Multi-source hierarchical data fusion for high-resolution AOD mapping in a forest fire event
Xiaoli Wei, Kaixu Bai, Ni‐Bin Chang, Wei Gao
Abstract
Satellite and ground‐based remote sensing images, as well as reanalysis data, are widely used to measure and/or model aerosol properties of Earth's atmosphere. However, none of these data sources are perfect: satellite data suffer from various sources of uncertainties and data gaps; ground observations have limited spatial coverage; and reanalysis data can’t provide high resolution information. In this study, we synergize these three data sources to develop a hierarchical data fusion algorithm based on the philosophy of Modified Quantile-Quantile Adjustment-Bayesian Maximum Entropy (MQQA-BME). Such efforts lead to improved data coverage, prediction accuracy, and spatiotemporal resolution simultaneously. Practical implementation of MQQA-BME was assessed by mapping the aerosol optical depth (AOD) of a forest fire event in California in November 2018. The proposed hierarchical data fusion scheme successfully synergizes the multi-source AOD data of MERRA2, GOES-16, and MAIAC, and the fused products are further calibrated using AERONET data. The estimated coefficient of determination (R2) and the root‐mean‐square error (RMSE) of the fused data set of MEERA2_GOES_MAIAC are 0.481 and 0.084, respectively. After calibrating with AERONET AOD data, the R2 and RMSE were improved to 0.694 and 0.072, respectively. The MQQA-BME algorithm has paved a new way to dynamically map AOD at high spatiotemporal resolution.