Synergy between UV and VIS-NIR sensor to improve aerosol retrievals of optical depth, absorption and height: Examples from collocated observations of OMPS and VIIRS

Abstract:

The collocation of near-ultraviolet (sensor OMPS-NM) and visible/near-infrared sensors (VIIRS) in the Suomi-NPP satellite provides a unique opportunity to address some of the current challenges in remote sensing of aerosols using algorithms currently customized to work with each of these sensors. The Dark Target (DT) aerosol algorithm for retrievals with VIIRS uses aerosol models with no absorption variability among them. In scenes with high concentration of absorbing aerosols (e.g. smoke plumes), the retrieved aerosol optical depth (AOD) exhibits important deviations from the 1-to-1 line when compared to independent observations. Many factors are at play in these deviations but in the case of biomass burning, one of the main causes has been traced to the lack of variability in absorption of the assumed aerosol models. In UV-only aerosol retrievals such as those from the OMAERUV algorithm, the methodology relies on solving an inversion where three physical quantities (AOD, single scatter albedo or SSA and aerosol height or Z) modulate the observed radiance at two wavelengths. The current approach , customized for the OMI sensor, uses climatological values of Z to constrain the retrieval of AOD and SSA at near UV ranges. With collocation of OMPS-NM and VIIRS, it is possible to tackle the above-mentioned issues by creating a synergistic algorithm.

The basic idea relies on the modifying both DT and OMAERUV algorithms to operate on collocated pixels and the output from each algorithm is utilized as input to the other algorithm. Specifically, the output AODs from the DT algorithm are used to constraint the retrieval of Z and SSA by the OMAERUV algorithm. In turn, the latter outputs an SSA that it is used as input into a DT algorithm (modified to ingest models with variable absorption). The algorithms are then run in loop until converge until both result in consistent outputs. The expectation is that both DT and OMAERUV outputs will improve when additional information is incorporated to aid each retrieval.

This presentation will illustrate the challenges to implement this methodology in two mature and well characterized algorithms. Cases studies of these synergistic retrievals will be shown with cases of smoke transport over the ocean and land.