Statistical modeling for long-term meteorological forecasting: a case study in Van Lake Basin

Abstract

Predicting environmental variables for a sustainable environment is vital for effective resource management and regional development, especially in sensitive regions such as the Lake Van basin in eastern T & uuml;rkiye. This study focuses on long-term annual forecasts of important meteorological variables such as mean annual atmospheric pressure, wind speed and surface evaporation in the Van Lake basin. Long-term forecasts made using R-based statistical models such as AUTO.ARIMA, TBATS, EST, NAIVE, THETAF and HOLT-WINTERS are evaluated using mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE). Here, it has been observed that the AUTO.ARIMA model consistently stands out with better performance than its counterparts in the field of time series analysis when predicting the variables mentioned above. Such scientific studies, which are of great importance especially for the regional structure, add valuable information to the literature by determining a superior prediction model for meteorological events in the specific geographical context of the Lake Van basin. The results of the study have far-reaching implications for further improving predictive modeling techniques, improving the reliability of long-term meteorological forecasts, and decision-making in climate-related research and applications.