th
ed
CS
es &
pular
categories
ories
southwest
es, valleys,
k and,
| side of
and 4700
evel.
rtheast and
mountain
n between
; above the
wet valley
ream head,
slope foot
XX) meters
v loot of
un, valley,
vegetation
/0.
1s below
is. below
is below
t of high
X0. meters
ind on the
Y SIS
transparent
se maps in
computers,
AAPINFO,
alysis of
AUTOCAD was used in complete these tasking and some necessary
programs were developed by using C and ADS.
e GIS construction
A self-developed GIS package, AUTOPLY, was used to complete
topology building, intersection, mosacking, geographical correction,
and intemal attribute database construction. Resulted data were
transferred from DXF to MIF and MID, a ASCII format accepted by
MAPINFO. Polygons of land/vegetation and base map, including
linear targets, were managed in separated strata respectively. Figure
| shows the resulted map of grassland classification.
e GIS analysis
Linking attached attribute databases of thematic maps, calculate and
summarize statistically the area information and related data of every
category and/or territory. Information query and layouts were built
up in the final system.
6. RESULTS AND DISCUSSIONS
Table 4 shows the final area summary of each category in Jianshe
scene.
e Table 4 Statistical summary of grassland classification
Category Polygons Area (sq. m.)
] 2673 4329828287. 34
11 625 1740668957. 54
12 303 276631698. 79
13 157 86325967. 21
14 1513 1891832255
141 278 376249739. 32
142 546 5231763567. 40
143 689 991818948. 44
15 19 334369408. 64
à 64 30814958. 70
3 7 1187575. 05
4 157 76819602. 61
41 149 62856681. 44
42 5 962482. 46
43 3 13060438. 71
5 146 536367299. 95
The total 3047 ; 4975077723. 65
6.1. Season for acquiring image
The substitute scene in. December, instead. of image in supposed
optimal late autumn or early winter, did not give good representation
of ground targets with simple tones, which is ditlicult to interpret. A
summer scene, however, shows much better results. Answer to tlus
is quite obvious. On the winter image, black beach does not stand
out like that in the field and is confused with other soil or dead grass,
while black beach can be easily identified on the summer scene.
There should be significant difference on phytomass between
degenerated grassland and nonnal grassland. Summer scene,
therefore, should be good for this task
It is suggested from this practice that a scene in summer, late spring
or early autumn be good for visual interpretation of grassland with
emphasis on degeneration. A scene in winter, more easy to acquire
without cloud cover but some snow, can be used as reference to
cancel cover of cloud, haze, or snow, if budget is allowable. Digital
image should be better on the enhancement of image to represent
vegetation coverage differentiation and its result should be more
reliable with the combination of computer-assistant
enhancement/classification and visual interpretation.
6.2. Interpretability of Landsat TM image
[t is quite suitable, from this practice, to map grassland (up to sub-
category) on the scale of 1:100,000, using Landsat TM image by
means of visual interpretation. Some special processing or
techniques should be involved if larger scale map is made fron TM
scene. One method is enlarging of image by digital interpolation and
stretch with the combination of more field work for examining and
compensating (Dai & Fu 1991). Another is to enlarge TM image by
mosaicking it to panchromatic SPOT image, witch is much cheaper
than multi-spectral data, which can make thematic map on the scale
of 1:25,000, even to 1:10,000. Types of the grassland classification
system of China basmg on a differentiation method of
phytocommunity and habitat, however, would be difficult to be
identified reliably on a Landsat TM scene.
ACKNOWLEDGEMENTS
I would like express my special thanks to the project office of
Qinghai Livestock Development Project China/EU at Xing,
Qinghai, China in funding the project Staff from both Qinghai
Bureau of Animal Husbandry and the CSARS attribute much in data
collecting, map-drawing, digitizing, and editing. All of their work
are indispensable to the success of the project
REFERENCES
Dai, Changda 1981. A Comprehensive Analysis Approach for Soil
Interpretation of Remotely Sensed Data. In Chinese.
PROCEEDINGS FOR SYMPOSIUM IV OF CHINESE
ASSOCIATION OF PEDOLOGY.
Dai, Changda and Deyong Fu 1991. Feasibility and Key Techniques
for Thematic Mapping on the Scale of 1:50000 Using Landsat TM
Image. In Chinese. REMOTE SENSING TECHNIQUE AND
APPLICATION. 6(2):13-16,56.
Dai, Changda and Liping Lei 1989. The Information Characteristics
of Thematic Mapper Data and the Optimal Band Combination. In
Chinese with English abstract. REMOTE SENSING OF
ENVIRONMENT CHINA. 4(4):282-292.
Department of Animal Husbandry and Veterinary of Ministry of
Agriculture of PRC and National General Station of Animal
Husbandry and Veterinarian 1996. GRASSLAND RESOURCES
OF CHINA. In Chinese and preface in English. Scientific and
Technological Press of China, Beijing. p608.
He, Yi 1985. Measures for Vegetation Recuperation on Deteriorated
Alpine Meadow of Guoluo Prefecture, Qinghaï Province.
AGRICULTURAL AND ANIMAL RESOURCES AND
PLANNING. (11):9-12.
Huang, Buoning and Li, Xilai 1993. Determination of Phytomass on
Deteriorate Grassland- "Heitutan" in Alpine meadow. In Chinese.
JOURNAL OF QINGHAI ANIMAL HUSBANDRY AND
VETERINARY MEDICINE COLLEGE. 23(4):10-12.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 303
