Water quality monitoring using Landsat Themate Mapper data with empirical algorithms in Chagan Lake, China

Lake Chagan represents a complex situation of major optical constituents and emergent spectral signals for remote sensing analysis of water quality in the Songnen Plain. As such it provides a good test of the combined radiometric correction methods developed for optical remote sensing data to monitor water quality. Landsat thematic mapper (TM) data and in situ water samples collected concurrently with satellite overpass were used for the analysis, in which four important water quality parameters are considered: chlorophyll-a, turbidity, total dissolved organic matter, and total phosphorus in surface water. Both empirical regressions and neural networks were established to analyze the relationship between the concentrations of these four water parameters and the satellite radiance signals. It is found that the neural network model performed at better accuracy than empirical regressions with TM visible and near-infrared bands as spectral variables. The relative root mean square error (RMSE) for the neural network was < 10%, while the RMSE for the regressions was less than 25% in general. Future work is needed on establishing the dynamic characteristic of Chagan Lake water quality with TM or other optical remote sensing data. The algorithms developed in this study need to be further tested and refined with multidate imagery data.