Estimating Secchi depth by simulated HSI of HJ-1A from in situ hyperspectral data: A case study in Sishili Bay, China

Abstract

We study the use of simulated Hyper-Spectral Imaging sensor (HSI) on Environmental Satellite 1A (HJ-1A) satellite remote sensing data for estimating Secchi depth of coastal waters. Field data such as Secchi depth of the Sishili Bay in Yantai's coastal waters were collected, meanwhile, hyperspectral remote sensing data were measured with Ocean Optics USB4000 spectrometer during two cruises carried out on 22nd and 23rd June 2009. The coastal water-leaving reflectance of HSI was simulated from in situ hyperspectral remote sensing spectrum with 0.19 nm spectral resolution. According to the spectral simulation, characteristics bands of HSI were identified for Secchi depth estimation model. On the basis of correlation analysis between Secchi depth and simulation spectra, a single band model for retrieving Secchi depth was established. Moreover, the relationship between Secchi depth and all the band combinations (band addition, band subtraction, and band ratio) were analyzed, and then a linear regression model of Secchi depth using simulated HSI was developed, which performed better than other models in Sishili Bay, with a mean relative error (RE) of 5.2% and relative mean square error (RMSE) of 0.28 m. The result indicates that the band ratio model of Rrs(508.42)/Rrs(513.56) of HSI could be used to estimate Secchi depth in coastal waters.