Retrieval of spectral slope of chromophoric dissolved organic matter (S275–295) in Laptev Sea

Abstract

Study region: Laptev Sea Study focus: The absorption slopes of colored dissolved organic matter (CDOM) in the spectral range of 275–295 nm (S275–295) is a reliable parameter of the source and transformation of CDOM and dissolved organic carbon (DOC) in the estuarine environment. Studying S275–295 in coastal areas can provide important information about CDOM and DOC. This study proposed a new multilayer backpropagation neural network (MBPNN) based on the relationship between S275–295 and aCDOM(443) (absorption of CDOM at 443 nm) as a customization function to invert S275–295. The model accurately estimated S275–295 with a MAPE and RMSE of 6.12 % and 0.0012 nm−1, respectively, and revealed the spatiotemporal distribution of S275–295 in the Laptev Sea between 2002 and 2022 (July to September). The effects of influencing factors on temporal and spatial variations of S275–295 were analyzed. New hydrological insights for the region: The MBPNN model with a customization function has good performance to invert S275–295 in Laptev Sea. The S275–295 in the Laptev Sea ranged from 0.014 to 0.022 nm−1. S275–295 increased gradually from coastal waters to the open sea. Inter-annual S275–295 fluctuated, but no significant trend was observed during the study period (2002–2022). S275–295 was negatively correlated with river discharge (r=-0.41), permafrost thaw depth (r=-0.42), ice extent (r=-0.55) and Normalized Difference Vegetation Index (r=-0.46) but positively correlated with salinity (r=0.86) and wind speed (r=0.46).

Keywords

• Colored dissolved organic matter
• Spectral slope
• Machine learning
• Oceanic carbon cycle