A geostatistical approach for producing daily Level-3 MODIS aerosol optical depth analyses

The daily Level-3 MODIS (dL3M) aerosol optical depth product is a global daily spatial aggregation of the Level-2 MODIS aerosol optical depth (10-km spatial resolution) into a regular grid with a resolution of 1° × 1°. Aerosol optical depth is a seminal parameter for surface solar radiation assessment, in particular, for those applications involving direct irradiance. However, the dL3M AOD is prone to data gaps originated mostly by the unfeasibility of retrieving reliable estimates under cloudy conditions. In addition, its usability is also constrained by regional biases owing to some other reasons. In this work we propose a methodology for bias reduction and data-gaps removal of the dL3M AOD dataset. The result is a database of daily regularly-gridded AOD suitable for use in surface solar radiation applications and large-scale and long-term studies involving AOD without requiring a previous costly data assimilation process involving numerical weather prediction models. The method consists of an empirical approach to bias reduction, data-gaps removal by kriging interpolation and, finally, where reliable ground observations are available, an optimal interpolation procedure. The method was tested in the North American region, where it was able to reduce the initial mean error from 0.067 to 0.001, the root mean square error from 0.130 to 0.057, and increase the squared correlation coefficient from 23% to 58%, as compared against ground measurements.