Allan, M.G, Hicks, B.J., Brabyn, L. (2007). Remote sensing of the Rotorua lakes for water quality. CBER Contract Report No. 51, client report prepared for Environment Bay of Plenty. Hamilton, New Zealand: Centre for Biodiversity and Ecology Research, Department of Biological Sciences, School of Science and Engineering, The University of Waikato.
Permanent Research Commons link: https://hdl.handle.net/10289/3785
The aim of this study was to determine empirical models between Landsat imagery and lake water quality variables (chlorophy11(ch1) a and Secchi depth) to enable water quality variables to be synoptically quantified. These models were then applied to past satellite images to determine temporal patterns in the spatial variation of water quality. Monitoring of lakes to determine temporal patterns in the spatial variation of water quality. Monitoring of lakes using traditional methods is expensive and lakes the ability to effectively monitor the spatial variability of water quality within and between lakes. Remote sensing can provide truly synoptic assessments of water quality, in particular the spatial distribution of phytoplankton. Recent studies monitoring lake water quality using Landsat series platforms have been successful in predicting water quality with a high accuracy. Analysis was carried out on two Landsat 7 Enhanced Thematic Mapper (ETM+) satellite images of the Rotorua lakes and Lake Taupo, for which most in situ observations were taken within two days of image capture. Regression equations were developed between the Band 1/Band 3 rations (B1/B3) from Landsat images from summer (25 Jan 2002) and spring (24 Oct 2002) and water quality variables measured in the lakes by Environment Bay of Plenty. For summer, the regression of in situ ch1 a concentration in µg/1 from ground data against the Band 1/Band 3 ratio (B1/B3) was Ln ch1 a = 14.141 – 5.0568 (B!/B3) (r² = 0.91, N=16, P<0.001). For spring, the regression equation was Ln ch1 a = 24.251-9.2806 (B1/B3) (r² = 0.83, N=13,P<0.001). Ch1 a water quality maps were than produced using these models which were also applied to other images without in situ observations near the time of image capture.