Geostatistical Methods for Filling Gaps in Level-3 Monthly-Mean Aerosol Optical Depth Data from Multi-Angle Imaging SpectroRadiometer

Abstract

The Aerosol Optical Depth (AOD) retrieved from satellite remote sensing measurements such as from MISR and MODIS, both onboard the Terra platform, are widely used for studying regional and global patterns of aerosol loading. Aerosol products from these sensors are also used for analyzing feedbacks and relationship between aerosols and climatic variables including clouds, precipitation, and radiation fluxes. Several statistical techniques leading to the understanding of such relationships, including empirical orthogonal function and temporal trend extraction methods, require spatially complete AOD data records. Inherent to remote sensing of aerosols, cloud cover significantly affects aerosol retrievals and results in missing data across the AOD products. This paper demonstrates widely-used geostatistical techniques, such as Co-Kriging (CK) and Regression Kriging (RK), for spatially-filling missing data in the MISR AOD product for the period 2001-2013. Among the unique characteristics of this data-filling algorithm is that it utilizes additional AOD information obtained from MODIS. The mean accuracy of the predicted MISR AOD using CK method is estimated to be 0.05, globally. The gap-filled MISR AOD data are also compared with 131 ground-based Aerosol Robotic Network (AERONET) stations, located around the world. It is found that Root Mean Squared Error of the gap-filled AOD dataset and the original MISR AOD product with respect to AERONET data are 0.143. The gap-filled AOD dataset can be used in applications where the presence of missing values is undesirable such as for global/regional aerosol variability and trend analysis.

Getting Down to Earth: Are Satellites Reliable for Measuring Air Pollutants That Cause Mortality in Low- and Middle-Income Countries?

“…In addition, they found that the dicted maps of PM 10 concentration were substantially different between kriging and co-kriging, with the co-kriging map capturing more of the expected special variation of PM 10 in the domain. In addition, Singh, Venkatachalam, and Gautam (2017) used co-kriging to fill in missing MISr AOD data using observations from the MODIS instruments. However, there is no example of co-kriging with satellite-based PM 2.5 products in the literature.…”

Empirical Correlation Weighting (ECW) Spatial Interpolation Method for Satellite Aerosol Optical Depth Products by MODIS AOD over Northern China in 2016

“…To ensure the data integrity and the accuracy of the interpolation results, several methods have been proposed. Combing multisource AOD data can provide additional AOD information and increase the availability of valid data [18,23,27], which can then be used to generate the desired products, even seamless aerosol products through spatial interpolation methods. In one approach, AOD data were recovered by synthetically weighing the AOD similarity, spatial proximity, and NDVI similarity based on the mechanism of satellite AOD retrieval [28].…”

Full-coverage 250 m monthly aerosol optical depth dataset (2000–2019) amended with environmental covariates by an ensemble machine learning model over arid and semi-arid areas, NW China

“…Considerable related work has concentrated on multisource AOD dataset fusion or AOD gap-filling methods using different models. The initial and most extensively applied method is interpolation, but the AODs obtained in this way show high spatiotemporal variability; thus, this method is not suitable for application to anticipate missing AOD data (Singh et al, 2017). Another widely used method involves merging mul-tiple AOD products; this method can improve the data quality but often fails to completely eliminate missing pixel values, even bringing about offsetting consequences (Bilal et al, 2017;Ali and Assiri, 2019;.…”