A novel remote sensing index for brine shrimp (Artemia) slick detection in salt lakes

Abstract

Artemia (brine shrimp) are a widely distributed zooplankton in inland salt lakes and artificial ponds throughout the tropical, subtropical, and temperate zones. Notably, they are of great importance to fishery production and salt lake ecosystem health. Recently, Artemia resources have been threatened by habitat deterioration and overfishing, which mandate adequate monitoring; however, advanced methods for detecting Artemia from remote sensing images have rarely been discussed. Accordingly, this study explored the possibility of detecting the adult Artemia (Artemia) slicks using multispectral optical sensors by analyzing their unique aggregation patterns and consistent optical properties. After measuring Artemia spectra in the field, reflectance troughs were observed at ∼431 nm and ∼ 560 nm, which were deemed likely related to the absorption of β-carotene and heme, respectively, resulting in their reddish color. These characteristics could also be used to differentiate Artemia from other confounding features, including clean water and chlorophyll-a (Chl-a)- or suspended particulate matter (SPM)- dominated water. Based on these spectral signatures, a brine shrimp index (BSI) was developed using the difference between the reflectance at NIR band and a linear baseline between green and SWIR1 bands of Landsat-8 Operational Land Imager (OLI) imagery. Further, a BSI-based Artemia slick pixel extraction method was proposed. Using observation geometry information in conjunction with the BSI value of entire lakes, the BSI thresholds for the Artemia-containing pixels were automatically determined, and > 80% overall accuracy was achieved. The index performance was affected by the presence of residual cloud shadows, high water turbidity, and mixed pixels with low Artemia density. Moreover, BSI values were found to be relatively insensitive to atmospheric signals. It was shown here that the BSI detection limit was >0.53% Artemia subpixel coverage under clean water conditions; whereas a higher coverage was required to separate Artemia from Chl-a- or SPM- dominated water. The mean absolute percentage difference of BSI values calculated from Rayleigh-corrected and top-of-atmosphere (TOA) reflectance was <4% compared with surface reflectance derived BSI, and the relative difference in coverage area was < ±10%. Lastly, time series analyses of Lake Aibi in northwest China and the Great Salt Lake in North America revealed that the annual areal coverage estimation of Artemia slicks based on BSI was a positive indicator of Artemia harvesting, which can facilitate the Artemia resource management.