Improving the empirical line method applied to hyperspectral inland water images by combining reference targets and in situ water measurements

ABSTRACT

Empirical line methods are frequently used to correct images from remote sensing. This method is performed in two steps: the first stage finds the calibration equation representing the data interval and the second step transforms the image data into the quantity established from the equation. Several works have been successfully applied empirical lines over land areas, but it is still a great challenge to correct images from waterbodies. Remote sensing of aquatic environments captures only a small amount of energy because the water absorbs much of it. The response signal of the water is smaller than the signal from other land surface targets. This work presents a new approach to calibrate empirical lines combining reference panels with a water point. For this purpose, we evaluated several combinations of targets using both linear and exponential fit. The best matching was provided with an exponential fit using a single grey reference panel combined with a water point resulting in a coefficient of determination about 0.87, a root-mean-squared error of 0.002 sr⁻¹ and a mean absolute percentage error of 18%. This approach presented suitable results to derive reflectance data from the raw digital numbers from images.

Acknowledgments

The authors would like to thank the Graduate Program in Cartographic Sciences (PPGCC) of the School of Sciences and Technology (UNESP), campus Presidente Prudente, for allowing the development of this research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The dataset, python algorithms and additional graph information that support the findings of this study are openly available in IEEE DataPort^TM at <http://dx.doi.org/10.21227/fbks-rj40> (do Carmo et al. 2019).

Additional information

Funding

The National Council for Scientific and Technological Development [CNPq, 400881/2013-6, 472131/2012-5 and 141909/2015-3]; and the Coordination for the Improvement of Higher Education Personnel (CAPES) for financial assistance dedicated to the project. Special thanks to the São Paulo Research Foundation [FAPESP, Process N 2012/19821-1; and N 2013/090457] by the financial support to the hyperspectral camera [2013/50426-4].