A novel hyperspectral index for quantifying chlorophyll-a concentration in productive waters

Abstract

Hyperspectral remote sensing has great potential for monitoring chlorophyll-a concentration (Chla) in optically complex waters. However, various hyperspectral indices currently used for retrieving Chla in productive waters exhibit drawbacks due to interference from other water parameters, imperfect atmospheric correction, and the constraints imposed by discrete spectral bands. To address these issues, we proposed a novel spectral index called the Red-Edge reflectance Peak Width Index (REPWI) for productive waters, which is defined as the horizontal wavelength distance from the fixed wavelength of red reflectance valley (678 nm) to the intersection of the right side of the red-edge reflectance peak curve. Firstly, the bio-optical model was used to clarify the theoretical basis and relationship between REPWI and Chla. Then, in situ remote sensing reflectance and Chla data from 57 water bodies worldwide were used to calibrate and validate the REPWI-based Chla retrieval model. The REPWI-based model achieved high accuracy (coefficient of determination (R²) of 0.92, mean relative error of 26.5 %, and root mean square error of 13.7 mg/m³), significantly outperforming other spectral indices. Furthermore, REPWI maintained a strong correlation with Chla (R² = 0.87–0.90) when applied to major on-orbit hyperspectral satellites equivalent spectra. Finally, the REPWI-based Chla retrieval model was applied to ZY1-02D hyperspectral imagery, which obtained high accuracy and demonstrated resistance to imperfect atmospheric correction. In summary, this novel hyperspectral index, REPWI, has demonstrated a solid theoretical foundation and significant advantages in monitoring Chla in productive waters.