SOME ASPECTS ON THE DEVELOPMENT OF HYDROLOGICAL FORECASTING
MODELS BASED ON MULTI-TEMPORAL SATELLITE REMOTE SENSING
A. Narayana Swamy', Pietro Alessandro Brivio^
* Department of Geophysics, Andhra University. Visakhapatnam 530 003, India — answamyæhotmail.com
' CNR-IREA, Via Bassini, 15 20133 Milano. Italy — brivio.pa@irea.cnr.it
Commission IV, WG IV/1
KEY WORDS: Remote Sensing, Digital, Hydrology, Snow, Classification, Landsat, Multi-spectral. Multi-temporal
ABSTRACT:
l'orecasting models for daily river discharge in the rivers of high mountainous catchments arc important for management of water
resources towards hydropower generation, irrigation, industrial and domestic water supply as well as flood control. The complex
topography in high mountainous terrain create extremely difficult situation for observation and sampling of meteorological and
hydrological parameters that are useful to model hydrological processes. Satellite remote sensing data in different wavelength
regions of electromagnetic spectrum arc most uscful to extract near real time hydro-geophysical parameters to support development
ol hydrological forecasting models. In the present study digital image processing of multi-spectral, multi-temporal and multi-sensor
satellite data has been carried out over two high mountainous river catchments of varied area extent viz.. Cordevole (248 sq. km.)
and La Vizza (7.5 sq. km.). The above catchments are located in the dolomites region of north-eastern Italian Alps. The elevation of
the region varies between 900 meters to 3150 meters above sea level, Nine sets of Landsat 4 and 5 Multi-spectral Scanner (MSS) and
Thematic Mapper (TM) computer compatible tapes covering a hydrological year have been processed in the study using digital
image processing techniques. Digital elevation model, slope, aspect and shadcd rclicf maps coinciding with satellite passes were
generated using locally developed Territorial Image Synthesis System (TISS) and utilized in the analyses. The areal extent of snow
cover has been extracted by taking the catchment as a single unit as well as three elevation zones. Supervised parallelepiped. ncarcst-
neighbourhood and maximum likelihood classification methodologies have been utilized to estimate the snow cover under different
categories. A threshold based second and third order polynomial fit approach has been attempted to generate accurate daily snow
cover distribution over the catchments as well as the elevation zones. A semi-distributed deterministic hydrological model has been
developed based on the catchment characteristics to generate river discharges. Model performance evaluation indicates excellent
correlation between measured and simulated discharges and the results are comparable with World Meteorological Organization
(WMO) test basins. :
|, INTRODUCTION are to a great extent responsible for the flow pattern in winter
and spring months,
Precipitation is one of the most important inpuls of
hydrological and water management systems. In the climate of The importance of satellite remote sensing data as a potential
several parts of the world solid precipitation and snow-cover source from complex high mountainous terrain has been widely
are of significance to the precipitation regime. About 30% of recognized. The capabilities of optical and microwave satellite
the carth's land surface is scasonally covered by snow. In high remotc scnsing technology enable us to map the snow cover on
mountains the proportion of snow in precipitation exceeds 50% regular basis under adverse weather conditions also. When
and the snowmelt becomes a dominant factor in the runoff snowmelt starts, the snow conditions change and the high
regime. The alpine snow cover in the middle latitudes have a mountain basins arc not accessible for regular snow surveys,
very large storage capability of water and consequently forms a However, daily observation by NOAA Advanced Very High
water reservoir during melting season. Snowmelt accounts for Resolution Radiometer (AVHRR) make it possible to monitor
30 to 80% of the annual stream flow of many rivers originating snow cover changes from satellite over large high mountainous
[rom Sierra Nevada, Rocky, Alps, Andes and Himalayas (Ferris regions, However, to study the snow cover changes due to the
and Congalton 1989). In high mountain regions the distribution impact of climate on small catchments moderate to high spatial
of snow controls the weather and climatic conditions and plays resolution sensors like Landsat — MSS and TM, SPOT — HRV,
an important role in the hydrological processes of the region. IRS — LISS I, II, IH, IV are quite uscful. However, the
Among the scientific community approaching the problems repetitive cycles of the above satellite sensors are insufficient
related to global changes, there is a general agreement that the for frequent periodical snow cover mapping. since all the
temperature and precipitation are affected by the rising of green images cannot be used due to extensive cloud cover or
house gases in the atmosphere. Such changes may influence acquisition after new snowfall.’ A successful forecast of the
significantly the runoff regime of high alpine environment. The melt water runoff’ needs fairly accurate measurements of the
snowmelt runoff is being used for hydropower generation, area covered by snow and continuous monitoring of the
domestic, irrigation and industrial water supply. Hence, changes about the gradually decreasing snow cover. Hence.
hydrological forecasting of snowmelt runoff from catchments various — mcthodologies are to be adopted — to
located in high mountainous terrain is very important for interpolate/extrapolate the snow cover derived from few
judicious management of waler resources. Any modeling satellite scenes for longer periods at shorter timc intervals, in
approach to the study of water balance in alpine areas should order to input the samc in to various hydrological forecasting
lakc in to account snow accumulation and melt processes which
1221
