Improved MODIS Dark Target aerosol optical depth algorithm over land

Improved MODIS Dark Target aerosol optical depth algorithm over land: Angular effect correction

Wu, Y. (TU Delft Optical and Laser Remote Sensing)
de Graaf, M. (TU Delft Atmospheric Remote Sensing; Royal Netherlands Meteorological Institute (KNMI))
Menenti, M. (TU Delft Optical and Laser Remote Sensing)

2016-11-23

Aerosol optical depth (AOD) product retrieved from MODerate Resolution Imaging Spectroradiometer (MODIS) measurements has greatly benefited scientific research in climate change and air quality due to its high quality and large coverage over the globe. However, the current product (e.g., Collection 6) over land needs to be further improved. The is because AOD retrieval still suffers large uncertainty from the surface reflectance (e.g., anisotropic reflection) although the impacts of the surface reflectance have been largely reduced using the Dark Target (DT) algorithm. It has been shown that the AOD retrieval over dark surface can be improved by considering surface bidirectional distribution reflectance function (BRDF) effects in previous study. However, the relationship of the surface reflectance between visible and shortwave infrared band that applied in the previous study can lead to an angular dependence of the AOD retrieval. This has at least two reasons. The relationship based on the assumption of isotropic reflection or Lambertian surface is not suitable for the surface bidirectional reflectance factor (BRF). However, although the relationship varies with the surface cover type by considering the vegetation index NDVISWIR, this index itself has a directional effect and affects the estimation of the surface reflection, and it can lead to some errors in the AOD retrieval. To improve this situation, we derived a new relationship for the spectral surface BRF in this study, using 3 years of data from AERONET-based Surface Reflectance Validation Network (ASRVN). To test the performance of the new algorithm, two case studies were used: 2 years of data from North America and 4 months of data from the global land. The results show that the angular effects of the AOD retrieval are largely reduced in most cases, including fewer occurrences of negative retrievals. Particularly, for the global land case, the AOD retrieval was improved by the new algorithm compared to the previous study and MODIS Collection 6 DT algorithm, with the increase of 2.0 and 4.5 % AOD retrievals falling within the expected accuracy envelope ±(0.05 + 15 %), respectively. This implies that the users can get more accurate data without angular bias, i.e., more meaningful AOD data.

aerosols
MODIS
surface albedoi

http://resolver.tudelft.nl/uuid:4a3d6362-36ca-4cd6-aace-cca35244a212

https://doi.org/10.5194/amt-9-5575-2016

1867-1381

Atmospheric Measurement Techniques, 9 (11), 5575-5589

Institutional Repository

journal article

© 2016 Y. Wu, M. de Graaf, M. Menenti