Improved red-edge chlorophyll-a detection for Sentinel 2

Abstract

Chlorophyll-a (chl-a) concentration is an indicator of algal biomass. The Sentinel 2 platform offers greatly improved spatial resolution over other satellite platforms designed for water based chl-a retrievals and includes a “red-edge” band at 704 nm not present on the Landsat 8 operational land imager. This study provides validation of an improved version of a well known semi-analytical chl-a retrieval algorithm. The algorithm is provided with several free image processing utilities and the improved approach can be implemented with minimal technology skills. The improved performance is the result of replacing a fixed chl-a specific absorption coefficient (a*) with a variable model. This method was applied to three Sentinel 2 images taken over the Lake Erie western basin correlating with an in-situ dataset of 24 samples where chl-a ranged from 1.89 mg m−3 to 70.20 mg m−3. The variable a* model produced chl-a retrievals with normalised root mean squared error of prediction (NRMSEP) = 7.5%, bias = -0.47 mg m−3, coefficient of determination (R2) = 0.91 and Nash-Sutcliffe efficiency (NSE) = 0.90). This represented a 23% reduction in NRMSEP, an 85% reduction in bias and an increase in NSE of 7% over the default algorithm using a fixed a* value. Creation of chl-a retrieval algorithms that consider the variability in a* should result in algorithms that perform better against a wide range of chl-a concentrations and are less likely to require local recalibration. Obtaining accurate chl-a retrievals from a satellite platform with the spatial resolution of Sentinel 2 will allow satellite monitoring of many more inland waters than previously possible.

Keywords

• Remote sensing
• Chlorophyll-a
• Sentinel 2
• MSI
• Red-edge