Spatiotemporal variation in summer net community production in the Amundsen Sea Polynya: A self-organizing map analysis approach

Abstract

We estimated weekly summer net community production (NCP) in the Amundsen Sea Polynya for 2010-2011 and 2011-2012 using the self-organizing map analysis technique, which is a type of artificial neural network useful for discovering underlying structure in datasets. The net community production estimates derived with four variables (sea surface temperature, mixed layer depth, chlorophyll-a, and photosynthetically available radiation) robustly reproduced the observed net community production in the Amundsen Sea Polynya. The mean net community productions were estimated as 0.42 +/- 0.09 and 0.39 +/- 0.07 gC m(-2) d(-1) in 2010-2011 and 2011-2012, respectively. The maximum weekly net community production of 1.29 gC m(-2) d(-1) in 2011-2012 was greater than in 2010-2011 by 0.32 gC m(-2) d(-1). The net community production in 2010-2011, derived using the self-organizing map, showed weekly variation similar to the trend of satellite-derived production of the Eppley-Vertically Generalized Production Model. However, in 2011-2012, it exhibited different temporal variation both in peak timing and in magnitude of the bloom. This implies the existence of complex processes not readily resolved by the four variables used in our self-organizing map analysis. Therefore, further observations during different blooming stages are required to improve self-organizing map-derived net community production estimates.