134
B.G.H. Gorte
cover induces net recharge. Examples are given of groundwater indicators and reference is made to the adaptation of vegetation to groundwater quality that may exist
in certain regions. Finally, most wetlands are situated in a hydrological environment
where vegetation reflects the variable inundation depths and durations, as influenced
by the micro-topography and corresponding soil textures. In coastal regions, water
salinity determines the vegetation associations.
7.2 Land cover Mapping with Remote Sensing
The various types of land cover govern much of the spectral reflection of the surface
of the earth, which is measured by sensors on various remote platforms, such as
multi-spectral and thermal scanners or push-broom arrays and active microwave
(radar) imaging systems.
In most parts of the world, the land cover is highly dynamic. Apart from effects
of seasonal rainfall, temperature and possible cyclicity of rainfall and droughts, man
has influenced the vegetation by converting natural vegetation into agricultural lands
where different crop rotations are practiced. Other dynamic aspects are the variable
grazing pressure of range lands, forest fires, destruction of flood plain vegetation by
floods, and so on. The frequent over-passes of the earth observation satellites allow
one to monitor the changes. Since satellite observations are available since the early
1970's, it is possible to relate e.g trends in cover densities to stream flow.
The above brief review highlights the major role of vegetation in hydrology, but
- implicitly - also points out that the effect of vegetation on hydrological processes
has to be understood through field studies in order to apply transfer functions to convert remotely sensed cover data into (relative) hydrologic quantities or to parameter
values for model input. The transfer functions are related e.g. to actual evapotranspiration, either through identification of crop types, followed by agro-ecologic models to predict transpiration or by functions relating evapotranspiration to remotely
sensed vegetation indices. Some hydrologic models have in-build transfer functions,
such as the SIMPLE model of Kouwen et al. (1990), as well as the models that use
the curve number methods. Physically based remotely sensed parameters related
to cover are perhaps limited to the actual evapotranspiration using surface energy
balance methods.
The influence of scattering on the radar signal (images) is still in the research
domain. Furthermore, classifications using remotely sensed data should aim at those
classes which are meaningful in a hydrological sense. For example, dense vegetation
near the surface strongly influences the surface hydrological processes, regardless
of the botanical composition of that cover. Hence, spectral cover classes related
to dense vegetation could be combined, unless experimental data is available to
do otherwise. The coupling to field observations is essential; forest canopies are
recorded by remote sensing, not the condition of the undergrowth, which determines
much of the processes.
The remainder of this chapter is divided in three parts, dealing with (a) vegetation indices, (b) image classification and (c) the use of radar imagery. The separa-
B.G.H. Gorte
cover induces net recharge. Examples are given of groundwater indicators and reference is made to the adaptation of vegetation to groundwater quality that may exist
in certain regions. Finally, most wetlands are situated in a hydrological environment
where vegetation reflects the variable inundation depths and durations, as influenced
by the micro-topography and corresponding soil textures. In coastal regions, water
salinity determines the vegetation associations.
7.2 Land cover Mapping with Remote Sensing
The various types of land cover govern much of the spectral reflection of the surface
of the earth, which is measured by sensors on various remote platforms, such as
multi-spectral and thermal scanners or push-broom arrays and active microwave
(radar) imaging systems.
In most parts of the world, the land cover is highly dynamic. Apart from effects
of seasonal rainfall, temperature and possible cyclicity of rainfall and droughts, man
has influenced the vegetation by converting natural vegetation into agricultural lands
where different crop rotations are practiced. Other dynamic aspects are the variable
grazing pressure of range lands, forest fires, destruction of flood plain vegetation by
floods, and so on. The frequent over-passes of the earth observation satellites allow
one to monitor the changes. Since satellite observations are available since the early
1970's, it is possible to relate e.g trends in cover densities to stream flow.
The above brief review highlights the major role of vegetation in hydrology, but
- implicitly - also points out that the effect of vegetation on hydrological processes
has to be understood through field studies in order to apply transfer functions to convert remotely sensed cover data into (relative) hydrologic quantities or to parameter
values for model input. The transfer functions are related e.g. to actual evapotranspiration, either through identification of crop types, followed by agro-ecologic models to predict transpiration or by functions relating evapotranspiration to remotely
sensed vegetation indices. Some hydrologic models have in-build transfer functions,
such as the SIMPLE model of Kouwen et al. (1990), as well as the models that use
the curve number methods. Physically based remotely sensed parameters related
to cover are perhaps limited to the actual evapotranspiration using surface energy
balance methods.
The influence of scattering on the radar signal (images) is still in the research
domain. Furthermore, classifications using remotely sensed data should aim at those
classes which are meaningful in a hydrological sense. For example, dense vegetation
near the surface strongly influences the surface hydrological processes, regardless
of the botanical composition of that cover. Hence, spectral cover classes related
to dense vegetation could be combined, unless experimental data is available to
do otherwise. The coupling to field observations is essential; forest canopies are
recorded by remote sensing, not the condition of the undergrowth, which determines
much of the processes.
The remainder of this chapter is divided in three parts, dealing with (a) vegetation indices, (b) image classification and (c) the use of radar imagery. The separa-