Results, Analysis and Discussion - Historical Land Use/Land Cover Classification Using Remote Sensing - page 163

There was a mixing between cultivated and uncultivated areas, particularly, in the
inactive volcano cones area, which is found to the east of the city of Arraqqa,
where dark colour soils are interrelated with fallow lands. Also, there were mixing
and misclassification concerning water and cultivated areas, particularly in river
areas with narrow width and shallow depth. It was possible to integrate the volcano
area into the uncultivated areas by manually drawing the borders of the area in the
form of vector-shapes. There was also significant mixing between fallow and bare
areas, meaning the separation and classification between artificial surfaces and
bare areas was impossible.
Each scene was classified separately after sub-setting based on the borders of
study area. After that, classification results scenes were collected in one mosaic-
scene (Fig. 6.2 ). Thus, the classification results were better rather than the above
mentioned situation, as it was possible to obtain the five needed general classes.
These results were improved for several reasons: technically, returning to the used
remotely sensed data itself, where, whatever the quality of the present algorithms
to correct satellite scenes that contain differences in their radiometric character-
istics, they were not accurate 100 % of the time. Hence, using every scene sep-
arately ensured the spectral behaviour towards all Earth surfaces features (and the
probable variations within each feature alone). The natural reasons for this
included: The large size and extension of the study area and its relation with the
geographical location that controlled the natural climatic characteristics (espe-
cially the precipitation-factor), led to variation in Earth surface features and
consequently in their spectral behaviour on satellite images. This is turn provided
Fig. 6.2 The spatial distributions classification of the five major LULC-categories in the ERB
for the data of LANDSAT-MSS acquired in June-1975