摘要
In China, the increase in exogenous-source pollutants from rivers is one of the most important causes of lake eutrophication. The application of remote sensing technology to water quality monitoring of rivers connected to these lakes has special significance for lake management at regional scales. Many research studies have estimated water clarity using Landsat imagery. However, most of this work focused on lakes or reservoirs, for which abundant water-only pixels (i.e., pure pixels of water, PPW) were available. Few of these studies have addressed rivers, especially rivers with an average width less than 100 m. In our study, we sought to determine whether water clarity in the rivers connected to Taihu Lake could be estimated using Landsat imagery. We obtained 18 Enhanced Thematic Mapper Plus (ETM ) images from 2009 for 13 rivers ranging from an average of 37.3 to 173.6 m wide. Three field campaigns conducted in May 2009, September 2009, and January 2010 were used to obtain field measurements of Secchi disk depth (SDD). Our results suggested that the widely used model, a(TM1/TM3) b(TM1) c, was suitable for the estimation of SDD for Taihu Lake. The brightness of the panchromatic band of ETM showed significant correlations with TM1, TM3 and TM1/TM3 (p < 0.001). As a result, SDD in the lake could also be estimated using the Landsat panchromatic band. The multispectral image of ETM did not provide adequate PPW for estimation of water clarity in rivers. However, PPW derived from the panchromatic image captured about 93 % of the variation in SDD, on average, for the every worst-case scenario in the 13 rivers. Using the PPW in rivers, a significant correlation was found between the brightness of the panchromatic image and SDD (R2 = 0.64, p < 0.001). Our results demonstrate that the panchromatic image of Landsat, but not the multispectral image, can be used to estimate water clarity in rivers with an average width greater than 40 m in the Taihu basin.