Content uploaded by Bingwei Tian
Author content
All content in this area was uploaded by Bingwei Tian on Aug 14, 2019
Content may be subject to copyright.
MAPPING DIGITAL DRAINAGE NETWORK USING GEOPROCESSING: A CASE STUDY
OF KALI GANDAKI RIVER BASIN, NEPAL HIMALAYA
1Feiyu Chen, 1Bingwei Tian*, 12Basanta Raj Adhikari, 1Xiaoyun Gou
1Institute for Disaster Management and Reconstruction Sichuan University-The Hong Kong Polytechnic
University
2Institute of Engineering, Tribhuvan University, Nepal
Shuangliu, Chengdu, 610207, China. * Email: bwtian@scu.edu.cn
ABSTRACT
The drainage network characteristics of watershed has a
significant impact on mountain landscape. The Kali
Gandaki River basin has various topographic variation with
geomorphology to be crossed by the Sino-Nepal road
corridor. In this research, the ALOS PRISM data is used to
extract the watershed DEM, and then the drainage network
is automatically extracted using geoprocessing methods.
Digital water networks extracted from DEM of different
filming years were compared. Automatic drainage
extraction has high precision and details information than
manual. The river morphology and local landscape has
change over the years due to climatic changes, mountain
hazards and human activities.
Index Terms—DEM, Kali Gandaki River Basin,
Drainage Network, Nepal-Himalaya
1. INTRODUCTION
The impact of climate change on water resources is
considered one of the major challenges facing the Hindu
Kush Himalayas[1–3], one of the most sensitive regions
because of the highly diverse climate and topography[4].
The Kali Gandaki River basin lies in the western part of
Nepal which has diverse topography and geomorphology.
Pokhara-Korala road corridor is one of the major
connecting road between Nepal and China under the broad
umbrella of Belt and Road Initiatives taken by Chinese
government. Apart from this, there are many hydropower
stations in this watershed i.e Kaligandaki Hydropower
Project (114 MW). Drainage network analysis is always
important for infrastructure construction and understand the
geomorphology. However, there are very few studies exist
in this area. Therefore, an attempt has been done to analyze
the drainage network typology and geometry using different
Remote sensing (RS) and Geographic Information System
(GIS) in this watershed. Drainage network topology and
geometry form the basis of many hydrological and
geomorphologic models. The digital basin feature is a
digital representation of the characteristics of drainage
networks. Accurate watershed boundaries extraction is
depends on geomorphology, original maps, and
measurement methods. Selecting a suitable grid-scale DEM
to extract digital basin characteristics such as river network,
slope, and watershed range is a basic task for establishing
distributed hydrological modeling and in-depth research. In
this research, ALOS PRISM data was used to generate
Digital Elevation Model (DEM). Depressions in the DEM
are identified and filled. The flow direction and catchment
area are determined using the D8 algorithms[5]. Then, the
Geoprocessing model is built to extract the digital basin
feature. Finally, drainage network from different years are
compared.
2. METHODS
2.1. Study Area And Data Sourse
Kali Gandaki River originates from the southern edge of the
Tibetan Plateau. This river basin is bounded by Neogene
extensional tectonics in the Tibetan Plateau and Himalaya,
and rugged glacial topography with some gentle slopes
(Figure 1). All of the tributaries originate from Himalayan
glaciers. For local residents, the mountain disaster and flood
risk have a profound impact on the production and daily life.
Figure 1: The map of the study area
3479978-1-5386-9154-0/19/$31.00 ©2019 IEEE IGARSS 2019
River data in 2001 has been prepared by merging sheet-
wise individual layers available from Survey Department,
Nepal. A 12.5m resolution DEM data is processed using
ALOS PRISM L1B2 data, shooting in 2016, to construct a
digital drainage network model[6]. Three kinds of DEM
(SRTM C-band, ALOS World 3D and ASTER GDEM
Version 2) are used to compare the change of fluvial
morphology[7].
Table 1: List of DEMs used for comparisons and
geomorphic analyses.
Dataset
Type
Time
Source
SRTM1-
V3
Radar /
30m
2000
https://earthexplorer.usgs.gov/
AW3D
Optical
/30m
2006-2011
https://www.eorc.jaxa.jp/ALO
S/en/aw3d/index_e.htm
Commercial
GDEM2
Optical
/ 30m
2006-2011
https://earthexplorer.usgs.gov/
ALOS
PRISM
Optical
/2.5m
2016
https://auig2.jaxa.jp/openam/U
I/Login
2.2. Processing Techniques
Based on the manually digitized watershed boundaries, a
buffer zone of 20km in the periphery is established. DEM
pretreatment and sag filling has been done in ArcGIS
platform. The slope runoff simulation has been done to
extract flow direction[8]. D8 algorithm is a single flow
algorithm and follows the basic principle that there are only
eight possible flows in a single grid. They are represented
by eight valid signatures 128, 1, 2, 4, 8, 16, 32, and 64. The
distance between the central grid and each adjacent grid is
calculated. The grid with the largest distance is the outgoing
grid of the central grid. If the cell size is 1, the distance
between two orthogonal cells is 1, and the distance between
two diagonal cells is 1.414. The algorithm is:
S = Z / D
Z is the height difference between the two grid cells, and D
is the distance between the centers of the two grid cells
(Figure 2).
Figure 2: The basic principle of the D8 algorithm
After the flow direction calculation, upstream confluence
matrix is calculated and then different critical support area
(CSA) are set to extract the raster map of the river network.
After that, the river network raster diagram is graded,
segmented and vector river network is extracted. At the
same time, according to the stream segment and flow
direction, the hydrological response unit is extracted. River
network has been classified based on Strahler grading[9].
Compared with Shreve grading[10], this algorithm has higher
extraction efficiency and clearer expression of the result.
Finally, the digital drainage model of the Kali Gandaki
River is prepared (Figure 3). All processes can be
implemented in the hydrological analysis module of ARC
GIS, and scripted using Python language.
Figure 3: Methodology chart
The Python scripts developed for this study will be bundled
into an Esri ArcGIS® toolbox for public download, making
them readily accessible to other researchers.
3. RESULTS AND DISCUSSION
3.1. Digital Drainage
In order to extract drainage system more accurately,
multiple critical support areas, ranging from 0.78 to 7.8
square kilometers, are set in this study to compare the
extraction effect of river network under different catchment
areas. The higher the density, the more fragmented the
ground, the greater the average slope, the lower the stability
of surface materials, and the formation of surface runoff,
soil erosion intensified. It is found that the density of river
network and river nodes are well correlated with the CSA
(Figure 4). The relation between the density and the CSA
can be well fitted (Figure 5) as the power function curve,
the formula is as follows:
01563.06585.0)(f 4939.0
xx
Calculation of open street map drainage density is 0.33 km-1.
According to the results of the power function fitting, when
the critical catchment area is 4.5 square kilometers, it is the
closest to the actual river network density. The outputs from
the model shows a good result with the existed watershed
map and OpenStreet map. The drainage system of the Kali
Gandaki River is oriented nearly in north-south direction.
The Syang Khola, Lumbuk Khola, Panda Khola, Jhon
Khola, Ghilumpa Khola, Narsing Khola, Tange Khola,
3480
Charan Khola, Dhechyan, Khola, Yak Khola and the
Hyujun Khola are the main tributaries of the Kali Gandaki
River. The watershed boundary is basically consistent with
the factual drainage network and with better details.
Figure 4: Drainage Network Extract by different CSA
Figure 5: The best fitted curve
3.2. Testing And Comparison
In order to understand the changes of Kali Gandaki River,
DEM data collected by ALOS PRISM in 2016 are
resampled into DEM with a resolution of 30m, and SRTM,
GDEM and AW3D30 data (30m resolution) were put into
the model to extract their digital watershed information
respectively (Table 2).
Table2. The drainage network characteristic
Figure 6: The contrast of drainage in SRTM1v3(2000)
ALOS3D (2006-2011) and ALOS prism (2016)
:
In the box
is the section with large deformation.
All processing share a same CSA, which is 5000 in raster
calculation. The results show that the local rivers have
undergone significant morphological changes in recent
years, reflecting the active erosion activities (Figure 6). The
Average fractal dimension is around 1.30 and shows an
increasing trend, which means the watershed landform is in
the infancy of erosion development. The river system is not
fully developed, and the river erosion is severe[11-12].
4. DISCUSSION
Rising temperature, erratic rainfall and change in monsoon
system in the Nepal Himalaya caused severe damage to
local livestock systems and agriculture[1,13]. The pattern
shows that the temperature rise is more active in high-
elevation than in low-elevation resulting glacial melting in
the Kali Gandaki river basin also. As temperatures warm
and glaciers retreat, many existing glacial lakes are growing
in size and new ones are forming[14,15]. As glaciers recede,
glacial lake levels are rising at an alarming rate, and the
threat of glacial lake outburst floods (GLOF) causing
catastrophic damage to people and infrastructure
downstream is growing[1]. These changes will also affect
the shape and flow of future water networks, creating new
challenges for local agricultural irrigation, production and
domestic water use.
The Kali Gandaki river cuts across the >8000 m high peaks
of Dhaulagiri and Annapurna, formed the world’s deepest
canyons in the gneiss of the greater Himalayas and the
metamorphic rocks (mainly quartzite and schist) of the
lesser Himalayas, where characterized by narrow terrain
and discontinuous schistose sedimentary terraces. Natural
disasters (such as landslide, collapse, glacial lake outburst,
etc.) often interact with rivers, causing river diversion and
landslide dam. This river basin also heavily affected by
DEM
Length
(km)
Subwatersh
ed No.
Source
No.
Area
(km2)
Density
(km-1)
FDavg
SRTM
2388.44
1253
890
6958.53
0.34
1.30
GDEM
2404.80
1277
903
6962.75
0.35
1.27
AW3D
2330.35
1273
884
6957.78
0.33
1.27
ALOS
PRISM
2259.22
1184
836
6956.81
0.32
1.34
FDavg:Average fractal dimension for selected area.
3481
such kinds of hazards causing sediment deposition and
change in local landscape. Anthropogenic activities i.e. road
construction, improper agricultural practices have been
largely affecting this river basin. Precipitation and
snowmelt are the most important variables affecting the
annual flow rate, and temperature is the most important
variable affecting the flow timing. The runoff of the
upstream and midstream of this river will increase by the
most or by 60% to 100%[1]. The synergistic effect of
temperature and precipitation will lead to the increase of
river flow and sink flow, and the final outlet flow may
increase by more than 50%[1]. Overall, local rainfall,
snowmelt, runoff and river flows will continue to increase,
and erosion may increase. The water network will not
shrink in the future, but may expand, especially at high
altitudes.
5. CONCLUSIONS
The utilization of GeoProcessing will help local water
resources utilization and management, as well as flood
disaster assessment and early warning. However, the
formation of drainage network is the common result of a
variety of natural and human factors within the basin. The
presence of large flat areas in depressions and plains makes
the spatial expression of topography and point elevation of
DEM inaccurate. The excavation of artificial channels and
canals will also affect the accuracy of extracting river
network water system based on DEM to varying degrees.
Climate change, mountain hazards and human activities
have significant effects on the spatial and temporal changes
of the water network in the Kali Gandaki basin.
6. ACKNOWLEDGEMENT
Funding: This work was supported by National Natural
Science Foundation of China(grant number: 71841027) and
Sichuan University Overseas Excellent Doctoral Research
Funding Scheme(grant number: skyb201603)
7. REFERENCES
[1] A.R. Bajracharya, S.R. Bajracharya, and A.B. Shrestha,
“Climate change impact assessment on the hydrological
regime of the Kaligandaki Basin”, Science of the Total
Environment, Nepal, pp. 837–848, 2018.
[2] PICC, Climate Change 2014–Impacts, Adaptation and
Vulnerability,Regional Aspects, Cambridge University
Press, 2014.
[3] W.W. Immerzeel, L.P. Vanbeek, M.F. Bierkens,
“Climate change will affect the Asian water towers,”
Science, American Association for the Advancement of
Science, vol.328, No.5984, pp. 1382–1385, 2010.
[4] P. Satyal , K. Shrestha, H. Ojha, “A new Himalayan
crisis? Exploring transformative resilience pathways,”
Environmental Development, Elsevier, vol.23, pp.47–56,
2017.
[5] A.B. Ariza-villaverde, F.J. Jimenez-hornero, G.D.
Ravee, “Influence of DEM resolution on drainage network
extraction: A multifractal analysis”, Geomorphology,
vol.241, pp.243–254, 2015.
[6] J. Takaku, N. Futamura, A. Goto, “High resolution
DEM generation from ALOS PRISM data,” Proc IEEE
Igarss, vol.57, No.5, pp. 405–407, 2004.
[7] B. Purinton, B. Bookhagen, “Validation of digital
elevation models (DEMs) and comparison of geomorphic
metrics on the southern Central Andean Plateau,” Earth
Surface Dynamics, vol.5, No.2, pp. 211–237, 2017.
[8] J.F. O’callaghan, D.M. Mark, “The extraction of
drainage networks from digital elevation data,” Computer
Vision Graphics & Image Processing, vol.28, No.3, pp.
323–344, 1984.
[9] L. Jiang, Q. Qi, A.Zhang, “River classification and river
network structuration in river auto-selection,” Geomatics &
Information Science of Wuhan University, vol.40, No.6, pp.
841–846, 2015.
[10] X. Xu, D. Zhang, S. Jia, “Automated extraction of
drainage in China based on DEM in GIS environment,”
Research And Environment In The Yangtze Basin, vol.13,
No.4, pp. 343–348, 2004.
[11] L. He, H. Zhao, “the fractal dimension of river network
and its interpretation,” Scientia Geographica Sinica, vol.16,
No.2, pp. 124–128, 1996.
[12] O.S. Al-kadi, “Texture measures combination for
improved meningioma classification of histopathological
images,” Pattern Recognition, vol.43, No.6, pp. 2043–2053,
2010, 43(6).
[13] Dahal, P., N.S. Shrestha, N. Krakauer, Climate change
and livestock system in mountain: Understanding from
Gandaki River basin of Nepal Himalaya, AGU Fall
Meeting Abstracts, 2015.
[14] X. Chen, P. Cui, Z. Yang, “Change in Glaciers and
Glacier Lakes in Boiqu River Basin, Middle Himalayas
during Last 15 Years,” Journal of Glaciology and
Geocryology, vol.27, No.6, pp. 793–800, 2005, 27(6).
[15] Ives, J.D., R.B. Shrestha, and P.K. Mool, Formation
of glacial lakes in the Hindu Kush-Himalayas and GLOF
risk assessment, 2010.
3482