The bottom mixed layer depth as an indicator of subsurfaceChlorophyll a distribution

Primary production dynamics are strongly associated with vertical density profiles in shelf waters. Variations in the vertical structure of the pycnocline in stratified shelf waters are likely to affect nutrient fluxes and hence the vertical distribution and production rate of phytoplankton. To understand the effects of physical changes on primary production, identifying the linkage between water column density and Chlorophyll a (Chl a ) profiles is essential. Here, the vertical distributions of density features describing three different portions of the pycnocline (the top, centre, and bottom) were compared to the vertical distribution of Chl a to provide auxiliary variables to estimate Chl a in shelf waters. The proximity of density features with deep Chl a maximum (DCM) was tested using the Spearman correlation, linear regression, and a major axis regression over 15 years in a shelf sea region (the northern North Sea) that exhibits stratified water columns. Out of 1237 observations, 78 % reported DCM above the bottom mixed layer depth (BMLD: depth between the bottom of the pycnocline and the mixed layer underneath) with an average distance of 2.74 +/- 5.21 m from each other. BMLD acts as a vertical boundary above which subsurface Chl a maxima are mostly found in shelf seas (depth <= 115 m). Overall, DCMs were correlated with the halfway pycnocline depth (HPD) ( rho S = 0.56) which, combined with BMLD, were better predictors of the locations of DCMs than surface mixed layer indicators and the maximum squared buoyancy frequency. These results suggest a significant contribution of deep mixing processes in defining the vertical distribution of subsurface production in stratified waters and indicate BMLD as a potential indicator of the Chl a spatiotemporal variability in shelf seas. An analytical approach integrating the threshold and the maximum angle method is proposed to extrapolate BMLD, the surface mixed layer, and DCM from in situ vertical samples.