| Summary: | 碩士 === 國立臺灣師範大學 === 地理學系 === 94 === The purpose of this thesis was to investigate the relationships that existed between landscape patterns and stream water quality, in the hopes of providing information to help land management at a watershed-level. Fourteen water quality parameters have been monitored on five sampling sites within fourteen months in upstream watershed of the Shui-Li creek. First, different statistical techniques were applied to interpret the spatial variation of, and to reduce the information of stream water quality. In addition, correlation analysis was performed to identify the relationships between land use compositions, landscape metrics and stream water quality variables. Moreover, this study addressed the issue of whether land use near streams influence steam water quality more than land use over the entire watershed.
The result of data analysis indicates that there are higher SO42- concentrations at the sampling sites of WC-4 and WC-5, which maybe result from the difference of geological conditions. Factor analysis shows that stream water quality data consists of five components, but there is no one that can explain the great amount of variation in the fourteen parameters. Cluster analysis suggests that monitoring sites and frequencies can be consolidated. NO3- concentrations was positively correlated with the land use of betel palm plantation within 25m buffer. The stream discharge will rise as a landscape tends to fragment. The variation of stream discharge would increase at a watershed where the proportion of betel palm plantation is large. However, forested land has the agent of regulating stream discharge. As patch area of forested land and grassland are larger and more compact, Cl- and NO3- concentrations become lower, respectively.
In short, the findings of this study can be drawn: one is that study of stream water quality can use multivariate analysis to identify the characteristics of water quality and optimize the design of monitoring programs within Taiwan watersheds in the future. The other, landscape metrics are useful to monitor the change of in-stream water chemistry.
|
|---|
