A merged aerosol dataset based on MODIS and MISR Aerosol Optical Depth products

Abstract

Aerosol Optical Depth (AOD) products available from MODIS and MISR observations are widely used for aerosol characterization, and global/environmental change studies. These products are based on different retrieval-algorithms, resolutions, sampling, and cloud-screening schemes, which have led to global/regional biases. Thus a merged product is desirable which bridges this gap by utilizing strengths from each of the sensors. In view of this, we have developed a “merged” AOD product based on MODIS and MISR AOD datasets, using Bayesian principles which takes error distributions from ground-based AOD measurements (from AERONET). Our methodology and resulting dataset are especially relevant in the scenario of combining multi-sensor retrievals for satellite-based climate data records; particularly for long-term studies involving AOD. Specifically for MISR AOD product, we also developed a methodology to produce a gap-filled dataset, using geostatistical methods (e.g. Kriging), taking advantage of available MODIS data. Merged and spatially-complete AOD datasets are inter-compared with other satellite products and with AERONET data at three stations- Kanpur, Jaipur and Gandhi College, in the Indo-Gangetic Plains. The RMSE of merged AOD (0.08-0.09) is lower than MISR (0.11-0.20) and MODIS (0.15-0.27). It is found that merged AOD has higher correlation with AERONET data (r within 0.92-0.95), compared to MISR (0.74-0.86) and MODIS (0.69-0.84) data. In terms of Expected Error, the accuracy of valid merged AOD is found to be superior as percent of merged AOD within error envelope are larger (71-92%), compared to MISR (43-61%) and MODIS (50-70%).