Modeling inland water quality using Landsat data

The water quality parameters turbidity and algal pigment concentration of freshwater lakes have been modeled and predicted using Landsat multispectral scanner data as multiple linear predictors. Satellite data for an area in South East Australia from seven occasions during 1978 and 1979 were used along with concurrent ground-based measurements from sampling sites on three lakes covering a wide range of water quality regimes. Date-independent models for turbidity and algal pigment were obtained using the satellite multispectral data and the water quality data from up to 21 sampling sites on one lake on six occasions. The sun elevation at the time of satellite overpass was included in the models to account for differences between dates, and the time of sample collection was included to compensate for diurnal variations in pigment fluorescence. These models were used to successfully predict these water quality parameters for this lake on a new occasion and for the two other lakes on three occasions.