Evaluating Retrieval Algorithm Climate Stability: Toward Obtaining Large-scale 3D Optical Depth Bias Distributions by Cloud Type

This presentation will focus on the beginning of our studies to evaluate the impact that time-varying algorithm artifacts may have on detecting trends in cloud properties. Specifically, these studies will focus on the effect that the multi-decadal stability of the 3D optical depth (τ_c) bias (i.e. the bias induced by using the 1D radiative transfer assumption in satellite retrievals) has on our ability to unambiguously detect trends in cloud optical depth. Previous studies have evaluated the nature and magnitude of the 3D τ_c bias on small, instantaneous scales (showing a strong dependence on cloud type and sun-view geometry), but none have studied the stability of this bias on large spatiotemporal scales relevant to climate trends.

Two requirements will enable us to separate a 3D τ_c bias trend from a true τ_c trend. First, large spatial scale distributions of the 3D τ_c bias for different cloud types are needed; second, projections of how the frequency of occurrence of different cloud types may change between two climate states must be estimated. After combining cloud type-dependent errors and cloud type changes, the 3D τ_c bias trend will be estimated.

A brute-force method to obtain large scale distributions of the 3D τ_c bias would be a computationally intractable challenge due to the computational expense of 3D radiative transfer modeling simulations. Our novel approach manages this challenge, limiting the number of 3D radiative transfer simulations to only those needed to establish a statistical relationship between the 3D τ_c bias for known cloud fields and a proxy of the bias.

This proxy of the 3D τ_c bias is the angular consistency metric that was developed using fused Moderate-Resolution Imaging Spectroradiometer (MODIS) and Multi-angle Imaging Spectroradiometer (MISR) measurements. This metric quantifies the departure of clouds from satisfying the 1D radiative transfer assumption and can therefore be considered a proxy of the 3D τ_c bias. This presentation will focus on our evaluation of this metric for different International Satellite Cloud Climatology Project (ISCCP) cloud types over ocean. Additionally, this presentation will show our analysis of how the proxy may change between two climate states using output from select CFMIP2/CMIP5 models that correctly implement the ISCCP instrument simulator.