applied, again, the cross-validation approach described in section 4 to forecast the
discharge at lead times from 1 to 15 days using the upstream satellite flood signal in the
regression model. Figure 6.a. shows time series of 1, 5 and 15 day forecasts overlaid the
observation for Ganges monsoon flood of 2003 at Hardinge Bridge. Past and current satellite
flood signals upstream of the forecast point at distances ranging from 63 to 1828 KM were used
as input to the forecasting model, while the discharge observation at the outlet was used for the
model training purposes. The 1 and 5 day lead forecast captured the peak flood of the September
20, 2003 correctly and they are not significantly far off the observation during the entire
11
monsoon season. The 15 day lead forecast, however, missed the peak flood of the September, 20,
2003 by almost 50%. Similar to the nowcast, the peak floods of the Brahmaputra 2007 monsoon
season (specifically July, 7 and September, 13), as shown in Fig. 6. b., were not captured by all
forecasts especially the rising limbs, but the falling limbs of the hydrographs were picked up
very well.
Let us look at the forecast of the entire time series instead of just one year example. Figure 7
presents the NS efficiency coefficient vs. lead time calculated for whole time period ranging
from December 8, 1997 to December 31, 2010. The NS efficiency score of the 1 day lead time
discharge forecast was 0.80 and declined to 0.52 for 15 day forecast in case of Ganges; similarly
for Brahmaputra it decreased from 0.80 for 1 day forecast to 0.56 for 15 day forecast. To account
for seasonal variability of the river flow, we performed the cross-validation based regression
separately for the dry (November-May) and wet (June-October) seasons, but no better forecast
skill due to seasonal classification was achieved. Overall, the results clearly indicate that the
remotely sensed flow signals provide useful information regarding surface water and could well
be used in these large rivers for flood forecasting with good skill, if not perfect.
5.3. Discharge forecast with combined SDF signal and persistence
Here, besides the satellite-based flood signals, we incorporated the current ground discharge
data (persistence) observed at the forecast point into the cross-validation forecast model to
examine how much the SDF improved the forecast skill with respect to the persistence. This
obviously relies on availability of current discharge observation at the forecast point, but if the
discharge is available then the combined use of the observed discharge with the satellite flood
signals is expected to improve the forecast skill. In fact, the NS efficiency coefficient improved
by 25%, 34% and 43% for 1, 5 and 15 day lead time respectively when persistence and satellite
12
flood signals were merged as opposed to SDF only. The contribution of the satellite flood signal
in the improvement of the forecast skill can be shown by comparing against the persistence. Fig.
8 shows the RMSE skill score of the 1 to 15 day lead forecast with reference to persistence for
both Ganges and Brahmaputra rivers. The microwave derived flood signals improved the
forecast RMSE skill score from 10% to 20% across the 15 day lead time.
6. Conclusion
This study showed that flood signals derived from passive microwave sensors are very useful
for near-real time river discharge forecasting of Ganges and Brahmaputra Rivers in Bangladesh.
The flood signals were very well correlated, albeit with different pattern, to the ground flow
measurements and are capable of tracking flood wave propagation going downstream the rivers.
The correlation pattern depends on the location, flow path length and lead time indicating that
the local ground conditions, topography, precipitation scale, and hydrologic response of the
watershed should be taken in to consideration before using the satellite signal for river flow
application; however the influence of each of the factors needs to be confirmed on separate
research.
The satellite based flood signals were used in cross-validation regression models for river
flow nowcasting and forecasting at 1-15 day lead times. The skill of the forecasts has improved
at all lead times compared to the persistence for both Ganges and Brahmaputra Rivers. This work
presents a substantial case for a proof of the utility of passive microwave remote sensing for
hydrologic forecast applications in data scarce regions. The remote sensing of river discharge
could play a significant role, as a main source of flow information, in ungauged basins and could
well be coupled with hydrologic models in data assimilation and model calibration framework
for flood forecasting purposes.
13
7. References
Birkinshaw, S.J, G.M. O’Donnell, P. Moore, C.G. Kilsby, H.J. Fowler and P.A.M. Berry, 2010:
Using satellite altimetry data to augment flow estimation techniques on the Mekong
River. Hydrol. Proc. DOI: 10.1002/hyp.7811.
Bjerklie, D. M., D. Moller, L. C. Smith, and S. L. Dingman, 2005: Estimating discharge in rivers
using remotely sensed hydraulic information, J. Hydrol., 309, 191– 209
Brakenridge, G. R., B. T. Tracy, and J. C. Knox, 1998: Orbital SAR remote sensing of a river
flood wave, Int. J. Remote Sens., 19(7), 1439– 1445
Brakenridge, G. R., S. V. Nghiem, E. Anderson, and R. Mic, 2007: Orbital microwave
measurement of river discharge and ice status, Water Resour. Res., 43, W04405,
doi:10.1029/2006WR005238.
Brakenridge, G. R., S. V. Nghiem, E. Anderson, and S. Chien, 2005: Space-based measurement
of river runoff, Eos Trans. AGU, 86(19), 185–188.
CEGIS, 2006: Sustainable end-to-end climate/flood forecast application through pilot projects
showing measurable improvements. CEGIS Base Line Rep., 78 pp.
Chow, V. T., D.R. Maidment and L.W. Mays, 1988: Applied Hydrology. McGraw-Hill, New
York
De Groeve, T., 2010: Flood monitoring and mapping using passive microwave remote sensing in
Namibia', Geomatics, Natural Hazards and Risk, 1(1), 19-35.
De Groeve, T., Z. Kugler, and G.R. Brakenridge, 2006: Near real time flood alerting for the
global disaster alert and coordination system. In Proceedings ISCRAM2007, B. Van De
Walle, P. Burghardt and C. Nieuwenhuis (Eds), pp. 33–39 (Newark, NJ: ISCRAM).
GDACS, Global Disaster Alert and Coordination System, Global Flood Detection System.
http://www.gdacs.org/floodmerge/. Accesed, January 2011.
Hopson, T.M, and P.J., Webster, 2010: A 1–10-Day Ensemble Forecasting Scheme for the Major
River Basins of Bangladesh: Forecasting Severe Floods of 2003–07, J. Hydromet., 11,
618-641. DOI: 10.1175/2009JHM1006.1.
Huffman, G. J., R. F. Adler, S. Curtis, D. T. Bolvin, and E. J. Nelkin, 2005: Global rainfall
analyses at monthly and 3-hr time scales. Measuring Precipitation from Space:
EURAINSAT and the Future, V. Levizzani, P. Bauer, and J. F. Turk, Eds., Springer, 722
pp.
——, and Coauthors, 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasiglobal, multiyear, combinedsensor precipitation estimates at fine scales. J. Hydrometeor.,
8, 38–55.
Joyce, R. J., J. E. Janowiak, P. A. Arkin, and P. Xie, 2004: CMORPH: A method that produces
global precipitation estimates from passive microwave and infrared data at high spatial
and temporal resolution. J. Hydrometeor., 5, 487–503.
Papa, F., F. Durand, W.B. Rossow, A. Rahman and S.K. Balla, 2010: Satellite altimeter-derived
monthly discharge of the Ganga-Brahmaputra River and its seasonal to interannual
variations from 1993 to 2008, J. Geophy. Res., 115, C12013, doi:10.1029/2009JC006075.
14
Smith, L. C., and T. M. Pavelsky, 2008: Estimation of river discharge, propagation speed, and
hydraulic geometry from space: Lena River, Siberia, Water Resour. Res., 44, W03427,
doi:10.1029/2007WR006133.
Smith, L.C., 1997: Satellite remote sensing of river inundation area, stage and discharge: A
review. Hvdrological Processes, 11, pp. 1427–1439.
Temimi, M., R. Leconte, F. Brissette, and N. Chaouch, 2005: Flood monitoring over the
Mackenzie River Basin using passive microwave data, Remote Sens. Environ., 98, 344–
355.
15
Ganges
Brahmaputra
Gauging location at Hardinge Bridge:
24.07N, 89.03E
GFDS Latitude Longitude
FPL
Site ID
(N)
(E)
(KM)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
11478
11488
11518
11522
11523
11524
11536
11537
11532
11528
11527
11539
11548
11559
11575
11588
11595
11606
11616
11623
11651
11691
24.209
24.469
25.341
25.402
25.415
25.409
25.660
25.722
25.672
25.585
25.513
25.620
25.938
26.149
26.423
26.852
27.179
27.494
27.738
28.003
28.812
29.259
88.699
88.290
87.030
86.670
86.379
85.950
85.069
84.587
84.150
83.700
83.430
81.519
81.207
80.815
80.439
80.123
79.786
79.470
79.110
78.674
78.131
78.035
63
121
340
370
420
550
650
676
690
725
800
1180
1220
1300
1320
1381
1431
1520
1590
1640
1761
1828
Gauging location at Bahadurabad:
25.09N, 89.67E
GFDS Latitude Longitude FPL
Site ID
(N)
(E)
(KM)
11533
11545
11558
11555
11554
11560
11570
11576
11579
11580
11583
11593
11603
11610
11619
11677
11681
11687
11685
11684
11675
11678
11679
25.451
25.875
26.014
26.221
26.148
26.205
26.383
26.574
26.671
26.776
26.853
27.089
27.394
27.603
27.836
29.296
29.300
29.369
29.295
29.334
29.303
29.232
29.267
89.707
89.910
90.282
90.738
91.214
91.683
92.119
92.586
93.074
93.555
94.062
94.456
94.748
95.040
95.293
91.305
90.854
89.441
88.966
88.443
88.049
85.230
84.709
53
117
145
204
285
330
385
475
496
590
630
660
712
750
837
1698
1737
1907
1929
1996
2045
2380
2443
Table 1. Details of the satellite-derived flow signal (“MagnitudeAvg” as labeled on GDACS
database) used for the study. The site ID, latitude, longitude and flow path length (FPL) have
been shown. The time period of all the data sets, including the satellite flood signal and gauge
discharge observations at Hardinge Bridge for Ganges and Bahadurabad for Brahmaputra, covers
from December 8, 1997 to December 31, 2010.
16
Fig. 1. Study Region, the Brahmaputra and Ganges rivers in south Asia. The satellite flood signal
observations have been overlaid on the main streams of the Brahmaputra (top right) and the
Ganges (bottom left) rivers. The observation sites have been shown in small dark triangles and
they are labeled by the GFDS site ID (see Table 1 for more details).
17
Fig 2.a. Correlation versus lag time between upstream satellite flood signals (only 3 shown here)
and gauge discharge at Hardinge Bridge of Ganges River in Bangladesh. The characteristics of
the correlation curve vary depending on the location of the satellite observation. The time at
which peak correlation occurs (shown as dark dot) is greater for longer flow path lengths (FPL).
18
Fig 2.b. Same as Fig. 2.a, but for Brahmaputra river.
19
Fig 3.a. shows plot of flow time (as estimated from the satellite flood signal data) versus distance
from the satellite flow detection point to the outlet (Hardinge bridge station) of the Ganges River.
The flow time is the lag time at which the peak correlation occurred, as shown in Fig.2.a. The
flow speed estimated from the slope of the fitted line is 2.85 m/s.
20
Fig 3.b. Same as Fig. 3.a., but for Brahmaputra river.
21
Fig 4.a. Lagged correlation map of daily satellite-derived flow signals calculated against the
discharge observation at Hardinge Bridge for Ganges River. The horizontal axis shows the
satellite flood signal sites (see Fig. 1) arranged in the order of increasing flow path length and the
vertical axis shows lag time (days).
22
Fig 4.b. Same as Fig. 4.a, but for Brahmaputra.
23
Fig 5.a. Daily time series of observed river discharge (solid) and nowcast (dash) based on the
river flow signal observed from satellite for Ganges River at Hardinge bridge station in
Bangladesh. Satellite-derived flow signals, at different locations with distance ranging from 63
KM to 1828 KM upstream the Hardinge bridge station as shown in Table 1, have been used for
the discharge estimation.
24
Fig. 5.b. Same as Fig. 5.a, but for Brahmaputra River.
25
Fig. 6.a. Daily time series of satellite-based river discharge forecast at different lead times shown
against observation during the 2003 flooding of Ganges River at Hardinge bridge station in
Bangladesh. Satellite-derived flow signals, at different locations with distance ranging from 63
KM to 1828 KM upstream the Hardinge bridge station as shown in Table 1, have been used for
the forecasting.
26
Fig. 6.b. Same as Fig. 6.a, but for Brahmaputra river.
27
Fig. 7. The Nash-Sutcliffe coefficient versus forecast lead time for Ganges and Brahmaputra
Rivers. Only satellite-derived flow signals were used for the forecast. The Nash-Sutcliffe
coefficients were calculated for the whole time period of record (December 8, 1997 to December
31, 2010).
28
Fig. 8. The root mean square error (RMSE) skill score versus forecast lead time for Ganges and
Brahmaputra Rivers. The current discharge observation (persistence) at the outlet point, along
with the upstream satellite flood signals, have been used for the forecasting, and therefore the
skill of the forecast has been estimated with respect to the RMSE of persistence. The skill scores
were calculated for the whole time period of record (December 8, 1997 to December 31, 2010).
29