Remote Sensing of Urban Lake Water Quality: A Preliminary Result from Spectral Angle Based Approach

Abstract

This study aims to develop a quick method based on spectral angle (SA) to evaluate overall water quality and spatial variation in urban lakes using in-situ water quality parameters of lakes or reservoirs and synchronous SPOT 5 remote sensing imagery, referring to the spectrum of a clear montanic Jiulongtan Reservoir in satellite image. The regression models between SA and water quality parameters were built for analysis, including chlorophyll a (Chl-a), chemical oxygen demand (COD), total phosphorous (TP), total nitrogen (TN) and integrated trophic state index (TSI). The results show that the grades of lake water quality in Guangzhou could be ordered by SA values from most desirable to least desirable as Jiulongtan Reservoir, Luhu Lake, Liwan Lake, Liuhua Lake, and Dongshan Lake. Further, the results also show that the SA of urban lakes correlates potentially with the parameters of water quality (Chl-a, R2 = 0.929569, p < 0.01; COD, R2 = 0.9767916, p < 0.01; TN, R2 = 0.58767495, p < 0.05; TP, R2 = 0.8705, p < 0.05) or TSI (R2 = 0.9066, p < 0.001) in spite of limited data samples collected in the study. The SA classification results by SPOT 5 multi-spectral images roughly reflect the grade difference of water quality as a whole and their spatial variations, i.e. consistent with concurrent result of lake water sampling analysis. The validation shows this approach can be helpfully used to quickly monitor the water quality status of lakes or reservoirs for broad region, to effectively identify the sampling locations for water sample taking and water quality analysis, and provide information for the management of urban lakes (by SPOT 5 or higher resolution satellite image) or large-middle sized reservoir (by Landsat ETM+).