Long-term aerosol optical depth analysis and prediction over the Antarctic Peninsula

The Earth's atmosphere is a system that significantly influences weather patterns and climate by regulating the radiation balance. Disturbances in the balance of incoming and reflected radiation filtered through the atmosphere constitute an interaction driving the planet's temperature, atmospheric circulation, and climate mechanism. The polar regions play a crucial role in maintaining the global climate balance due to their influence on atmospheric circulation patterns. In this context, monitoring atmospheric variables and their changes over time is critical for understanding and predicting global climate change. This study investigated the long-term variations of aerosol optical depth (AOD) in Antarctica, a region highly sensitive to atmospheric changes. The research utilized Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua and Terra AOD products across three different-sized areas. The temporal variations of AOD over a twenty-year period were examined on a spatial basis to understand the patterns and trends of AOD. Furthermore, a comparison was conducted using data from the Marambio Aerosol Robotic Network (AERONET) station and satellite-based data. The evaluations were carried out separately using data from the Terra and Aqua satellites, and similar increasing trends in AOD values were observed for both datasets from 2002 to 2022. The correlation between the four-month mean AOD of Aqua and AERONET, as well as Terra and AERONET, was calculated as 0.6488 and 0.6190, respectively. A Seasonal Autoregressive Integrated Moving Average (SARIMA) model was applied to satellite-based data to predict the dataset's behavior in the near future. The predictive analysis based on these datasets indicated that the most appropriate models were SARIMA(3,1,2)(3,1,0)₁₂ for Aqua and SARIMA(2,1,1)(0,1,3)₁₂ for Terra. The root mean square error values derived from predictions based on these model configurations were 0.030 for the Aqua dataset and 0.018 for the Terra dataset, indicating the accuracy of the models in forecasting the respective data. The predictions demonstrated strong agreement with the observed data, revealing a sustained increase in AOD values over time.