1. پیشبینی روند تغییرات صید ماهی تون زرد باله در آبهای جنوبی کشور)Thunnus albacares Bonnaterre, 1788()NN(و شبکة عصبی)ARIMA(براساس مدلهای آریما
- Author
-
سید احمدرضا هاشمی and مسطوره دوستدار
- Abstract
The aim of this study is to develop different models of aquatic forecasting and try to predict yellowfin tuna catch in the southern waters of the country with minimum possible errors. The average catch (Yi ± S. D) and logarithm of catch (LogYi ± S. D) for the years 1997 to 2021 were 35,378 ± 13,744 tons (95% confidence interval 21,634 - 49,129 tons) and 4.51 ± 0.18 tons (95% CI The confidence interval was 4.33-4.69 tons), respectively. According to the Mann-Kendall test, the average catch has increased significantly during the mentioned period (over the last two decades) (Z = 3.80, P < 0.05). Different ARIMA combined prediction models (ARIMA, (p, d, q)) were tested based on the AIC index, and the ARIMA model (1, 0, 0) had the best fit with the change trend of yellowfin tuna in the southern waters of the country (AIC=- 24). The predict of yellowfin tuna catch results in the neural network (NN) models was show that feed forward neural networks (FFNN) have better performance than other models and with less error (MAE=0.02 and RMSE=0.03). Also, according to the results of ARIMA time series and neural network models, it can be concluded that feed forward neural networks simulate catch this species fish with higher accuracy than time series models. It seems, forecasting the trend of aquatic catch can be an important tool for fisheries managers and planners for better and sustainable management of aquatic resources and should be given more attention. [ABSTRACT FROM AUTHOR]
- Published
- 2023