Primary Productivity Algorithm Round Robin of Satellite-based Models for the Arctic Ocean

Integrated net primary productivity (NPP) was estimated from 32 ocean color models (OCMs) and compared to in situ NPP. OCMs were most sensitive to the source of surface chlorophyll-a (CHL), generally performing better with in situ than with satellite-derived CHL. OCMs underestimated the variability of NPP and they over(under)-estimated NPP in low(high)-productivity regions/seasons. OCMs performed better when no subsurface CHL maxima (SCM) were present, but NPP was overestimated, which may be offset by underestimation of NPP due to absence of SCM, or dampened by averaging temporally on a pan-Arctic scale. Summary The Arctic Ocean (AO) poses special challenges to ocean-color based Net Primary Productivity (NPP) models because of the simultaneous occurrence of subsurface chlorophyll-a maxima (SCM), enhanced coastal turbidity, sea-ice presence during the growth season, and the orbit of the satellite per se over polar regions. Historically the scarcity of in situ data to validate simulations has affected our ability to accurately simulate contemporary mean NPP and variability in the AO [Wassmann and Reigstad, 2011]. In order to better understand the strengths and limitations of currently available NPP estimates for the AO, we conducted a Primary Productivity Algorithm Round Robin (PPARR) activity to assess the skill and sensitivity of NPP models in the AO (Fig. 1). This study is based on 32 ocean color models using input fields from a combination of in situ, reanalysis, and satellite data (Fig. 2), i.e., surface chlorophyll-a (CHL), photosynthetically available radiation (PAR), sea surface temperature (SST), and mixed layer depth (MLD) as well as satellite-derived properties: Rrs, QAA a443, GIOP aph, GIOP bbp443, and Zeu_lee