<![CDATA[High-frequency wind speed data collected via environmental monitoring systems often contain significant stochastic noise that can obscure underlying patterns and degrade the reliability of statistical models. A two-stage modeling framework (integrating a Kalman Filter (KF) for signal purification and Autoregressive Integrated Moving Average (ARIMA) for predictive modeling) was developed and applied to five-minute interval wind speed data in Pontianak, West Kalimantan. The dataset, comprising 3,742 observations recorded from December 11 to 24, 2024, was utilized to evaluate the effectiveness of the KF in enhancing model fitting. The model quality was further assessed using Individual Moving Range (IMR) control charts to monitor residual stability and detect localized anomalies. Results demonstrate that the KF-ARIMA approach significantly improves performance, reducing the Root Mean Square Error (RMSE) from 1.123 m/s to 0.145 m/s, representing an 87.1\% improvement in precision compared to the standalone ARIMA model. The I-MR charts confirmed that the KF-ARIMA residuals remained consistently within the $3\sigma$ control limits, effectively identifying transient variations that standard diagnostic tests might overlook. This integrated framework proves that combining state-space filtering with traditional time-series models provides a robust approach for characterizing high-frequency meteorological data in equatorial regions.]]>
Copyrights © 2026
