15 Introduction to and General Aspects of Water Management
333
series, has been in the domain of crop yield monitoring and forecasting (Groten,
1993). Regionalization studies could eventually benefit from the information once
correlations have been established between the NDVI values aggregated over gauged
watersheds and rainfall minus streamflow figures. The hydrological significance of
a change has to be assessed and the water manager is faced with the question whether
a trend can be detected in the observed changes or whether the changes are within
natural variations in time for the phenomenon under study. A study for the water
management of Amboseli National Park (south Kenya) illustrates how remotely
sensed data could contribute to the answer. Important fresh water springs in that area
feed large marshes in an otherwise saline, dry plain. Large herds of herbivores depend
on the marshes. In a period of a few years only, the marshes expanded and a small
lake was formed, which caused problems for the tourist infrastructure. No discharge
records were available. During a field study in 1992, it was found that the discharges
equaled the size of the marsh times the potential evapotranspiration rate. Aerial
photographs and satellite imagery of nine different dates, during the period 1950 -
1992, have been used to measure the size of the swamps and therefore indirectly the
spring discharges. Because appreciable changes in swamp size were observed, it was
concluded that the noted recent increase of the discharge of the springs were within
the expected range of variation (Meijerink and van Wijngaarden, 1996).
Upstream-downstream problems are common in many catchments. Deforestation
in the upper watersheds can cause severe problems in the downstream alluvial areas
and coastal zones. The identification and quantification of the downstream damages
is required to draw the attention of the decision makers involved, often this can only
be done with the aid of remote sensing. An example is given of the damages to
productive rice lands on an alluvial floodplain of the Komering River in South
Sumatra, Indonesia. In the Komering alluvial plain, upstream of the inland delta (see
Fig.15.3 in Sect. 15.3.3), loss of rice lands in the backswamps occurs due to waterlogging, and is becoming worse year by year. The waterlogging is caused by a rise of
the sandy river bed, which prevents the drainage of water out of the backswamps into
the river after the rainy season. The progressive aggradation of the river bed can be
attributed to an increased supply of sand and gravel from the upper catchment where
important deforestation took place of hilly areas consisting of sandy tuffs (Meijerink
et aI., 1988). Remote sensing allowed identification and mapping of the damages in
the floodplain area. The relative depths of the back swamps before the changes can
be accurately mapped on infrared aerial photography of 1976 (see Color Plate 15.A),
because rice is planted in stages which follow the recession of the flood waters after
the monsoon. At the time of photography the deeper parts of the backswamps still
contained aquatic weeds in the rice fields (high reflectance). The rice, planted on the
higher parts of the backswamps, had partial canopy cover, medium reflectance and
the intermediate part was still under water for transplanting (no reflectance). A more
recent situation was assessed using a SPOT color composite image (Colour Plate
15.A), overlaid with a relative height classification, to locate the areas (backswamps)
where changes could have taken place (compare expansion of swamp areas between
the two inset maps on Colour Plate 15.A). If field patterns were not present in the
deeper backswamps, the area had developed in a swamp. No other interpretations or
333
series, has been in the domain of crop yield monitoring and forecasting (Groten,
1993). Regionalization studies could eventually benefit from the information once
correlations have been established between the NDVI values aggregated over gauged
watersheds and rainfall minus streamflow figures. The hydrological significance of
a change has to be assessed and the water manager is faced with the question whether
a trend can be detected in the observed changes or whether the changes are within
natural variations in time for the phenomenon under study. A study for the water
management of Amboseli National Park (south Kenya) illustrates how remotely
sensed data could contribute to the answer. Important fresh water springs in that area
feed large marshes in an otherwise saline, dry plain. Large herds of herbivores depend
on the marshes. In a period of a few years only, the marshes expanded and a small
lake was formed, which caused problems for the tourist infrastructure. No discharge
records were available. During a field study in 1992, it was found that the discharges
equaled the size of the marsh times the potential evapotranspiration rate. Aerial
photographs and satellite imagery of nine different dates, during the period 1950 -
1992, have been used to measure the size of the swamps and therefore indirectly the
spring discharges. Because appreciable changes in swamp size were observed, it was
concluded that the noted recent increase of the discharge of the springs were within
the expected range of variation (Meijerink and van Wijngaarden, 1996).
Upstream-downstream problems are common in many catchments. Deforestation
in the upper watersheds can cause severe problems in the downstream alluvial areas
and coastal zones. The identification and quantification of the downstream damages
is required to draw the attention of the decision makers involved, often this can only
be done with the aid of remote sensing. An example is given of the damages to
productive rice lands on an alluvial floodplain of the Komering River in South
Sumatra, Indonesia. In the Komering alluvial plain, upstream of the inland delta (see
Fig.15.3 in Sect. 15.3.3), loss of rice lands in the backswamps occurs due to waterlogging, and is becoming worse year by year. The waterlogging is caused by a rise of
the sandy river bed, which prevents the drainage of water out of the backswamps into
the river after the rainy season. The progressive aggradation of the river bed can be
attributed to an increased supply of sand and gravel from the upper catchment where
important deforestation took place of hilly areas consisting of sandy tuffs (Meijerink
et aI., 1988). Remote sensing allowed identification and mapping of the damages in
the floodplain area. The relative depths of the back swamps before the changes can
be accurately mapped on infrared aerial photography of 1976 (see Color Plate 15.A),
because rice is planted in stages which follow the recession of the flood waters after
the monsoon. At the time of photography the deeper parts of the backswamps still
contained aquatic weeds in the rice fields (high reflectance). The rice, planted on the
higher parts of the backswamps, had partial canopy cover, medium reflectance and
the intermediate part was still under water for transplanting (no reflectance). A more
recent situation was assessed using a SPOT color composite image (Colour Plate
15.A), overlaid with a relative height classification, to locate the areas (backswamps)
where changes could have taken place (compare expansion of swamp areas between
the two inset maps on Colour Plate 15.A). If field patterns were not present in the
deeper backswamps, the area had developed in a swamp. No other interpretations or