To evaluate the effect of sampling frequency on the global monthly mean aerosol optical thickness (AOT), we use 6 years of geographical coordinates of Moderate Resolution Imaging Spectroradiometer (MODIS) L2 aerosol data, daily global aerosol fields generated by the Goddard Institute for Space Studies General Circulation Model and the chemical transport models Global Ozone Chemistry Aerosol Radiation and Transport, Spectral Radiation-transport Model for Aerosol Species and Transport Model 5, at a spatial resolution between 1.125°× 1.125° and 2°× 3°: the analysis is restricted to 60°S–60°N geographical latitude. We found that, in general, the MODIS coverage causes an underestimate of the global mean AOT over the ocean. The long-term mean absolute monthly difference between all and dark target (DT) pixels was 0.01–0.02 over the ocean and 0.03–0.09 over the land, depending on the model dataset. Negative DT biases peak during boreal summers, reaching 0.07–0.12 (30–45% of the global long-term mean AOT). Addition of the Deep Blue pixels tempers the seasonal dependence of the DT biases and reduces the mean AOT difference over land by 0.01–0.02. These results provide a quantitative measure of the effect the pixel exclusion due to cloud contamination, ocean sun-glint and land type has on the MODIS estimates of the global monthly mean AOT. We also simulate global monthly mean AOT estimates from measurements provided by pixel-wide along-track instruments such as the Aerosol Polarimetry Sensor and the Cloud–Aerosol LiDAR with Orthogonal Polarization. We estimate the probable range of the global AOT standard error for an along-track sensor to be 0.0005–0.0015 (ocean) and 0.0029–0.01 (land) or 0.5–1.2% and 1.1–4% of the corresponding global means. These estimates represent errors due to sampling only and do not include potential retrieval errors. They are smaller than or comparable to the published estimate of 0.01 as being a climatologically significant change in the global mean AOT, suggesting that sampling density is unlikely to limit the use of such instruments for climate applications at least on a global, monthly scale.
I Geogdzhayev, B Cairns, MI Mishchenko, K Tsigaridis, T van Noije. Model-based estimation of sampling-caused uncertainty in aerosol remote sensing for climate research applications
Journal: Q.J.R. Meteorol. Soc., Volume: 140, Year: 2013, First page: 2353, Last page: 2363, doi: 10.1002/qj.2305