Environmental Engineering Reference
In-Depth Information
TABLE 2.1
List of the Landsat images used.
RRN
c
Purpose
d
Date
Landsat
Sensor
Scene
Nominal
Sun
Sun
RMSE
No.
Type
Location
IFOV (m)
Elevation
Azimuth
(Control
(degree)
(degree)
Point No.)
4-13-73
1
MSS
north Atlanta
57 × 79
54.16
129.37
0.58 (13)
Yes
1
4-13-73
1
MSS
south Atlanta
57 × 79
54.83
127.39
6-11-79
3
MSS
north Atlanta
57 × 79
61.65
105.65
0.46 (13)
Yes
1
6-11-79
3
MSS
south Atlanta
57 × 79
61.74
102.77
6-29-87
5
TM
center-shifted
a
28
.
5
×
28
.
5
61.84
103.71
0.22 (13)
No
1
5-14-88
5
MSS
center-shifted
57 × 79
61.61
115.51
0.51 (12)
ref.
4
7-31-93
5
TM
center-shifted
28
.
5
×
28
.
5
57.00
110.00
0.40 (15)
No
1
7-10-97
5
TM
center-shifted
28
.
5 × 28
.
5
61.00
106.00
ref.
No
2and4
1-02-98
5
TM
center-shifted
28
.
5 × 28
.
5
27.00
150.00
0.27 (14)
No
3
28
.
5 × 28
.
5
b
9-09-99
7
ETM+
north Atlanta
53.8
140.10
0.52 (15)
No
1and4
28
.
5 × 28
.
5
b
9-09-99
7
ETM+
south Atlanta
54.8
138.40
0.44 (13)
5-19-07
5
TM
north Atlanta
30 × 30
65.65
121.91
0.42 (14)
No
1
5-19-07
5
TM
south Atlanta
30 × 30
66.17
118.68
a
The panchromatic band has a nominal IFOV (instantaneous field of view) of 15 m.
b
The center of the north scene has been shifted by 50%. The scene size is approximately
185 × 185
km
2
.
c
RRN
Relative radiometric normalization.
d
Code: 1
=
land use/cover mapping; 2
reference data for geometric correction; 3
reference data for ground truth in the Winter season; and 4
land
=
=
=
=
use/cover mapping accuracy assessment.
Based on the Landsat images, the ancillary data, and the
knowledge gained during the field survey, we carefully designed
a mixed Anderson Level I/II land use/cover classification scheme
(Anderson
et al
., 1976), with the following six major classes:
remote sensing and relevant geospatial techniques, including
data preprocessing, image classification, spatial reclassification,
and thematic accuracy assessment.
2.3.3.1
Image preprocessing
Both geometric rectification and radiometric normalization were
conducted. The geometric correction strategy used is actually an
image-to-image rectification. The reference image was the 1997
Landsat TM image (see Table 2.1) that has been orthorectified by
EOSAT to UTM projection (Zone 16), NAD83 horizontal datum,
and GRS80 ellipsoid. All other scenes were rectified to the TM
image by using a first-degree polynomial transformation, given
the relatively even terrain relief in the study site. The number
of control points and RMSEs (root mean squared errors) are
summarized in Table 2.1.
The two MSS images used for image classification were
acquired by Landsat-1 and Landsat-3, and are very different from
each other in contrast although they were processed identically.
To help restore a common radiometric response among them,
the relative radiometric normalization method developed by Hall
et al
. (1991) was applied to the 1973 and 1979 MSS images by
using the 1988 MSS image as the reference (Yang and Lo, 2000).
Note that the RRN method developed by Hall
et al
. (1991) is
based on the use of radiometric control sets that should have little
or no variation through time. These control sets were extracted
by using the two non-vegetated extremes of the Kauth - Thomas
(KT) greenness-brightness scattergram which was constructed
using the first two bands of a Tasseled Cap transformation of
the raw image. Radiometric normalization was not attempted to
●
high-density urban use: mostly large commercial and indus-
trial buildings, large transportation facilities, and high-density
residential areas in the city cores;
●
low-density urban use: mostly single/multiple family houses,
apartment complexes, yards, local roads, and small open
spaces;
●
cultivated/exposed land: mainly non-impervious areas with
sparse vegetation, such as clear-cuts, quarries, barren rock or
sand along river/stream beaches;
●
cropland/grassland: crop fields and pasture as well as cultured
grasses (such as golf courses, lawns, city parks);
forest land: deciduous, coniferous, and mixed forest land; and
●
water: streams, rivers, lakes, and reservoirs.
●
2.3.3
Image processing of remotely
sensed data
The design of the image processing procedures to be used
in this study was based on a thorough understanding of the
research objective, image characteristics, landscape complex-
ity, and the status of technological development. The image
processing
procedures
identified
here
involved
a
variety
of
Search WWH ::
Custom Search