This study explains the use of remote sensing data for spatially distributed hydrological modeling using the MIKE-SHE software used in Tarim River Basin CHINA
1. SEMINAR PRESENTATION
on
“On the usefulness of remote sensing input data for
spatially distributed hydrological modelling: case of the
Tarim River basin in China”
T. Liu, P. Willems, X. W. Feng, Q. Li,Y. Huang, An. M. Bao, X. Chen,
F. Veroustraete and Q. H. Dong (2012)
SHYAM MOHAN CHAUDHARY
17AG62R13
Land and Water Resources Engineering
Agricultural and Food Engineering Department
IIT KHARAGPUR
2. CONTENTS
• INTRODUCTION
• REVIEW OF LITERATURE
• PROBLEM STATEMENT
• OBJECTIVES
• METHODOLOGY
• RESULTS AND DISCUSSION
• CONCLUSIONS
3. INTRODUCTION
• Since the water shortage problem is becoming globally realized,
the demands on hydrological models to simulate scenarios and
develop efficient water resource management strategies are
increasing.
• Physically based hydrological models describe the natural system
using equations of mass, momentum and energy. The parameters
generally have direct physical significance.
• Remote sensing datasets of precipitation can be obtained using
GPCP, CMAP, TRMM , land surface temperature from NOAA-
AVHRR, MODIS, vegetation products Like global NDVI from NOAA-
AVHRR, vegetation indices using TERRA MODIS.
4. REVIEW OF LITERATURE
AUTHORS YEAR STUDY
Grayson et al. 2002
The applicability of spatially distributed
hydrological models is still rather
limited because of lack of spatially
variable input data or reference data
for model calibration
and validation.
Chen et.al. 2005
Implemented various spatial
meteorological data sets and leaf area
index (LAI) derived from LANDSAT in a
small Canadian watershed to track the
potential Evapotranspiration.
5. PROBLEM STATEMENT
• Since the 1950s excessive land reclamation, over-grazing and
unreasonable utilization of water resources in the upper
reaches of the Tarim river basin has intensified environmental
deterioration.
• Modelling activities are, however, not easy in the region,
because of the remote location in central Asia, the low
development and the related data poverty.
• There is a high need for spatially distributed hydrological
models, which allow agricultural and ecological impact and
scenario investigations.
6. OBJECTIVES
• To implement spatially distributed hydrological model for
selected sub-basins in the Tarim River basin, using the MIKE-
SHE modeling software.
• To compare the conventional use of Station Based (SB) data
versus the use of Remote Sensing (RS) data for evaluation of
observed daily discharges at river gauging stations and daily
snow cover maps.
7. METHODOLOGY
• STUDY AREA
– Tarim River Basin has area more than 550000 sq. km. and the
altitude ranges from 805 m to 5298 m. River has a length of
1324 km, and around 9 million inhabitants in the valleys along
this river.
– The maximum temperature in the Tarim basin is around 43 °C
and the minimum temperature around -4°C
– The average annual rainfall depth equals 33 to 270 mm
and the daily average sunlight around 8.2 h.
– The most dominant soil types are silt loam, fine sand, coarse
sand and sandy loam.
8. continued….
– The flood plains around the river reach are formed by
alluvial soils and are surrounded by dunes formed by wind
erosion.
– The vegetation cover consists of mostly drought tolerant
trees and shrubs, which have maximum root depths of
almost 4 m.
9. • MIKE SHE model description
The MIKE-SHE software is well known to split the water
movement into five parts:
• Overland flow
• Channel flow
• Evapotranspiration
• Unsaturated flow
• Saturated flow
10. continued….
• The overland flow calculation is done by
– Using the diffusive wave approximation of De Saint Venant
equations.
Use of the diffusive wave approximation allows the depth
of flow to vary significantly between neighboring cells and
backwater conditions to be simulated.
• The channel flow is calculated by
– The one-dimensional simulation of river flows and water
levels using the fully dynamic Saint Venant equations.
– The simulation of a wide range of hydraulic control
structures, such as weirs, gates and culverts.
12. continued…
In MIKE SHE, the ET processes are split up and modeled in the
following order :
• A proportion of the rainfall is intercepted by the vegetation
canopy, from which part of the water evaporates.
• Part of the infiltrating water is evaporated from the upper part
of the root zone or transpired by the plant roots.
• The remainder of the infiltrating water recharges the
groundwater in the saturated zone where it will be extracted
directly if the roots reach the water table, or indirectly if
capillarity draws groundwater upwards to replace water
removed from the unsaturated zone by the roots.
13. continued…
In MIKE SHE for calculating vertical flow in the
unsaturated zone, following options can be used:
• Richards Equation
• Gravity Flow
The simplified gravity flow procedure assumes a uniform
vertical gradient and ignores capillary forces.
14. continued….
The saturated flow modeling is based on the three-dimensional
Darcy equation
where Kxx, Kyy, Kzz are the hydraulic conductivity along the x, y
and z axes ,
h is the hydraulic head,
Q represents the source/sink terms, and
S is the storage coefficient
16. Model Setup
• The model is implemented on one headwater region, the
Kaidu River subbasin, and the area of the lower Tarim River
reach.
• The total modeled domain covers 132800 sq. km.
• A spatial resolution of 5 km was considered, providing a good
balance between spatial detail and limited computational
time.
• For precipitation, the correction lapse rate is 25% per 100 m
for altitudes below 2500 m and 6% per 100 m above 2500 m.
• The temperature correction lapse rate was set as 0.5 °C per
100 m altitude change.
• A crop coefficient of 0.2 is used to transform pan evaporation
to potential ET.
17. continued…
• The degree-day coefficient is set as 2.5 mm/°C/day. The initial
total snow storage was based on the snow cover RS data and
the cumulative precipitation depth of early winter period.
• The LAI was calculated based on the NDVI derived from
MODIS satellite RS data.
• LAI = 0.57 e 2.33 x NDVI
• The Chinese soil classification was transformed to the
standard triangle classification to relate the classes to
hydrological properties
18. Input data
XMB – Xinjiang Meteorological Bureau
TMB – Tarim Water Resources Bureau
XIEG – Xinjiang Institute of Ecology and
Geography
VITO – Flemish Institute for
Technological Research
A period of 2 years (from 1st Jan2000 till 31st
December 2001) with daily SB input data was
used for calibration, while RS products for 8
months (from 1 May 2005 till 31st Dec 2005)
and SB input data for 12 months (from 1st
January 2005 till 31st December 2005) were
used in support of model validation.
21. Validation of daily river flow results at BYBLK and DSK gauging stations using RS input data
22. Evaluation of daily river flow results at DSK and BYBLK station after SB and RS input data
When the model results after use of
RS versus SB inputs are being
compared, the snow accumulation
during the winter period is higher
and closer to the observed snow
cover area after RS input than that
based on the SB input. This shows
an advantage of the RS input due to
its precipitation spatial distribution.
23. CONCLUSIONS
• A spatially distributed hydrological model has been implemented
for the Kaidu and lower Tarim basins to obtain a decision support
tool for the management of the water resources at a regional level.
• This study confirmed that the availability of RS data provides an
interesting and valuable opportunity to partly overcome the lack of
necessary hydrological model input data in developing or remote
regions.
• RS data might have limitations of accuracy and time step. One
limitation encountered in the study for the snow cover product is
that this product of MODIS does not consider the frozen soil, which
is of high importance for the melted snow modeling.