332
A.MJ. Meijerink and C.M.M. Mannaerts
15.2.2 Spatial analysis and regionalization
Remote sensing also makes it possible to prepare a quantitative analysis of water
balance components at a wide range of scales, ranging from poor water distribution
problems in irrigation areas (Menenti et aI., 1989) to delineations of main hydrological terrain units, which may be termed ,,hydrotopes" after Engelen and Venneker
(1990). Terrain units group natural associations between lithology, geomorphology
and soils, all of which influence the local hydrology. Conceptually these spatial units
can be considered as areas with a typical set of hydrologic responses. An essential
characteristic of hydrological terrain units is that their boundaries can in many cases
be deduced from remotely sensed imagery, using a pragmatic, open-ended classification scheme. Knowledge of the effects of terrain factors on the hydrology should
assist in formulation of criteria for their delineation. Without analytical image
interpretation, the hydrological terrain units can, to a certain extent, be compiled in
a GIS environment by combining a digital geologic map, topographic derivatives
(e.g., slopes), hydrological features (e.g., drainage, lakes, wetlands) and a vegetation
cover classification. Land use and vegetation patterns generally reflect the spatial
distribution of terrain units, except where large-scale human interventions of land
cover alterations have been implemented. Examples of the latter are large developments of agricl..lltural farmland or tree plantations in many temperate or tropical
countries. Descriptions of some examples (e.g., Colour Plate 14.B) will further
illustrate hydrologic terrain units, how they can be differentiated on imagery and what
their role in the regionalization is. For example, it is obvious that runoff, groundwater
and sediment data of one of the four units of Fig. 15.4 (see Sect. 15.4.2 and illustrating the Handeni area in Tanzania), cannot be extrapolated to anyone of the other
units in that zone, despite the fact the total area considered is small in size. In other
regions, a hydrologic terrain unit can be huge and extend over several thousands of
square kilometers, for example, the vast tracts of the plateau basalts - Deccan traps
- in India. This is due to uniformity in lithology, geomorphology and soil associations, apparent on the imagery. Hydrologic data and water management practices from
a sub-catchment in such an environment can be meaningfully extrapolated to neighboring sub-watersheds. In complex terrain, extrapolation is much more difficult.
15.2.3 Monitoring and forecasting
The prognosis and monitoring of hydrologic phenomena by remote sensing usually
rely on the use of image time series or multi temporal images from a same area. The
idea is to find an empirical correlation between features measured on imagery and
ground hydrometric data. Ifthe two are correlated, the relationship can be used to
reduce hydrometric ground operations, usually difficult or expensive, or to fill in gaps
in the record. Obvious applications are evaporation estimations from seasonally
variable swamp areas, or prediction of snowmelt runoff from snow cover.
For many parts of the world normalized differential vegetation index or NDVI maps
are produced for lO-day periods in a routine fashion (Hielkema, 1990). The NDVI
values are related to vegetation density and therefore to actual evapotranspiration loss
but also reflect the occurrence of rainfall. Until now, the operational use of these time
A.MJ. Meijerink and C.M.M. Mannaerts
15.2.2 Spatial analysis and regionalization
Remote sensing also makes it possible to prepare a quantitative analysis of water
balance components at a wide range of scales, ranging from poor water distribution
problems in irrigation areas (Menenti et aI., 1989) to delineations of main hydrological terrain units, which may be termed ,,hydrotopes" after Engelen and Venneker
(1990). Terrain units group natural associations between lithology, geomorphology
and soils, all of which influence the local hydrology. Conceptually these spatial units
can be considered as areas with a typical set of hydrologic responses. An essential
characteristic of hydrological terrain units is that their boundaries can in many cases
be deduced from remotely sensed imagery, using a pragmatic, open-ended classification scheme. Knowledge of the effects of terrain factors on the hydrology should
assist in formulation of criteria for their delineation. Without analytical image
interpretation, the hydrological terrain units can, to a certain extent, be compiled in
a GIS environment by combining a digital geologic map, topographic derivatives
(e.g., slopes), hydrological features (e.g., drainage, lakes, wetlands) and a vegetation
cover classification. Land use and vegetation patterns generally reflect the spatial
distribution of terrain units, except where large-scale human interventions of land
cover alterations have been implemented. Examples of the latter are large developments of agricl..lltural farmland or tree plantations in many temperate or tropical
countries. Descriptions of some examples (e.g., Colour Plate 14.B) will further
illustrate hydrologic terrain units, how they can be differentiated on imagery and what
their role in the regionalization is. For example, it is obvious that runoff, groundwater
and sediment data of one of the four units of Fig. 15.4 (see Sect. 15.4.2 and illustrating the Handeni area in Tanzania), cannot be extrapolated to anyone of the other
units in that zone, despite the fact the total area considered is small in size. In other
regions, a hydrologic terrain unit can be huge and extend over several thousands of
square kilometers, for example, the vast tracts of the plateau basalts - Deccan traps
- in India. This is due to uniformity in lithology, geomorphology and soil associations, apparent on the imagery. Hydrologic data and water management practices from
a sub-catchment in such an environment can be meaningfully extrapolated to neighboring sub-watersheds. In complex terrain, extrapolation is much more difficult.
15.2.3 Monitoring and forecasting
The prognosis and monitoring of hydrologic phenomena by remote sensing usually
rely on the use of image time series or multi temporal images from a same area. The
idea is to find an empirical correlation between features measured on imagery and
ground hydrometric data. Ifthe two are correlated, the relationship can be used to
reduce hydrometric ground operations, usually difficult or expensive, or to fill in gaps
in the record. Obvious applications are evaporation estimations from seasonally
variable swamp areas, or prediction of snowmelt runoff from snow cover.
For many parts of the world normalized differential vegetation index or NDVI maps
are produced for lO-day periods in a routine fashion (Hielkema, 1990). The NDVI
values are related to vegetation density and therefore to actual evapotranspiration loss
but also reflect the occurrence of rainfall. Until now, the operational use of these time