Exploring the use of Landsat imagery in monitoring chlorophyll-a and lake surface temperature in Lake Victoria

Abstract

In Uganda, Lake Victoria is a livelihood of about 40 million people and a home to aquatic life. It is utilized for the different activities like tourism, industry, transport and fisheries. However, it contains substances derived from the environment due to both natural and anthropogenic activities and these cause eutrophic challenges which is a negative impact on the water quality. Traditionally, in situ platforms have been used and they provide accurate measurements of water quality parameters with minimum disturbances from the atmosphere. In addition, they offer high sampling intervals but it is not always possible to make immediate observations over a large region due to high costs, labor-intensiveness and environmental /weather conditions. This research has investigated the use of Landsat8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS) sensors in monitoring water quality in Lake Victoria. 39 stations were sampled for Chlororphyll-a and lake surface temperature parameters of both in situ and Landsat8 OLI data on 12th &13th March 2015 and 15th March 2015 respectively. OLI digital image corrected atmospherically by Quick Atmospheric Correction (QUAC) obtained the first five bands for Chl-a and the 10 and 11 thermal bands for temperature. Different models were used to estimate the parameters in the Lake from the satellite after which band rationing was used to get the best bands to model Chl-a. The best model had R2 = 0.696 and using the band ratios, a model was generated yielding a correlation coefficient R2 of 0.806. The developed model was validated producing an R2 of 0.601 with a RMSE of 1.8 µg/l. Band 10 among the thermal bands showed a better correlation of R2 of 0.646 with a RMSE of 2.84 0C with the in situ temperature than band 11. A standard model for temperature using band 10 yielded R2 of 0.683 and on validation, the developed model had a coefficient of determination of 0.617 and a RMSE of 0.003 0C. Conclusively, the research gives a good correlation between the satellite data and in situ data making the use of Landsat imagery possible in monitoring the water quality of lake Victoria. More studies should be done to investigate the applicability and consistency of the developed models for estimating the Chl-a concentration and lake surface temperature even on other waters.