were polluted by local pyrites mining, which could interfere
with the measurement of some water quality variables
(Wei et al. 1989).
Results
Watershed characteristics
The catchment area of the sampling sites was ranged from
4.97 to 1,963 km2
, with a mean value of 346 km2
(Table 1). The mean slope and elevation were ranged from
14.3 to 28.4, 1,153 to 2,424 m, respectively.
Land use in the study region was dominated by forest,
with the value ranged from 36.4 to 92.0% (mean 64.0%)
among sampling sites. Shrub ? grass was the next most
prevalent cover, ranging from 0.86 to 43.0% (mean 18%).
The mean values of agriculture and urban were 14.3 and
0.10%, respectively. The GMS River and JCH River
watersheds had the highest forest coverage, whereas the GF
River, GL River and the downstream of XX River watershed
have larger agriculture area due to their relatively flat
topography and lower elevation (Fig. 2). The agriculture in
the upland of the study region watershed is dominated by
tea, corn and tobacco, whereas the middle and lower
regions are typified by orchards and paddy. Urban areas
were mainly distributed in the middle and lower region of
the watershed near the river bank.
Spatial pattern of water quality
The SOM algorithm clustered 81 sampling sites into five
groups according to the similarity in water quality (Fig. 3a).
Table 1 Basic statistic information of watershed characteristics,
which was used to describe influence of land use (anthropogenic) and
topography (natural) on river water quality
Parameters Range Mean ± CV
Land use
Forest (%) 36.4–92.0 64.0 ± 22
Shrub and grass (%) 0.86–43.0 19.8 ± 56
Permenant vegetation (%) 75.4–95.5 83.7 ± 6.15
Agriculture (%) 0.00–24.6 14.3 ± 47
Urban (%) 0.00–0.51 0.10 ± 144
Topography
Catchment area(km2
) 4.97–1963 346 ± 134
Mean catchment slope (o
) 14.3–28.4 23.2 ± 16
Mean elevation (m) 1153–2424 1541 ± 20
Fig. 2 The spatial distribution of forest, shrub ? grass, agric