Akademik Veri Yönetim Sistemi
ECOLOGICAL MODELLING, cilt.144, ss.295-314, 2001 (SCI-Expanded)
Using the Black Sea ecosystem as an example, the phytoplankton seasonal cycle is simulated by several coupled physical-ecosystem models with different vertical resolutions, but the same biological setting. First, a high resolution multi-level model having a vertical grid spacing of about 3 m is shown to reproduce the observed annual phytoplankton structure reasonably well. This simulation is then compared with its multi-layer alternatives to investigate feasibility of using a relatively simpler model, and to look for its optimum vertical configuration. The simplest model involving a three layer structure provides only general features of the multi-level model simulation. It is able to reproduce the autumn and spring blooms taking place in the mixed layer, but it is not equally successful for simulating the summer production within the intermediate layer below the seasonal thermocline. It is found that this deficiency of the model is related to its poor nutrient recycling capability. Resolving the intermediate layer in the form of two sub-layers (i.e. increasing the model resolution to the four layer case) is shown to improve the efficiency of nutrient recycling and lead to a much better agreement of the results with those of its multi-level model. Further resolution introduced into the mixed layer, however, does not improve the performance of the layer models further. The key conclusion from our analysis is that, despite simplicity of its vertical configuration, the four layer model emerges as a practical alternative tool to its more complex, and computationally more demanding multi-level counterpart, particularly for three dimensional applications. Moreover, the Kraus-Turner type bulk surface-layer dynamics implemented in the multi-layer models are shown to be very efficient in simulating observed mixed layer characteristics as well as those deduced from the more sophisticated Mellor-Yamada level 2.5 turbulence closure parameterization of the multi-level model. (C) 2001 Elsevier Science B.V. All rights reserved.