Geography Reference
In-Depth Information
Remote sensing techniques were approved and applied on the remotely sensed
data (see Chap. 4.1 ) of 1975, 1987, 2005 and 2007 for the four major emphases
(LULC-classification, LULC-change detection, irrigation mapping, and irrigated
agriculture classification). There were no known determined adopted techniques
that could be directly applied for the emphases. Therefore, it was necessary to set
suitable methods that would be compatible with the used data, the thesis questions
and the privacy of the study area environment. Most of the applied methods in this
work were already recognized, some were modified and some were combined.
It is true that a lot of remotely sensed data and some processing programs are
becoming more and more accessible to researchers at little or no cost, but remotely
sensed data processing and interpretation techniques are still time consuming and
not suitable for all regions of the Earth at the same level of accuracy. It has been
shown in this study that the geometric, atmospheric and radiometric correction
processes (see Chap. 5.2 ) are not always necessary for each image, each sensor
data and each date. Geometric correction processes, especially for the geometric
registration, were not difficult and were achieved at very high accuracies. How-
ever, atmospheric correction was impossible for the relatively old data (MSS-
1975), where the weather parameters were difficult to obtain. Radiometric cor-
rection was applicable for all data, but this did not mean that it produced suitable
results for the whole dataset. Therefore, when neither the raw data or the enhanced
data after applying the atmospheric and/or the radiometric corrections gave good
results (especially for use in mosaicing), then the data were processed and clas-
sified separately, i.e., each image alone. ATCOR-2 was used for atmospheric
correction, while iMAD was used for radiometric correction. Both applications
were relatively easy to use and required no additional external information. These
programs were found to give better results than those methods that are time
consuming and needed more external data such as MFF and 6S (Chavez 1996 ).
Where the results of the classifications were cartographic products, all spatial
data were standardized, and were transformed and geometrically corrected to a
general reference system: a UTM-projection of Zone 37 N with the international
general ellipsoid/spheroid WGS84 and datum WGS84.
Geometric correction, geo-referencing and geometric registration formed the
basis for mosaicing more than one image (see Chap. 5.2.5 ) for fusion of different
remotely sensed data, i.e., the ASTER-data were fused with LANDSAT-ETM+-
data (see Chap. 5.2.4 ) for detection of changes (see Chap. 5.12 ) .
Precise mosaicing was very important for further remotely sensed data pro-
cessing and interpretation (e.g., classification, change detection, etc.). The general
algorithms of imagery mosaicing were not always able to produce a one mosaic-
image with a consistent appearance in which the values of the histograms of each
image were combined together in one mosaic-image. This gave an unsuitable
presentation of the various LULC-features on the mosaic-image. In these situa-
tions, the MAD-technique was applied to satisfy a radiometric consistent mosaic.
This was a comfortable relative radiometric calibration technique that built a data
Search WWH ::




Custom Search