A trigonometric polynomial is used to assign values at any model time and for all of the grid points. Initial phytoplankton values Sirolimus for January and December are very limited, so a constant value of 0.1 mgC m−3 is defined; but the model is not sensitive to the initial conditions of phytoplankton concentration (in January). Also, the data for the detritus content at the bottom are not available, so the instantaneous sinking of detritus is a more arbitrary model assumption. The initial amount of detritus at the bottom is prescribed as 200 mgC m−2 for the whole Baltic Sea.

The initial values for total inorganic nitrogen are taken from SCOBI 3D-model for January. The initial vertical distributions of nutrient, phytoplankton, zooplankton and detritus pool are known: Phyt(x, y, z, 0)=Phyt0(x, y, z)0≤z≤H,Nutr(x, y, z, 0)=Nutr0(x, y, z)0≤z≤H,Detr(x, y, z, 0)=Detr0(x, y, H)z=H.The

vertical gradients of the phytoplankton and nutrient concentration fluxes are zero at the sea surface (z = 0): FPhyt(x, y, 0, t)≡Kz∂Phyt(x, y, z, t)∂z|z=0−wzPhyt(x, y, 0, t)=0,FNutr(x, y, 0, t)≡Kz∂Nutr(z, t)∂z|z=0=0. The bottom flux condition for phytoplankton and nutrient is given by FPhyt(x, y, H, t)≡−wzPhyt(x, y, H, t),FNutr(x, y, H, t)≡Kz∂Nutr(x, y,z, t)∂z|z=H=gNREMD.This flux Fphyt(H) enters the benthic detritus equation as a source term. The boundary condition provides Ibrutinib manufacturer the mechanism by which the water column is replenished by nutrients derived from benthic remineralization. In order to assess the accuracy of the CEMBSv1 model for determining the parameters of the Baltic ecosystem, we compared the temperatures and chlorophyll a concentrations obtained from the model with those measured in situ and in water samples for five years Lepirudin (2000–2004). For these comparisons

the relevant errors of these simulations were calculated in accordance with the principles of arithmetic and logarithmic statistics: 1. Arithmetic statistics: 2. Logarithmic statistics: a) Relative mean error:〈ε〉〈ε〉 [%] (systematic) 〈ε〉=1N∑iεiwhere εi=xi, mod−xi, exp/xi, expεi=xi, mod−xi, exp/xi, exp e) Mean logarithmic error: g〈ε〉〈ε〉g [%] (systematic) g〈ε〉=10〈L〉−1〈ε〉g=10〈L〉−1where L=log(xi, mod/xi, exp)L=log(xi, mod/xi, exp) b) Standard deviation of ε: σε [%] σε=1N(∑i(εi−〈ε〉)2) f) Standard error factor: χ χ=10σLχ=10σLwhere σL is standard deviation of L c) Absolute mean error: 〈ε′〉〈ε′〉 [%] 〈ε′〉=1N∑iεi′where εi′=xi, mod−xi, exp g) Statistical logarithmic errors: σ–, σ+ [%] σ−=1/χ−1σ+=χ−1 d) Standard deviation of ε′: σε′ [%] σε′=1N(∑i(εi′−〈ε′〉)2) Full-size table Table options View in workspace Download as CSVwhere xi, mod – calculated values, xi, exp – measured values. The following aspects were taken into account in the assessment of the modelled ecosystem parameters: 1.