Uncertainty in observational estimates of aerosol direct radiative effect and forcing

The most elementary understanding of how aerosols influence forcing of the Earth's climate begins with the aerosol direct radiative forcing (DRF). However, global constraints from current satellite observations are relatively poor. Even for the more basic aerosol direct radiative effect (DRE, the radiative effect of all aerosols both natural and anthropogenic), satellite remote sensing estimates vary by +/- 1.25 W/m^2 over clear-sky ocean (Yu et al. 2006; Henderson et al. 2013; Oikawa et al. 2013; Matus et al. 2015). This range is considerably larger than the stated uncertainties in each of these studies demonstrating the difficulty in even determining the uncertainties in the aerosol DRF/DRE. The importance of quantifying aerosol direct effects, and our continued difficulty in doing so, has earned it a position as an A-CCP science objective.

In this study, we focus on improving uncertainty estimates in remote sensing based calculations of the aerosol DRF/DRE by deriving a detailed set of aerosol DRE radiative kernels (Jacobians) from MERRA-2 reanalysis data. This comprehensive set of kernels provides a convenient way to assess the aerosol DRF/DRE uncertainties that result from observational or model-based uncertainties. Kernels are also a pragmatic way to identify how the current observational uncertainties might be reduced by future a A-CCP observing system. The A-CCP SATM (Science and Applications Traceability Matrix) geophysical variables and their desired accuracies have been chosen using these aerosol DRE kernels as a guide.

The aerosol DRE kernels have been used to assess the typical spectral and vertical simplifications inherent to remote sensing observations along with quantifying the relative importance of observing the aerosol DRE in different scene types. To obtain a lower bound on the current observational uncertainty in the aerosol DRF, the kernels are applied to the systematic uncertainties found in the AERONET version 3 database using a series of optimistic assumptions. This uncertainty estimate is compared to that found in modeling-based estimates of aerosol DRF and other estimates of the observational uncertainty given in previous studies.