National Aeronautics and Space Administration

The MODIS MAIAC products are currently considered to be at validation stage 1

The Collection 6 MAIAC land surface reflectance suite (MCD19A1) provides spectral bi-directional reflectance factor (BRF), or surface reflectance (SR), daily at 500m and 1km, and 8-day spectral BRDF and albedo for MODIS bands 1-8 (MCD19A3) at 1km.

MAIAC surface reflectance products have been extensively evaluated by the MAIAC team during algorithm development and testing. The test criteria were visual inspection for the lack of cloud/cloud shadow/aerosol artifacts, the realism of RGB images, and the consistency of the time series [Lyapustin et al., 2012]. Our current large-scale analysis is nearing completion and will be reported here soon. Initial results show highly accurate BRF from improved cloud detection, aerosol retrievals and the use of a Green’s function solution, instead of a Lambertian approximation, to derive BRF. This study also found that MAIAC BRF has a low sensitivity to variations in AOD. In cloudy conditions, we found MAIAC provides more high-quality retrievals (up to a factor of 3), confirming an earlier analysis by Hilker et al. [2012] in the highly cloudy Amazon. These results are consistent with those from Chen et al. [2017], based on a limited set of data showing an increase in spatial coverage of the best quality, high-precision LAI retrievals, using MAIAC input.

MAIAC surface reflectance and BRDF-normalization to the fixed view geometry was most thoroughly investigated and used over the tropics by the land community, particularly over the often cloudy Amazon basin (Hilker et al. [2012; 2014; 2015]; Bi et al. [2016]; Maeda et al. [2016] and others). A global inter-comparison with standard MODIS products is pending.

In addition to the reflectance layers, the MCD19A1 includes a Snow QA bit mask, as well as sub-pixel snow fraction and snow grain size layers at 1km. The MAIAC snow mask was evaluated by Cooper et al. [2018] over North America in 2015 against the ground Global Historical Climatology Network-Daily (GHCN-D) database of over 10,000 measurements. The study reported an 82% F-score for MAIAC (the F-score provides a balanced performance assessment combining accuracy and precision, or false positives and false negatives), and an average improvement of 30% in snow detection compared to the NOAA Near-real-time Ice and Snow Extent (NISE) and standard MODIS snow products in 2015.

The MAIAC snow grain size layer was validated against ground measurements in Alaska and Japan (2001-2005) in the case of pure snow (Lyapustin et al., 2009). The retrievals correlate well with measurements in the range of radii ˜0.1–1 mm, although retrieved optical diameter may be about a factor of 1.5 lower than the physical measured diameter.

The Snow Fraction layer (sub-pixel, at 1km) has thus far only been visually evaluated, as part of the MAIAC development and assessment, using new snowfalls, periods of melting and snow seasonality. A validation using high-resolution Landsat data is pending.