The versatility of the neural network (NN) technique allows it to be successfully applied in many fields of science and to a great\nvariety of problems. For each problem or class of problems, a generic NN technique (e.g., multilayer perceptron (MLP)) usually\nrequires some adjustments, which often are crucial for the development of a successful application. In this paper, we introduce a\nNN application that demonstrates the importance of such adjustments;moreover, in this case, the adjustments applied to a generic\nNN technique may be successfully used in many other NN applications. We introduce a NN technique, linking chlorophyll â??aâ?\n(chl-a) variabilityâ??primarily driven by biological processesâ??with the physical processes of the upper ocean using a NN-based\nempirical biological model for chl-a. In this study, satellite-derived surface parameter fields, sea-surface temperature (SST) and\nsea-surface height (SSH), as well as gridded salinity and temperature profiles from 0 to 75m depth are employed as signatures\nof upper-ocean dynamics. Chlorophyll-a fields from NOAAâ??s operational Visible Imaging Infrared Radiometer Suite (VIIRS) are\nused, aswell asModerateResolution Imaging Spectroradiometer (MODIS) andSea-ViewingWide Field-of-ViewSensor (SeaWiFS)\nchl-a concentrations. Different methods of optimizing the NN technique are investigated. Results are assessed using the rootmean-\nsquare error (RMSE) metric and cross-correlations between observed ocean color (OC) fields and NN output. To reduce the\nimpact of noise in the data and to obtain a stable computation of the NN Jacobian, an ensemble of NN with different weights is\nconstructed. This study demonstrates that the NN technique provides an accurate, computationally cheapmethod to generate long\n(up to 10 years) time series of consistent chl-a concentration that are in good agreementwith chl-a data observed by different satellite\nsensors during the relevant period. The presented NN demonstrates a very good ability to generalize in terms of both space and\ntime. Consequently, the NN-based empirical biological model for chl-a can be used in oceanic models, coupled climate prediction\nsystems, and data assimilation systems to dynamically consider biological processes in the upper ocean.
Loading....