| Abstract: | Primary production (PP) models of the Kara Sea are developed based on data collected on fall expeditions (September–October 1993, 2007, and 2011) and their precision assessment utilizes the dataset collected in September 2013. The algorithms for different model types (depth-integrated and depth-resolved) are compared. The depth-resolved model performs slightly better than the depth-integrated one (the rootmean- square-difference (RMSD) are 0.29 and 0.31, respectively). These algorithms utilize the daily assimilation number (DAN) and photosynthetic efficiency (ψ) as the model coefficients, and surface chlorophyll a (chl a) and photosynthetically active radiation (PAR) as input variables. These algorithms perform better than the models that use chl a alone. Our results suggest that an increase in the performance of the Kara Sea PP models depends on the input of the photophysiological characteristics of phytoplankton (DAN and ψ) and PAR. To a lesser extent, this concerns the advantages of the depth-resolved model over the depth-integrated one. The constructed region-specific Kara Sea PP models combined with satellite-derived chl a and PAR can be used to estimate annual values and long-term variation of PP in hydrologically and hydrochemically similar waters of the Arctic Ocean. |
