Development of bio-optical algorithms to estimate chlorophyll in the Great Salt Lake and New England lakes using in situ hyperspectral measurements

Chlorophyll is widely used to evaluate lake water quality, effectively integrating the chemical, physical and biological state of a lake. Assessment of chlorophyll conditions in lakes can be greatly enhanced by the use of remote sensing, allowing information to be gathered at spatial and temporal scales not possible with traditional limnological sampling methods. In order for remote sensing methods to provide accurate estimates of chlorophyll concentration, algorithms need to be developed with high-quality spectral data paired with water quality measurements and optimized for regional lake differences. In this study, in situ hyperspectral optical measurements were used to develop algorithms to estimate chlorophyll for the Great Salt Lake and New England lakes. The spectral data were used to mimic bands utilized by the MODIS, MERIS, and SeaWiFS sensors, as well as for a theoretical hyperspectral sensor with 3-nm wide bands, providing the capability to evaluate algorithm performance in all of these sensors. In addition to the traditional bands used in these algorithms, alternate band combinations were examined for both ocean color chlorophyll (OC) and maximum chlorophyll index (MCI) algorithms. A simulated 709 nm band was created for MODIS using the 754 nm band, providing a method for testing MODIS with algorithms relying on the key 705 nm to 715 nm wavelength range. In New England lakes, the most effective algorithm for hyperspectral bands (RMS = 0.206, in log decades) and MERIS (RMS = 0.218) was a version of MCI. For MODIS and SeaWiFS, the most effective algorithm used an OC approach with 489 nm as the blue band, yielding an RMS of 0.242 and 0.231, respectively. In the Great Salt Lake, the most effective algorithms for hyperspectral bands and MERIS were based on a single ratio of 709 nm / 675 nm, providing an RMS of 0.236 and 0.249, respectively. For MODIS and SeaWiFS, the most effective algorithm was the OC method using 489 nm as the blue band, which resulted in an RMS of 0.246 and 0.255, respectively