Forecasting environmental factors and zooplankton of Bakreswar reservoir in India using time series model

Time-series models have vast advantages in the study of dynamic systems, especially if the aims are to determine structure and stability of population or finding regime shifts in dynamic characters of ecological systems. These models can also be used with precise goals for extracting specific demographic functions and their impacts. Auto Regressive Integrated Moving-Average or ARIMA models are one of the most general class of time-series fore-casting models. In the present study, eight different environmental factors we2re chosen as the target groups for studying time series variations, that ranged from physico-chemical (e.g. air and water temperature, humidity, etc.) to biological (zooplankton) factors. Most of the ARIMA models were able to capture the trends of variation in the observed data. However, some linear trends were also observed in few of the forecasted series (increasing for some and decreasing for others). To improve on these forecasts, a hybrid ARIMA-ANN method was utilized which successfully increased the accuracy of future predictions showing seasonal variations in the forecasted values.