Time series analysis and stochastic modeling of aerosol optical depth distribution using satellite- and ground-based observations

Authors: Xueke Li*, University of Connecticut, Chuanrong Zhang, University of Connecticut, Weidong Li, University of Connecticut
Topics: Remote Sensing, Environmental Science
Keywords: MODIS C6, Aerosol optical depth, time series, ARIMA
Session Type: Paper
Day: 4/11/2018
Start / End Time: 8:00 AM / 9:40 AM
Room: Astor Ballroom I, Astor, 2nd Floor
Presentation File: No File Uploaded

Long-term trend analysis and modeling of aerosol optical depth (AOD) distribution is of paramount importance to study radiative forcing, climate change, and human health. This study is focused on analyzing and modeling the trends and variations of AOD at six stations spreading across United States and China during 2003 to 2015, using satellite-retrieved Moderate Resolution Imaging Spectrometer (MODIS) Collection 6 retrievals and ground measurements derived from Aerosol Robotic NETwork (AERONET). An autoregressive integrated moving average (ARIMA) model is employed to simulate and predict AOD values. The R2, adjusted R2, Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Bayesian Information Criterion (BIC) are used as indices to select the best fitted model. Results show that there is a persistent decreasing trend in AOD for both MODIS data and AERONET data over three stations. Monthly and seasonal AOD variations reveal consistent aerosol patterns at stations along mid-latitudes. Regional differences caused by impacts of climatology and land cover types are observed for the selected stations. Statistical validation of time series models indicates that the non-seasonal ARIMA model performs better for AERONET AOD data than for MODIS AOD data at most stations, suggesting the method works better for data with higher quality. On the contrary, the seasonal ARIMA model reproduces the seasonal variations of MODIS AOD data much more accurately. Overall, the reasonably predicted results indicate the applicability and feasibility of the stochastic ARIMA modeling technique for forecasting future and missing AOD values.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login