Prediction of the solar radiation using RBF neural networks and ground-to-sky images

In this study, Artificial Neural Networks are applied to multistep long term solar radiation prediction. The networks are trained as one-step-ahead predictors and iterated over time to obtain multi-step longer term predictions. Auto-regressive and Auto-regressive with exogenous inputs solar radiationmodels are compared, considering cloudiness indices as inputs in the latter case. These indices are obtained through pixel classification of ground-to-sky images. The input-output structure of the neural network models is selected using evolutionary computation methods.

Population structure of the black seabream Spondyliosoma cantharus along the south-west Portuguese coast inferred from otolith chemistry

The chemistry of black seabream Spondyliosoma cantharus otoliths from three main fishery grounds (Olhao, Sagres and Sesimbra) located along c. 400 km of the Portuguese south and west coasts was examined. Element:Ca ratios were determined in whole otoliths and otolith cores of young adult specimens of 23 years of age. Using the data from whole otoliths, it was possible to discriminate among S. cantharus from the three fishing grounds with an average accuracy of 91%. Differences among fishing grounds were significant for all element:Ca ratios, and otoliths from Sagres had significantly higher levels of all ratios compared to the other fishing grounds. In contrast, the chemical composition of the otolith core, representative of the larval stage, showed limited variation among the fishing grounds, with an average discrimination accuracy of only 44%, although the Mg:Ca ratio of the otolith cores was also significantly higher for the Sagres samples. The data suggest that larval stages experienced a homogenous environment consistent with an offshore oceanic spawning. Juveniles appeared to display local residency on the inshore fishing grounds, areas probably characterized by greater environmental heterogeneity. Spondyliosoma cantharus population structure is consistent with distinct local population units that share a spawning ground providing recruits to different coastal fishery areas.

Structure and temporal variations of fish assemblages of the Castro Marim salt marsh, southern Portugal

We studied the ichthyofauna of the Castro Marim salt marsh based on monthly sampling surveys at five sites between September 2000 and August 2001. Sampling took place at night during rising neap tides using a 40-m long beach seine. We sampled a total of 7955 fish specimens (37 995.7 g), comprising 34 species and 17 families. The occurrence of most species was occasional, with Pomatoschistus microps (51.9%) and Atherina spp. (10.3%) being the most abundant species, accounting for 62.2% of the total fish captured. Biomass was dominated by the marine species Liza ramado (15.9%), Mullus surmuletus (13.5%), and Liza aurata (13.4%). Temperature and salinity showed a seasonal pattern, with minimums during the winter months and maximums during the summer months. In contrast, river flow peaked in winter and was lowest during summer. This pattern in river flow appears to be correlated with variations in the fish assemblages, which present two distinct compositions during the two periods. A few species characterise the winter fish assemblage, with dominance by residents and the presence of freshwater species, while the summer assemblage is characterised by the presence of many marine visitors that use the salt marsh in their first months/years of life. Temporal variations in total abundance and biomass reflect fluctuations in the dominant species. Resident species presented the highest abundance values, while marine adventitious species and marine species that use the salt marsh as a nursery ground contributed most to community species richness. Castro Marim salt marsh constitutes an important ecosystem for fishes, providing habitat for many species, especially juveniles, which find conditions within the salt marsh suitable for their development. (c) 2006 Elsevier Ltd. All rights reserved.