Geography Reference
In-Depth Information
features out of classification process (i.e., insignificant classes in irrigated agri-
cultural projects such as bare areas), or postponing classification to the next stage
if spectrally possible. Here, the significance of the accredited methodological
approach in building up the used classification system emerged, in the reduction of
the number of classes in each stage of classification and the securing of some
reduction in the spectral mixture.
One of the positives of the hierarchal principle is to reduce the effect of the
geographical location and the natural and climatic properties that affect the
spectral behaviour of the studied Earth surface features, specifically, if the study
area is within a wide geographic distribution, peppered with large diversion in
natural and climatic characteristics. For example, making a mask of the distribu-
tion of irrigated plantations gives a natural harmonious area, since all the culti-
vations here are irrigated and the majority of soils have close colours and close
content of humidity, etc. The greater the study area with a geographic and spatial
distribution featuring the same or similar natural and climatic characteristics, the
more likely homogeneity will be achieved in the spectral response of the Earth
surface features contained in the study area. This trait does not exist in the geo-
graphical and spatial distribution of the Euphrates River Basin, since, for example,
the spectral behaviour of bare areas will be in the dozens.
One of the more significant positives of the masking process was the reduction
in the problem of spatial correlation which produces classification errors, as well
as separating borders areas and/or the mutual areas between two classes or more
(the negative impact increases wherever the spatial resolution decreases and the
spectral variation rises). This was most effective when the areas of the class (other
crops) were over-classified. Taking into consideration that the mask layer resulted
in application of the NDVI has meant a few of the agricultural areas were
neglected.
Making automated classification on sensory data after applying the mask
(through use of the option: apply mask in ENVI-program) and integrating the mask
layer with/or on the satellite scene with spectral bands (layer stacked), will
decrease the separability between the spectral signatures created from the training
samples. Consequently, automated classification was conducted directly on sen-
sory data through selecting the option of using mask in the ENVI-program.
The classification of water was more effective when using the remotely sensed
data acquired during the summer season (e.g., August).
The reason that the corn was over-classified in some cases, was probably to do
with the mixture between the lands that were cultivated previously with wheat
during the winter season. This occurred when the wheat residues were not burnt or
tilled, and the fields remained covered in yellowish dry residues (straw). In
addition, some plants grew naturally after the harvesting of the wheat.
TERRA-ASTER-May-2005 data fused with LANDSAT-ETM+-May-2005
data. There was no negative or positive impact for using thermal spectral and
panchromatic bands on the spectral discrimination and the separation between
various crops, trees and other surface features in the study area. Therefore, they
were dispensed in the automated classification, especially, the panchromatic
Search WWH ::
Custom Search