New band locations for remote sensors provide more accurate information about oceans

From nearly 400 hyperspectral measurements of water color, the primary spectral bands for ocean-color remote sensing are determined via first- and second-order derivatives.
11 December 2007
ZhongPing Lee, Kendall Carder, Robert Arnone and Ming-Xia He

For remote sensing of optical and biogeochemical properties in the ocean or coastal waters, several operational satellite sensors are currently in orbit, including SeaWiFS, MODIS, and MERIS. These sensors collect spectral information coming out of the water at a few spectral bands in the visible domain (seven for SeaWiFS, eight for MODIS, and 12 for MERIS). The major properties derived from this information include phytoplankton biomass, colored dissolved organic matter, suspended sediments, and bottom properties.

Recently, Lee and Carder1 demonstrated that to adequately derive such properties from observation of water color in both oceanic and coastal environments, a sensor should have ∼15 bands in the 400–800nm range. As a follow-up to that study, we have determined the spectral positioning of the 15 bands (see Figure 1) from nearly 400 hyperspectral (∼350–900nm with a ∼3nm resolution) measurements of remote-sensing reflectance (a measure of water color). The measurements were taken in both coastal and oceanic waters covering optically deep and optically shallow waters.


Figure 1. Spectral locations proposed by this study (gold color), overlaid with the bands of current operational sensors, along with samples (black curves) of spectral remote-sensing reflectance (Rrs).

We determined the band locations based on the frequency of zero values of first- and second-order derivatives at each wavelength. Such zero values indicate extrema (a local maximum or minimum) of the reflectance spectrum or inflections of the spectral curvature. Placing the bands of a sensor at these wavelengths maximizes the potential to capture (and then restore) the spectral curve, and thus maximizes the potential to accurately derive the properties of aquatic environments. Interestingly, the band locations in this study not only validate the band selections of the operational sensors, but they also nearly encompass the band suggestions made in previous studies2–4 aimed at different objectives or different water environments.


ZhongPing Lee, Robert Arnone
Naval Research Laboratory
Stennis Space Center, MS
Kendall Carder
University of South Florida
St. Petersburg
Ming-Xia He
Ocean University of China
Qingdao, China

Recent News
PREMIUM CONTENT
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research