| im- 
tility 
king 
land 
mon- 
p ter- 
on of 
hills 
sand- 
, the 
com- 
uired 
pho- 
ed in 
veral 
pe at 
tation 
s and 
ireas, 
)Ogra- 
968). 
on the 
ids of 
ion of 
photomorphic characteristics, especially colour tones. In 
other words, each kind of region has own notable photo- 
morphic features, in particular colours and their combina- 
tion. Region A is characteristic of long and very narrow 
white stripe accompanied with wide brown-red stripes on 
both flanks, sometimes on one flank. Region B appears to 
be dark green broad stripes as well as light brown or light 
yellow patches. Region C is dominanted by redish yellow 
colour together with light green narrow stripes. In region 
D, the major colours are brown-red and green, or brown- 
yellow, brown-red and greyish white. In region E, grey- 
ish white is the dominant colour, along with patched 
brown-yellow, light yellow or brown-red colours. 
By consulting the land cover/use map and other ancillary 
data and through reconnaissance surveys an integrated 
analysis on the relationship between photomorphic charac- 
teristics of each region and its probable land system was 
conducted. As a result, each of these regions was assigned 
to a given land system with some indispensable modifica- 
tions in its boundary. The resultant land system map was 
shown in Figure 2, and the main characteristics of each 
land system were briefly described in Table 1. 
Table 1 The Main Characteristics of Land Systmes 
Land system Characteristics Total area 
symbol (sq. km) 
A River beds in greyish white colour; and river terraces cultivated and in brown-red 161. 5 
colour. with patches of woodland in dark brown colour or grassland in light yel- 
low colour. 
B Loess hillside, cultivated and in light brown colour or grass-covered and in light 443. 9 
yellow colour; and sandstone gullies, mostly bared and in dark green colour. 
C Sandstone hillside, mostly grass-covered and in reddish yellow colour, and part- 390. 4 
ly cultivated and in light brown colour: and sandstone gullies, mostly bared and 
in light green colour, and large gully bottoms in brownish red colour. 
D Sand-mantled loess hillside, forested and in brown-yellow colour, or cultivated 679.5 
and in brownish red colour, or bared and in greyish white colour: and grass-cov- 
ered loess gullies in yellow green colour or cultivated and in brown-red colour. 
E Sand-mantled and nearly level to gently undulating ground, mostly bared and in 171.2 
white colour and partly forested and in reddish yellow colour, or cultivated and 
in brownish red colour; and sand-mantled shallower gullies, mostly cultivated 
and in red colour, or forested and in brown-red colour. 
3.2 Interpretation and Mapping of Land Units 
  
Land units are the broader components of a land system. 
Each land unit formed on a basic topographic unit has its 
corresponding soil and vegetation. In order to delineate 
land units in a land system, a number of sub-region with 
relatively homogeneous photomorphic characteristics were 
distinguished on the TM image. In addition, one sampling 
area was selected in each sub-region and its boundary was 
as accurately as possible depicted on 1:50, 000 scale aerial 
photographs. As a result, land units were interpreted 
through analysis on the relationship between photomor- 
phic characteristics and physiographic features combined 
with the consulting of the thematic maps and field obser- 
vation. 
In the interpretation of land units such information as 
landform readily to interprete was extracted in the first 
and could be taken as the starting for further interpreta- 
179 
tion. Land cover/use types are distinguishable based 
mainly on colour variation on the photographs. For exam- 
ple, cultivated farmland, fallow, pasture, woodland and 
settlement cold be distinguished by colours of red, green- 
ish grey. yellowish green, brownish red. and greyish 
white, respectively. Besides, size, texture and shadow of 
ground features on the photographs also were used, for in- 
stance, to make a differentiation between tree, shrub and 
grass. Shadows are particularly useful for the estimation 
of dissection intensity of gullies. Nevertheless, there are 
still a certain amount of information related to land unit 
which are not readily to interprete directly on the pho- 
tographs. Consequently, the deductive inferences added 
some advantages, for instance, to the differentiation of 
such ground features as irrigated farmland and dry-farm- 
ing land based on irrigation canals or special large wells. 
Through these stages all land units in the sampling area 
were recognized and mapped on the 1:50, 000 scale base 
map.