390
M. Menenti
Traditional irrigation. The spatial resolution of current space-borne sensors is
adequate for detailed studies of relatively small irrigation schemes, even in
complex environments such as the Ecuador's Andes region studied by Goulven et
al. (1995) to characterise traditional irrigation systems. Practical methods were
developed on the Urcuqui irrigation system (12 635 ha) using stereo-pairs of
SPOT image data. Crops were identified with an accuracy of ",,70%, and 60% of
infrastructures were correctly mapped. A study of old irrigation systems in
Turkmenistan was presented by Volovik and Kochergin (1993).
Numeric classification algorithms are widely used in studies of land cover with
numeric satellite data. Criteria have been developed to assess the perfonnance of
numeric classification techniques. The performance of several Landsat 5 Thematic
Mapper (TM) image classification methods for crop estimates in an irrigation
district was studied by Barbosa et al. (1996). The agricultural land cover in a
26300 ha. irrigation district was classified utilizing two Landsat 5 TM scenes.
Manual and automatic selection of training areas for the classification of two
single subscenes and a combined multitemporal sub scene result in several
differently classified images. The classifications based on the training sets selected
manually provided the most precise results, with the exception of the single image
spring classification of rice. D'Urso and Menenti (1996 b) presented a general
approach to rank numeric classification procedures on the basis of statistical
indicators of accuracy, reliability and discriminability.
17.3.3 Crop water requirements - Visible and Near Infrared
Reflectance spectra in the 400 to 2500 nm is related to green biomass, Leaf Area
Index, leaf water content and chlorophyll, although this relation is not straightforward and crop-dependent. Bausch (1995) demonstrated that better estimates of
crop coefficients (and therefore of crop water requirements) were obtained for a
maize crop using spectral reflectance instead of traditional methods. Moran et al.
(1989) evaluated various spectral indices in relation with crop water stress. The
Perpendicular Vegetation Index (PVI) decreased with the stress-induced change in
architecture while the near-IR:red ratio (IRJR) remained relatively constant. As a
result, the near-IR:red ratio was more successful in estimating Medicago sativa
biomass than was the PVI. Fernandez et al. (1994) compared the sensitivity of
radiometric characteristics of a wheat crop to water and nitrogen stress. The
reflectance of parts of the spectrum was modified by the applied treatments. The
normalized difference vegetation index (NDVI) was the most suitable spectral
combination for estimating total leaf area and leaf area index (LAI) across
different treatments. Plant N content could be estimated independently of
treatment through a linear combination of green and red canopy reflectance,
whereas chlorophyll content could not be estimated independently of plant
treatment from radiometric measurements. On the other hand Mani et al. (1991)
studied the spectral reflectance of a sorghum crop as influenced by nutrient,
irrigation and plant canopy. The spectral ratio, (IRJR) was highest between 55 and
M. Menenti
Traditional irrigation. The spatial resolution of current space-borne sensors is
adequate for detailed studies of relatively small irrigation schemes, even in
complex environments such as the Ecuador's Andes region studied by Goulven et
al. (1995) to characterise traditional irrigation systems. Practical methods were
developed on the Urcuqui irrigation system (12 635 ha) using stereo-pairs of
SPOT image data. Crops were identified with an accuracy of ",,70%, and 60% of
infrastructures were correctly mapped. A study of old irrigation systems in
Turkmenistan was presented by Volovik and Kochergin (1993).
Numeric classification algorithms are widely used in studies of land cover with
numeric satellite data. Criteria have been developed to assess the perfonnance of
numeric classification techniques. The performance of several Landsat 5 Thematic
Mapper (TM) image classification methods for crop estimates in an irrigation
district was studied by Barbosa et al. (1996). The agricultural land cover in a
26300 ha. irrigation district was classified utilizing two Landsat 5 TM scenes.
Manual and automatic selection of training areas for the classification of two
single subscenes and a combined multitemporal sub scene result in several
differently classified images. The classifications based on the training sets selected
manually provided the most precise results, with the exception of the single image
spring classification of rice. D'Urso and Menenti (1996 b) presented a general
approach to rank numeric classification procedures on the basis of statistical
indicators of accuracy, reliability and discriminability.
17.3.3 Crop water requirements - Visible and Near Infrared
Reflectance spectra in the 400 to 2500 nm is related to green biomass, Leaf Area
Index, leaf water content and chlorophyll, although this relation is not straightforward and crop-dependent. Bausch (1995) demonstrated that better estimates of
crop coefficients (and therefore of crop water requirements) were obtained for a
maize crop using spectral reflectance instead of traditional methods. Moran et al.
(1989) evaluated various spectral indices in relation with crop water stress. The
Perpendicular Vegetation Index (PVI) decreased with the stress-induced change in
architecture while the near-IR:red ratio (IRJR) remained relatively constant. As a
result, the near-IR:red ratio was more successful in estimating Medicago sativa
biomass than was the PVI. Fernandez et al. (1994) compared the sensitivity of
radiometric characteristics of a wheat crop to water and nitrogen stress. The
reflectance of parts of the spectrum was modified by the applied treatments. The
normalized difference vegetation index (NDVI) was the most suitable spectral
combination for estimating total leaf area and leaf area index (LAI) across
different treatments. Plant N content could be estimated independently of
treatment through a linear combination of green and red canopy reflectance,
whereas chlorophyll content could not be estimated independently of plant
treatment from radiometric measurements. On the other hand Mani et al. (1991)
studied the spectral reflectance of a sorghum crop as influenced by nutrient,
irrigation and plant canopy. The spectral ratio, (IRJR) was highest between 55 and