Monitoring water quality characteristics over Asprokremmos and Kourris dams using remote sensing

Abstract

Remote sensing techniques were utilised in order to investigate water quality parameters of the two largest dams located in Cyprus, Asprokremmos and Kourris. Recent studies showed that dams were classified as oligotrophic to mesotrophic states but only concerned classification for a short period of time. In this study, a ten year approach from 2006 until present was conducted for both dams in order to examine their overall behaviour in terms of Chlorophyll-a (Chl-a) and Turbidity concentrations values and thereafter to their correlation with Trophic State Index (TSI). In total, ten digital images from Landsat 5 Thematic Mapper (TM) and five from Landsat 8 Operational Land Imager (OLI) were acquired from United States Geological Survey (USGS) concerning the abovementioned time of period. For the pre-processing of all digital images, radiometric and atmospheric correction was performed to obtain reflectance values. It was observed that dams of Asprokremmos and Kourris indicated rational water reflectance values for all the selected bands, Red-Green-Blue (RGB), and where it was possible, validation was done utilising in-situ spectral signatures. Otherwise, findings were compared with other studies reflectance’s outputs. Furthermore, four algorithms were applied to retrieve water quality parameters as the post-processed part of this study. The turbidity and TSI algorithms which were originally developed for Asprokremmos dam and Landsat 5 TM, was also tested for Kourris dam and Landsat 8 OLI. For Chl-a concentrations, two proposed algorithms were found in literature concerning the two satellites bands’ characteristics and other study areas were also tested. From the final results, it was concluded that the turbidity algorithm performed well at Kourris dam. For Asprokremmos dam, the estimated turbidity mean values data showed similar findings with empirical in-situ data collected from past campaigns. Chl-a algorithms also presented rational concentrations at both dams but it was assumed that due to no calibration, final results may be questioned, especially for Kourris dam. Linear regression models were created weighting differently for each parameter and dam and a correlation with the estimated TSI values was achieved scoring up to 0.59 for Asprokremmos and 0.37 for Kouris. Low values of correlation were justified by the fact that the time periods among the acquisition dates been examined where not always even and thereafter influenced the overall variation of the parameters’ mean values.