Granted, the plane-parallel/1D RT assumption is not unreasonable for spatially extended stratiform cloud layers, especially those climatically important ones that occur in the mid-latitudes off the western coasts of the continents, which are indeed known to be highly susceptible to indirect aerosol impacts. It is also tenable for smoothly distributed background aerosol layers, especially over the uniform and dark surface of the oceans, which indeed dominate the aerosol direct impacts on climate. These applications draw respectively on the optically thick and optically thin-to-intermediate regimes in classic 1D RT.
However, these 1D RT-friendly scenarios exclude cases that are critically important for climate physics. For starts, 1D RT---hence operational cloud remote sensing---fails catastrophically for cumuliform clouds that have fully 3D outer boundaries and internal structures as well as microphysics driven by shallow or deep convective dynamics. Yet these are also clouds that are known react to the presence of anthropogenic aerosol; they are also responsible for processing aerosols, in particular via precipitation. Furthermore, powerful sources of aerosols, especially of the absorbing type, start in the form of opaque plumes emanating from volcanoes, biomass burning, and so on. Aerosol modelers need accurate characterizations of these sources to predict correctly the long-range transport, hence climate impacts, of aerosols. As for 3D-shaped convective clouds, these plumes are not amenable to remote sensing diagnostics based on 1D RT.
In this presentation, we deliver a proof-of-concept for an entirely new kind of practical remote sensing applicable to 3D clouds and dense aerosol plumes (Bal et al., 2015). It is based on highly simplified 3D RT and only aims to reconstruct the outer shape of the cloud/plume from a multi-angular suite of images of the object at high spatial resolution. Such data can readily be collected with an airborne sensor like AirMSPI (Diner et al., 2013). AirMSPI is currently being developed at JPL and has been already been extensively tested on NASA's high-altitude ER-2 platform, and subsequently deployed during several field campaigns (PODEX, SEACR4S, CalWater2-2015).
The key ingredient of the reconstruction algorithm is a sophisticated solution of the nonlinear inverse problem via linearization of the forward model and a robust iteration scheme supported, when necessary, by adaptive regularization. In its current state, the demo uses a 2D ("flatland") setting to show how either vertical profiles or horizontal slices of the object can be accurately reconstructed. We believe that extension to 3D volumes is straightforward. We will also modify the shape retrieval algorithm to accommodate images at lower spatial resolution; such as can be readily obtained from the MISR instrument on NASA's Terra satellite.
In summary, we have made substantial progress toward an entirely new paradigm in the characterization of 3D clouds and dense aerosol plumes (i.e., near powerful sources, both natural and anthropogenic) by passive remote sensing methods. We recall that passive remote sensing, either airborne or space-based, is always more cost effective than in-situ probing in terms of spatial and temporal sampling. From the remote sensing perspective, in-situ data from cloud/plume-penetrating aircraft or drop-sondes can be used for algorithm validation by enabling a form of retrieval error quantification. We will discuss other validation methodologies that complement in-situ data, and other directions of future research.
G. Bal, J. Chen, and A.B. Davis. Reconstruction of cloud geometry from multi-angle images, submitted to Inverse Problems in Imaging (2015).
D.J. Diner, F. Xu, M.J. Garay, J.V. Martonchik, B.E. Rheingans, S. Geier, A. Davis, B.R. Hancock, V.M. Jovanovic, M.A. Bull, K. Capraro, R.A. Chipman, and S.C. McClain. The Airborne Multiangle Spectro-Polarimetric Imager (AirMSPI): A new tool for aerosol and cloud remote sensing, Atmospheric Measurement Techniques, 6(8), 2007-2025 (2013).