Retrieval of chlorophyll content from multitemporal IRS-1C/LISS-III data
For the year 1997 a multitemporal set of IRS-1C/LISS-III data including four dates (May 4th, June 2nd, September 1st, and September 25th) was made available. Due to the few broad multispectral bands, only one parameter, i.g. chlorophyll, could be determined. For that task, the green peak was correlated to chlorophyll content. The comparison of different methods showed that linear spectral unmixing gave the best interpolation results. For each scene the spectrum of the lake with the lowest chlorophyll concentration (representing clear water: lake Wumm in most cases) and the lake with highest chlorophyll concentration (representing chlorophyll laden water, lake Bramin in all cases) were taken as endmembers. The resulting band containing the relative spectral similarity to lake Bramin was multiplied by the chlorophyll content of lake Bramin as retrieved from the field campaigns at the same or similar date. The mean standard error could be estimated from the scene recorded on September 1st with additional field sampling of 11 lakes. With 13 µg/l it was in the same order of magnitude as for the hyperspectral data.