Geography Reference
In-Depth Information
Table 5.8 The remotely sensed data classification based statistical results for the 21,000 ha
irrigation project for 1975, 1987, 2005 and 2007
The 21,000 ha-project_all classes
1975
1987
2005
2007
Cultivated areas
23,834
25,806
34,792
Trees ? shrubs
648
872
58
Herbaceous (winter crops)
12,561
22,377
23,366
Alfalfa
275
459
198
Wheat
3,669
11,693
10,560
Barley
5,305
4,150
1,211
Sugar beet
3,454
6,075
2,359
Other crops
62
200
9,038
Fallow
10,927
2,796
11,970
Herbaceous (summer crops)
50,499
24,272
13,303
Cotton
2,950
8,412
2,284
Corn
2,099
463
650
Other crops
8,143
127
10,369
Fallow
10,063
15,270
22,277
Water
0
0
0
0
Uncultivated areas
46,324
22,494
20,501
11,489
sieve filter. This caused a smoothing of the class boundaries. A small pixels-cluster
of an individual class was added to the surrounding area of a larger class, and the
boundaries of the LULC-classes were generalized and clearly identified. The sieve
filter was used in addition to the majority filter to clean the classification result of
further small pixel-clusters that were not eliminated from the majority filter.
Clusters with less than 10 pixels were removed by merging them with their largest
neighbor.
5.12 Automated Change Detection Mapping
5.12.1 Pre-Classification Approach
This approach was essential in mapping the increasing changes in the agricultural
irrigated areas in the ERB. Data from the LANDSAT-program, MSS and TM
spanning the period between 1975 and 2007 were chosen from a similar time of
the year in order to allow each LULC-class of interest a similar spectral response
and similar illumination conditions. The MSS-data set of six images (Fig.
5.60
)
were pre-processed (see
Sect. 5.2
) for radiometric normalization using iMAD. The
master-scene was p185r035. It was impossible to correct the atmospheric effects
because it was difficult to obtain weather parameters for such relatively old dates.
However, it was possible to carry out radiometry and atmosphere corrections for
the TM-data set of six images. The master-scene for this was p172r035. Each data
Search WWH ::
Custom Search