CURRENT SCIENCE, VOL. 97, NO. 3, 10 AUGUST 2009 429
to detecting two spatially separated clusters of P. juliflora,
the L-band cross-polarized SAR was found to be best
suited as two distinct peaks were observed for it. Whereas
the C-band cross-polarized SAR could not give two dis-
tinct peaks corresponding to two tree clusters. Instead the
two clusters appeared as more or less as a single cluster
on the C-band cross-polarized SAR image. The X band
could not detect the presence of thin vegetation volume
for any of the three cases taken up in this study, as the
thin vegetation failed to have any impact on the X band
SAR in comparison to its surrounding features. Thus the
overall L-band cross-polarized SAR was found to be the
most suitable for detecting thin vegetation volume in the
1. Hsu, C. C., Wang, L., Kong, J. A., Souyris, J. C. and Le Toan, T.,
Theoretical modeling for microwave remote sensing of forest. In
International Symposium on the Retrieval of Bio- and Geophysical
Parameters from SAR Data for Land Applications, Toulouse,
2. Kasischke, E. S. and Christensen, N. L., Connecting forest ecosys-
tem and microwave backscatter models. Int. J. Remote Sensing,
1990, 11, 1277–1298.
3. Souyris, J. C., Le Toan, T., Hsu, C. C. and Kong, J. A., Inversion
of land forest biomass using SIR-C/X-SAR data: Experiment and
theory. In International Symposium on the Retrieval of Bio- and
Geophysical Parameters from SAR Data for Land Applications,
Toulouse, France, 1995.
4. Srivastava, H. S., Patel, P., Sharma, Y. and Navalgund, R. R., De-
tection and density mapping of forested areas using SAR interfero-
metry technique. Inter. J. Geoinf., 2007, 3, 1–10.
5. Ulaby, F. T. and Dobson, M. C., Handbook of Radar Scattering
Statistics for Terrain, Artech House, Norwood, MA, 1989.
6. Henderson, F. M. and Lewis, J., Principles and Applications of
Imaging Radar: Manual of Remote Sensing, John Wiley & Sons,
Inc., NY, 1998, vol. 2, 3rd edn, pp. 461–465.
7. Dobson, M. C., Ulaby, F. T., LeToan, T., Beaudoin, A., Kasis-
chke, E. S. and Christensen, N., Dependence of radar backscatter
on coniferous forest biomass. IEEE Trans. Geosci. Remote Sens-
ing, 1992, 30, 412–415.
8. Dobson, M. C. et al., Estimation of forest biophysical characteris-
tics in northern Michigan with SIR-C/X-SAR. IEEE Trans. Geo-
sci. Remote Sensing, 1995, 33, 877–895.
9. Ulaby, F. T., Moore, R. K. and Fung, A. K., Microwave Remote
Sensing: Active and Passive, Artech House, Norwood, 1990, vols
II & III.
10. Patel, P., Srivastava, H. S., Panigrahy, S. and Parihar, J. S., Com-
parative evaluation of the sensitivity of multi-polarized multi-
frequency SAR backscatter to plant density. Int. J. Remote Sens-
ing, 1996, 27, 293–305.
12. Kierein-Young, K. S., Lefkoff, A. B. and Kruse, F. A., Radar
Analysis and Visualization Environment (RAVEN): Software for
Polarimetric Radar Analysis. In Third Annual JPL Airborne Geo-
science Workshop (AIRSAR Workshop), JPL Publication, 1992,
vol. 3, pp. 78–80.
13. Lee, J. S., Speckle suppression and analysis for SAR images. Opt.
Eng., 1986, 25, 636–643.
ACKNOWLEDGEMENTS. We thank Dr V. Jayaraman, Director,
National Remote Sensing Centre, Hyderabad for keen interest in this
study. H.S.S. and P.P. thank Dr Y. V. N. Krishnamurthy, Director,
Regional Remote Sensing Service Centre/National Natural Resources
Management System (RRSSC/NNMRS), ISRO Head Quarters, Banga-
lore; Dr J. S. Parihar, Deputy Director, Remote Sensing Applications
Area, Space Applications Centre, Ahmedabad; Dr K. P. Sharma, Head-
In-charge, RRSSC-Dehradun; Dr M. Chakraborty, Group Director,
Geo-Informatics and Techniques Development Group and Dr S.
Mohan, Head, Advance Techniques Development Division for support
Received 4 September 2008; revised accepted 10 June 2009
The Great avulsion of Kosi on
18 August 2008
Engineering Geosciences Group, Department of Civil Engineering,
Indian Institute of Technology, Kanpur 208 016, India
The 18 August 2008 avulsion of the Kosi River drain-
ing the parts of north Bihar in eastern India may well
be regarded as one of the greatest avulsions in a large
river in recent years. The Kosi River shifted by ~120 km
eastward, triggered by the breach of the eastern afflux
bund at Kusaha in Nepal at a location 12 km upstream
of the Kosi barrage. This event was widely perceived
as a major flood in the media and scientific circles.
Although a large area was indeed inundated after this
event, it is important to appreciate that this inunda-
tion was different from a regular flooding event.
Keywords: Floods, Ganga plains, Kosi barrage, river
dynamics, river management.
RIVERS play a critical role in human society and history
as they are the major source of fresh water, transporta-
tion, and resources. However, this relationship is often
‘troubled’ because changes in river discharge (floods or
droughts) or position can play havoc with permanent set-
tlements. Such changes can be caused by natural forcing
as well as human interventions, or a combination of both.
Natural processes may include short-term changes in
sediment load or water volume as well as long-term
changes in relative sea level or climate change. The human
interventions impact changes in sediment load or run-off
through water resource management schemes such as
dams, barrages and embankments. Human alterations of
river systems can have many important consequences,
primarily because river systems are dynamic and highly
integrated systems and, any change in any part of the
river can easily propagate and affect the whole system.
The Kosi River is an important tributary of the Ganga
in the eastern India (Figure 1 a) and has distinctive hydro-
CURRENT SCIENCE, VOL. 97, NO. 3, 10 AUGUST 2009
Figure 1. a, Location map showing the position of the Kosi River in eastern Ganga plains (G, Ganga; Y,
Yamuna; K, Kosi). b, The Kosi River has a large mountainous catchment in Nepal and a rather small alluvial area
in north Bihar giving rise to a large upland/alluvial (u/p) ratio; c, The historical records suggest a dominant west-
ward migration of the Kosi River during the last ~200 years before the river was embanked on both sides by
Table 1. Major hydrological and sediment transport characteristics of
the Kosi River. Note that the Kosi has a high sediment yield which has
to be accommodated in a rather small alluvial area (u/p ratio is 5.3,
see Figure 1 b)
Parameter Kosi Ganga Amazon
Catchment area (103 km2) 101 1073 7180
Total length (km) 1216 2700 6518
Average annual discharge (m3/s) 2036 15,000 480,000
Annual sediment load at 43 1670 1000
river mouth (mt/yr)
Discharge/area 20 14 25
Sediment yield (mt/y/km2) 0.43 1.56 0.14
logical and sediment transport characteristics (Table 1).
The dynamics of the Kosi River, generally described as
‘avulsive’ shifts, has been well documented by previous
workers1–3 and a preferentially westward movement of
150 km in the last 200 years has been recorded (Figure
1 c). Various explanations for this unidirectional shift of
the Kosi include sedimentation in a braided stream4, cone
building activity2, active tectonics5,6, and autocyclic and
stochastic movements3. Avulsion involves a sudden move-
ment around a nodal point (divergence point) and occurs
when an event of sufficient magnitude (usually a flood)
occurs along a river that is at or near avulsion threshold7
defined by the changing channel instability through time.
It also implies that avulsion may not always be triggered
by the largest flood in a given river, and that even a small
flood can trigger an avulsion if the river is close to avul-
sion threshold. One of the most common mechanisms of
avulsion is ‘channel reoccupation’ (rapid) when the new
channel occupies a pre-existing channel in the vicinity.
On the other hand, ‘crevasse splaying’ involves a gradual
process of breaching through the banks and development
of a new channel through time. Not just Kosi, several rivers
draining the plains of north Bihar are known for frequent
and rapid avulsions8–10 and the area is prone to fluvial
Unlike the previous westward avulsions (Figure 1 c),
the 18 August 2008 avulsion of the Kosi River recorded
an eastward jump of ~120 km which is an order of mag-
nitude higher than any single avulsive shift recorded in
historical times. The avulsion was triggered by a breach
in the eastern afflux bund of the Kosi at Kusaha, 12 km
upstream of the Kosi barrage (Figure 2 a and b). This
avulsed channel ‘reoccupied’ one of the palaeochannels
of the Kosi and 80–85% flow of the river was diverted
into the new course. Since the new course had a much
lower carrying capacity, the water flowed like a sheet,
15–20 km wide and 150 km long with a velocity of 1 m/s
at the time of breach. Interestingly, the new course did
not join back the Kosi nor did this find through-drainage
into the Ganga, as a result of which a very large area
remained inundated/waterlogged for more than four
months after the breach. This single event affected more
than 30 million people and there is still no reliable esti-
mate of loss of life and property.
Several lines of evidence coupled with ground observa-
tions support that this event was a ‘mega-avulsion’ rather
than a regular flood.
CURRENT SCIENCE, VOL. 97, NO. 3, 10 AUGUST 2009 431
Figure 2. a, Google image showing the course of the Kosi River before avulsion and the breach point at Kusaha; b, Part
of the eastern afflux washed away after the breach and the river avulsed towards the east; c, Seepage channel outside the
eastern afflux bund causing significant toe erosion; d, The western side of river bed and the afflux bund – river bed is 4–
5 m higher than the adjoining flood plain.
Simulations for understanding the avulsion mechanism
suggest that avulsion points shift up-valley followed by
an abrupt down-valley shift as a result of continued
growth of alluvial ridges and increase in cross-valley
slope upstream of avulsion locations13. Specific simula-
tions for the Kosi also predicted the pseudo-nodal style of
the progressive westward shift and it was also suggested
that channel belt would start shifting towards the east
once the avulsing channel belts encounter a barrier created
by the depositional topography of the fan13. Although the
observational data from the Kosi2,3 show a down-valley
shift in avulsion sites through time, the river has a
tendency to periodically return to the upstream nodal
It has also been suggested that a decrease in inter-
avulsion period occurs through time7,13,14 because the
probability of avulsion increases due to decrease in over-
all channel belt slope. Data from Kosi shows a decrease
in inter-avulsion period7,14 between 1700 and 1955. A
sharp increase in the inter-avulsion period around 1955
coincides with the construction of embankments along
both banks of the Kosi which should have stabilized the
The breach at Kusaha occurred at a discharge of
144,000 cusecs which is much less than the design dis-
charge of 950,000 cusecs for the barrage upstream and
the afflux bunds. The river avulsed following the breach,
occupied one of its palaeochannels and inundated large areas
as the channel capacity of the new course was very small.
Repeated satellite images show that the Kosi River
around Kusaha was flowing very close to the eastern
afflux bund at least since 1999 and there are reports that a
couple of spurs upstream of the breach point were eroded
in the last few years. During the field visit, it was also
observed that a well-defined seepage channel (Figure 2 c)
outside and parallel to the eastern afflux bund formed
some years ago. This channel has also been causing signi-
ficant toe erosion of the afflux bund. Further, the afflux
bunds are more than 50 years old and have been poorly
maintained which may have facilitated the breach and the
In contrast, the channel has been aggrading on the
western side with an accelerated rate after the construc-
tion of embankment. This is not surprising as the Kosi is
among the highest sediment-laden river in the world15
(0.43 mt/y/km2; see Table 1). The river bed around the
western afflux bund was observed to be at least 4–5 m
higher than the surrounding floodplain level (Figure 2 d)
although no measurement of the rate of channel bed
aggradation is available at this stage. This suggests that
most of the sediment load was trapped within the em-
bankment and the river developed a ‘gradient advantage’
as the cross valley slope exceeded the down-valley slope
in this region (Figure 3). This made the eastern afflux
bund vulnerable to breach and pushed the river close to
This avulsion triggered by the breach, has once again
questioned the efficacy of the embankment strategy for
CURRENT SCIENCE, VOL. 97, NO. 3, 10 AUGUST 2009
flood control. Even a casual look at the data and interac-
tions with local people in the Kosi region would reveal
that there has been no appreciable flood moderation in
the Kosi and other rivers of north Bihar even after the
construction of embankments and the barrage (designed
with a flood cushion). The embankment strategy for flood
control as well as several other human interventions such
as highways and railway embankments have also pro-
duced severe drainage congestion in the region which
results in longer inundation of large areas almost every
year. There is an urgent need to adopt an ‘integrated’
river basin management which requires a rigorous under-
standing of the physical processes by which river channels
are formed and maintained, and encompasses all physical
attributes of the earth’s surface involved in water cycle.
The 18 August avulsion of the Kosi triggered by the
breach at Kusaha occurred partly due to incorrect strate-
gies of river management and partly due to human negli-
gence and poor maintenance of the afflux bund for the
last several years. This means that this avulsion was partly
a ‘human disaster’ rather than regular flooding.
The Kosi River was diverted back into the old course
through the barrage on 26 January 2009 after restoring
about 2000 m long embankment which was breached on
18 August 2008. The big question now is: how long the
river will stay in this course and when will the next
breach/avulsion occur? The conditions which led to the
breach and avulsion of the river, i.e. aggraded river bed
within the poorly maintained embankment, remain as
they were before the avulsion. The sustenance of the
plugging of breach is questionable and the possibility of
another breach in the near future at other locations cannot
be ruled out. A good possibility should have been to use
the new course as a diversion channel for the excess water
Figure 3. Schematic model for the 18 August 2008 avulsion of the
Kosi River. a, Natural course of the river before embankment construc-
tion; large floods passed through the channel and avulsion threshold
was higher. b, River embanked on both sides resulting in accelerated
aggradation; frequent breaches through small floods, reduction in
stream power and lowering of avulsion threshold. c, Crossing of avul-
sion threshold triggered by ‘gradient advantage’ (increase in Sa/Se)
caused by aggradation.
during floods and to follow the age-old practice of ‘con-
trolled’ flooding. Perhaps a few other palaeochannels of
the Kosi can be surveyed and a system of channel net-
work can be developed as a long-term effort. The course
of the river through the barrage and within the embank-
ment would need significant channel improvement per-
haps through dredging in selected reaches. Till today, no
attempts have been made to improve the channel or
strengthen the embankment.
While a great deal of research needs to be done to find
a long-term solution to the Kosi avulsion and flooding, it
is important that a ‘system’ approach to river manage-
ment should be adopted keeping in view the dynamic
behaviour of the Kosi. Further, there has been a paradigm
shift globally from ‘river control’ primarily involving an
engineering approach addressing the ‘effect’ at a local
scale to ‘river management’ which emphasizes an inte-
grated approach at a crossover of scales and addresses the
‘cause’ rather than the effect16–19. Even though India is a
country bestowed with several large rivers, our river
management strategies are rather rudimentary and our
planners are yet to embrace modern approaches such as
satellite-based monitoring and multi-criteria decision
support system. Some efforts in this direction20,21 have
already shown encouraging results but a large-scale,
coordinated effort is needed to save a large population
from repeated miseries of fluvial hazards year after year.
A process-based understanding of the Kosi and the cou-
pling between river form and processes are needed to find
long-term solutions to river dynamics and floods.
1. Shillingfield, 1893; cited in Gole and Chitale, 1966 (see ref. 2).
2. Gole, C. V. and Chitale, S. V., Inland delta building activity of
Kosi River. J. Hydraul. Div., ASCE, 1966, 92, 111–126.
3. Wells, N. A. and Dorr, J. A., Shifting of the Kosi River, northern
India. Geology, 1987, 15, 204–207.
4. Leopold, L. B. and Maddock, T., Flood control problems. J. Soil.
Water Conserv. India, 1955, 3, 169–173.
5. Arogyaswamy, R. N. P., Some geological factors influencing the
behaviour of the Kosi. Rec. Geol. Surv. India, 1971, 96, 42–52.
6. Agarwal, R. P. and Bhoj, R., Evolution of Kosi river fan, India:
structural implication and geomorphic significance. Int. J. Remote
Sensing, 1992, 13, 1891–1901.
7. Jones and Schumm, S. A., Causes of avulsion: an overview. In
Fluvial Sedimentology VI (eds Smith, N. D. and Rogers), Spl. Pub.
Int. Ass. Sedimentol., 1999, vol. 28, pp. 171–178.
8. Sinha, R., Channel avulsion and floodplain structure in the Gan-
dak–Kosi interfan, north Bihar plains, India. Z. Geomorph.,
Suppl.-Bd, 1996, 03, 249–268.
9. Jain, V. and Sinha, R., Fluvial dynamics of an anabranching river
system in Himalayan foreland basin, Baghmati River, north Bihar
plains, India. Geomorphology, 2004, 60, 147–170.
10. Jain, V. and Sinha, R., Hyperavulsive-anabranching Baghmati
river system, north Bihar Plains, eastern India. Z. Geomorph.,
2003, 47, 101–116.
11. Sinha, R. and Jain, V., Flood hazards of North Bihar Rivers, Indo-
Gangetic Plains. Mem. Geol. Soc. India, 1998, 41, 27–52.
12. Sinha, R., On the controls of fluvial hazards in the north Bihar
plains, eastern India. Engineering Geology Spl. Publication 15,
Geological Society of London, 1998, pp. 35–40.
CURRENT SCIENCE, VOL. 97, NO. 3, 10 AUGUST 2009 433
13. Mackey, S. D. and Bridge, J. S., Three dimensional model of allu-
vial stratigraphy: theory and application. J. Sed. Res., 1995, 365,
14. Stouthamer, E. and Berendsen, H. J. A., Avulsion: the relative
roles of autogenic and allogenic processes. Sed. Geol., 2007, 198,
15. Sinha, R. and Friend, P. F., River systems and their sediment flux,
Indo-Gangetic plains, northern Bihar, India. Sedimentology, 1994,
16. Gilvear, D. J., Fluvial geomorphology and river engineering:
future role utilizing a fluvial hydrosystems framewok. Geomor-
phology, 1999, 31, 229–245.
17. Downs, P. W. and Gregory, K. J., River Channel Management,
Arnold Press, 2004, p. 395.
18. Brierley, G. J. and Fryirs, K., Geomorphology and River Man-
agement: Applications of the River Styles Framework, Blackwell
Publishing, Oxford, 2005, p. 398.
19. Brierley, G. J. and Fryirs, K., River Futures. An Integrative Scien-
tific Approach to River Repair, Island Press, Washington, 2005, p.
20. Bapalu, G. V. and Sinha, R., In Flood Hazard Mapping: A Case
Study of Kosi River Basin, GIS@Development online edition,
21. Sinha, R., Bapalu, G. V., Singh, L. K. and Rath, B., Flood risk
analysis in the Kosi river basin, north Bihar using multi-
parametric approach of Analytical Hierarchy Process (AHP). J.
Indian Soc. Remote Sens., 2008, 36, 293–307.
ACKNOWLEDGEMENTS. I thank Barh Jan Ayog Bihar for organiz-
ing a visit to the flood affected areas in Nepal and north Bihar includ-
ing the breach point at Kusaha. Discussions with several persons
namely, Dr Manas Bihari Verma, Dr Ajaya Dixit, Dr Deepak Gywali,
Mr Vijay Kumar, Narayajee and several others during the field excur-
sion helped immensely to develop the initial ideas in this paper.
Received 2 June 2009; accepted 7 July 2009
A novel nesting behaviour of a
treefrog, Rhacophorus lateralis in
the Western Ghats, India
S. D. Biju
Systematics Lab, Centre for Environmental Management of
Degraded Ecosystems, School of Environmental Studies,
University of Delhi, Delhi 110 007, India
Nest building by leaf folding is a rare behaviour in
anuran amphibians, with previous reports for only two
genera, the Subsaharan African Afrixalus, and Cen-
tral and South American Phyllomedusa. This commu-
nication reports a specialized nest building behaviour
of an Indian treefrog Rhacophorus lateralis, which was
observed in natural habitat at Kalpetta in Wayanad
District, Kerala. This behaviour of leaf folding is the
first report in the family Rhacophoridae, and in the
Asiatic amphibians. Nesting behaviour of R. lateralis is
unique among Rhacophorus – a purse-like nest is
made over water by folding a single leaf around the
egg mass (embryos and translucent foam) by the
female alone after oviposition. The function of this
parental investment is to prevent desiccation of eggs in
open sunlight. This paper also documents the multiple
leaf nesting behaviour of other two species of this
genus, R. calcadensis and R. pseudomalabaricus, and
the previously documented nesting behaviour of R.
malabaricus using more than one leaf.
Keywords: Leaf folding, leaf nesting, Rhacophorus lat-
eralis, treefrog, Western Ghats.
IN addition to typical aquatic habitats, anuran amphibians
deposit eggs in a wide range of places including under-
ground1, arboreal foam nests2, tree holes3 and stream
banks4. Among the 262 anuran amphibians reported from
India5, above-ground nest construction using multiple leaf
is known only in Rhacophorus malabaricus6.
Rhacophorus lateralis is a small sized Rhacophorid
treefrog (snout to vent size-male: 28.6–30.1 mm, N = 5;
female: 33.5–34.8 mm, N = 3) having bright green or
light reddish-green dorsal colour with a prominent golden
yellowish streak from snout along the side of head to near
the vent. Scientific knowledge on R. lateralis is sparse
other than the original description based on a sole pre-
served animal7, followed by rediscovery after a gap of
more than 100 years from the Western Ghats of Kerala8
and Karnataka9. Rhacophorus lateralis is an endangered
species10, thus a better understanding of the breeding bio-
logy of this frog is critical for its conservation manage-
A breeding population of R. lateralis was observed
over two breeding seasons during which courtship, mating
and leaf nesting behaviour were studied. The primary ob-
jective of this communication is to document leaf nest
construction behaviour of R. lateralis and determine the
possible function of this behaviour based on field obser-
vations and laboratory studies. The complete sequence of
courtship and mating behaviour of this species is beyond
the scope of this communication. This study is based on
observation of 65 nests, including nine sequences from
pair detachment after oviposition to completion of leaf
The study was conducted during 2000 and 2005 breed-
ing seasons (June–September) at Kalpetta (11°36′N,
76°05′E; 980 m asl), Wayanad District, Kerala. Amplexed
pairs were located by active searching guided by choruses
or by making repeated observations of single females un-
til they mated. Amplexus is axillary and duration of egg
laying varies from 35 to 50 min (N = 9). Fieldwork was
undertaken at a natural breeding pool between 19:00 and
23:00 h using a dim or red flashlight. The pool was 3.2 m
wide and had a maximum depth of 0.9 m. Grasses, low
herbs (Ludwigia sp., Lantana sp., Chromolaena sp., etc.).