Modeling Framework to Study the Influence of Environmental Variables for Forecasting the Quarterly Landing of Total Fish
- PDF / 585,305 Bytes
- 4 Pages / 595.276 x 790.866 pts Page_size
- 76 Downloads / 157 Views
SHORT COMMUNICATION
Modeling Framework to Study the Influence of Environmental Variables for Forecasting the Quarterly Landing of Total Fish Catch and Catch of Small Major Pelagic Fish of North-West Maharashtra Coast of India Vinod K. Yadav1,2
•
Shrinivas Jahageerdar2 • J. Adinarayana1
Received: 11 October 2018 / Revised: 24 September 2019 / Accepted: 21 January 2020 The National Academy of Sciences, India 2020
Abstract Forecasting fish landings is a critical element tool for fisheries managers and policymakers to short-term quantitative recommendations for fisheries management. In this study, the forecasting of a quarterly landing of total fish catch and the catch of major pelagic fish species (Indian Mackerel and Bombay duck) was done by nonlinear autoregressive with exogenous inputs (NARX), an artificial neural network model. The quarterly landings data of total fish catch and the catch of major pelagic fish along with quarterly average data on the mean value of environmental variables were used for building the model and forecasting. The developed NARX model was validated with the actual fish catch on holdout data with prediction error 2.45–11.42%. Further, the developed NARX model was used to forecast fish catch for the next 20 quarters (5 years) and was compared and found good agreement with the actual catch reported by Central Marine Fisheries Research Institute, Kochi, annual report(Year- 2014, 2015 and 2016). The developed NARX model in the present case study is of the first time to forecast the fish catch landing using exogenous input in the Maharashtra region. Keywords NARX Forecasting Fish catch landing Maharashtra
& Vinod K. Yadav [email protected] 1
Centre of Studies in Resource Engineering (CSRE), Indian Institute of Technology, Powai, Bombay 76, India
2
Central Institute of Fisheries Education (CIFE), Panch Marg, Off Yari Road, Versova, Andheri(W), Mumbai 61, India
Forecasting is used to analyze the past and current behavior and to predict the future fish landing which may be used in decision-making and planning. Autoregressive integrated moving average (ARIMA) model is the most widely used model for forecasting time series data [1, 2]. One of the main drawbacks of this model is the presumption of linearity. Sometimes, the time series often contain nonlinear components; under such condition, the ARIMA models are not adequate in modeling and forecasting. The artificial neural network (ANN) is the most widely used machine learning techniques to model and forecast the time series data which contain nonlinear component. To model the series which contains nonlinear patterns, the artificial intelligence technique- nonlinear autoregressive (NAR) artificial neural network (ANN), commonly employed to forecast samples framed in a one-dimensional time series. Several well-known methods and models such as autoregressive integrated moving average (ARIMA), vector autoregression (VAR), neural network, wavelet were used by many researchers in the past for short-term catch forecasting [3], but
Data Loading...
