115
Annex D.Innovative methods of mapping and monitoring
saline and alkaline soils
Table D.1. Land use distribution within the Kurdamir District
№
Parameters
Units of
measure
Years
1999
2000
2001
2002
2003
2004
2005
1
The total land area
ha
116190
116190
116190
116190
116190
116190
116190
2
Including agricultural lands
ha
83359
83359
83359
83359
83359
83359
83359
3
Croplands
ha
28997
43401
48074
47946
44491
48362
49833
4
Perennial
plantations
ha
2403
2178
2144
2146
2254
2373
2384
5
Pasture
ha
51959
37780
33141
33267
36614
32624
31142
6
Forest belts
ha
739
739
739
739
739
739
739
7
Fallows
ha
22500
22500
22500
22500
22500
22500
22500
8
Other
ha
9592
9592
9592
9592
9592
9592
9592
The soil cover of the Kurdamir District is dominated by Meadow Greyzems (28 thousand ha) and
Light Meadow Greyzems (23 thousand ha), which have irrigated areas of 15 and 14.7 thousand
ha, respectively. Salt-affected soils are found mostly in the southern part of the district and have a
total area of 3.6 thousand ha.
New methods of soil mapping based on the computer analysis of high-resolution satellite images
were tested at the study site (300 km
2
) located to the east of Kurdamir town. A network of field
test points was set within the study site to ensure that the complete range of soil types and land use
types was represented. The network included a total of 100 points located along 6 transects across
the site with varied spacing that was designed to adequately represent changes in soil type and use.
Field data recorded at each
point included GPS coordinates, land use type, crop species, plough
depth and the presence of surface salt concentrations and salt- tolerant plant species.
Most points
were also photographed.
Soil samples were taken from the surface (0-5 cm) and the 5-10 cm depth at each point, with three
extra samples at a distance of less than 1 m around each point. The contents (%) of salt ions (Cl-
and SO4
2-
) and carbonates (CaCO
3
) and pH were then determined in the samples. At selected test
points, morphological descriptions and photographic documentation of soil profiles 50 cm deep
(the maximal possible pit depth due to high soil density) were also carried out.
All surface horizons studied were salt-affected (although their salt concentrations were below 2%,
i.e., did not meet the salic criterion). Mean salt concentrations determined
at each point varied
from 0.8 to 1.55%, with electrical conductivity (EC) values from 16 to 31 dS/m, which indicated
a strong salinization of soils. The surface horizons were highly calcareous, with calcium carbonate
contents of 10-13%.
The WorldView-2 multispectral satellite image of 18.08.2011 was analysed.
The image underwent
standard radiometric, atmospheric and geometric corrections and ortho- transformation.
Processing of the digital image with the aim of soil mapping was performed using the Erdas
Imagine 9.2 program and included the following stages:
1. Supervised Classification of Images on the basis of visual separation of image sets.
2. Parallelepiped Classification of Images (where brightness values were used for determination
of image class boundaries) with the aim of distinguishing bare and vegetated surfaces.
3. Image Segmentation using the eCognition algorithm to distinguish uniform areas followed
116
Soil salinity manаgement manual | Part II.Tutorial
examples, guidelines and exercises
by visual and statistical comparison of the output image with soil salinity values at 100 test
points. The results of correlation analysis showed that salinity values within the 5-10 cm soil
layer were not connected with spectral characteristics of the surface. Moreover,
it was found
that the output image mainly reflected the changes in soil taxa (types or subtypes) and land
use types.
4. Inclusion of land use type into the eCognition algorithm. Several land use classes were
identified: fields of cotton and wheat, ploughland, abandoned fields, scrub, meadows, bogs,
etc. Histograms for most classes, apart from ploughland, had similar
spectral characteristics
and could not be differentiated.
5. Nearest Neighbor Image Classification by land use type. The identified classes were as follows:
ploughland, fields of cotton and wheat, meadows and pastures, bogs, abandoned lands, forest
and scrub. The area of each class was determined.
6. Nearest Neighbor Image Classification by soil type, with land use type taken into account.
As a result, 17 soil classes were identified. There were unambiguous and ambiguous classes,
with the latter including two or three soil types. The unambiguous classes included Meadow
Greyzems, Greyzemic Meadow soils and Light Meadow Greyzems (classified with the
precision levels of 90%, 79%, 83%, respectively) as well as Wet Meadow soils identified with
the highest precision (94%) and Dark Greyzemic Meadow soils identified least certainly (with
a precision level below 60%).
The soil cover and soil salinity map of the study area. The data obtained were used to show the
distribution of salt-affected soils on the latest soil map of the Kurdamir District, which was created
by combining old and new information sources. This map was designed to represent the main soil
types and subtypes and the degrees of soil salinization at a 1:50000 scale (Fig. D.9).
117
Annex D.Innovative methods of mapping and monitoring saline and alkaline soils
Do'stlaringiz bilan baham: 
