to be
of the
d the
rface,
000).
irface
issive
erties
er, in
| land
bility
s are
could
Since
> the
y can
ration
aphic
h for
logic
small
suited
naller
ns of
ented,
rface
reater
1ents.
-time
ighly
and
soil
cover
e the
eters
data
d lies
ndia.
year,
rain
field
6 ie.
es to
f the
nonts
d by
> soil
The
The
| 30?
itude
any
e a methodology crisis
There are hardly any means to unravel all of these crises
together. If you do not have good quality data, your model
cannot give accurate results and similarly if your methodology
is not proven, the results will be wrong. Therefore all the three
crises are interrelated to each other. In the data crises it is very
difficult to get good quality data. In most cases needed data
either does not exist or are not available in full. Even if the
needed data are available, problems remain with regard to
incompleteness, inaccuracy and inhomogeneity of data. Hence
interpolation and extrapolation techniques becomes useful. The
temporal and the spatial scales are the most important
contributing factors in data crises. The temporal domain is an
indicator of the phenology while spatial scale is dependent on
the level of information and can be useful to classify models
into small-watershed, medium-size watershed and large
watershed models (Singh, 1995). Therefore a classification is
arbitrary and is experimental rather than conceptual and is
governed by data availability rather than physical meaning
(Singh, 1995). To alleviate the problem of lack of adequate data
in hydrology, remote sensing plays an important role to
quantitatively describe a hydrologic process accurately
(Schultz, 1988).
The model crisis lies between the selections of empirical
models versus physically based models. Empirical models are
easy to use, but their application is limited to the areas where
they were developed. They are based on statistical observations.
While physically based models are of most universal use but
they require a lot of data, which are rarely available (Morgan,
1986) and are based on physical laws.
There are some queries that what are the most important
sources of hydrological variability in a given environment and
can they be detected by remote sensing. Presently remote
sensing data is now routinely used for land use classification at
different spatial and temporal scales. However, the
multispectral sensor data provide no direct estimates of
hydrological variables. Other electromagnetic region such as
active microwave and SAR sensors are also dependent on the
parameterization of surface roughness and emissivity before
any estimation of hydrological variables can be achieved. This
is true also of estimates of surface reflectance and estimates of
leaf area index, and estimates of surface resistance and
evapotranspiration. All these derived estimates of
hydrologically relevant variables and parameters depend
themselves on models, models that have parameters must be
calibrated. It has been found in several studies that calibration
of even simple models may not be very robust; and the values
determined for particular parameters may be dependent on the
other parameters in the model as well on the model structure
and input data set (Beven, 1993; Freer et al., 1996). There is not
just one parameter set (or model structure) that is compatible
with the data. The implication of this study would be that any
physical interpretation of parameter values derived in this way
must be made with extreme care, and that extrapolation of such
values to other circumstances may be difficult. It will not be
therefore be easy to assign parameter values on the basis of
vegetation type. Since vegetation type might be considered to
be primary variable in terms of derivation from remote sensing,
secondary variables such as soil moisture and evapo-
transpiration estimates should be in general are expected to be
highly uncertain. This uncertainty does not bode well for any
solution to the problem, which will ultimately depend on
knowledge of patterns of hydrologically relevant variables for
both theory development and verification. Several studies had
been carried out which provide reviews of the possibilities for
IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring”, Hyderabad, India, 2002
743
the use of remote sensing in hydrology for water and soil
erosion (Vigiak, O.and G. Sterk, 2002; Vrieling, V. et al.,
2002). Remote sensing may be used in the estimation of input
data (including topography, rainfall and evapotranspiration
rates), state variables (including soil moisture) and parametric
values (mostly derived through the classification of soil and
vegetation types from remote sensing) (Beven, K.J., 2002)
The methodology crisis arises from the fact that the complexity
of a watershed cannot be measured with current measurement
techniques (Beven and Kirkby, 1979). Each catchment presents
a unique set of topography, soil, vegetation and anthropogenic
characteristics, whose representation in a model leads to an
often overlooked degree of uncertainty (Beven, 1997). On the
other hand, there is a need for explorations, in order to support
decision-makers with information concerning the consequences
of their plans. This need should be met with available data, or
data whose collection is feasible and economically sound and
whose uncertainty are made explicit.
The present work is under the umbrella of IIRS-ITC, IHE, WU
Phase II project on Environmental Analysis and Disaster
Management (GEONEDIS). The objective of the work is to use
microwave remote sensing for distributed water erosion
modeling in a Himalayan catchment. The study envisages
following objectives:
e To estimate effective parametric values for
hydrological modeling
e Evaluation and validation of model using remote
sensing data sets (optical and microwave)
3. METHODOLOGY
To estimate runoff for the entire watershed, three types of data
will be used i.e. remote sensing data (optical and microwave),
climatic data (automatic weather station and rain gauges) and
field data. To derive the effective rainfall-runoff parameters of
the entire watershed, it is suggested to divide the entire
watershed into sub units (mesh) of equal area (50x50 m) or
uniform grid size. Those effective parameters can be estimated
by measuring the individual sub unit, which can further be
extrapolated depending upon the similar spatial land use
characteristics. The composition of the sub units includes land
cover type, soil attributes (soil type, infiltration, texture etc),
weather data, topographic attributes (microwave data) and soil
profiles (microwave data). All the dataset generated will be
used in KINEROS model for the runoff estimation. The model
will be validated using runoff recorder. The estimated runoff
values will be major input for soil erosion loss (Fig.1)
Climatic data
(Automatic Weather Station
and Rain guage)
[ Remote Sensing Data
Ground Truth
Field data
Sampling
"Soil type
Microwave
Surface
| Precise DEM Roughness
| Soil Moisture
Weather parameters
SAC KINEROS *Rainfall data
*Texture *Evapotranspiration
iyeiraulic Process Based Model »Relative humidity
conductivity «Temperature
*Volumetric
moisture
Validation of the model
using
Runoff Recorder
Soil Erosion Loss
