Quantifying uncertainties of cloud microphysical property retrievals with a perturbation method

摘要

Quantifying the uncertainty of cloud retrievals is an emerging topic important for both cloud process studies and modeling studies. This paper presents a general approach to estimate uncertainties in ground-based retrievals of cloud properties. This approach, called the perturbation method, quantifies the cloud retrieval uncertainties by perturbing the cloud retrieval influential factors (like inputs and parameters) within their error ranges. The error ranges for the cloud retrieval inputs and parameters are determined by either instrument limitations or comparisons against aircraft observations. With the knowledge from observations and the retrieval algorithms, the perturbation method can provide an estimate of the cloud retrieval uncertainties, regardless of the complexity (like nonlinearity) of the retrieval algorithm. The relative contribution to the uncertainties of retrieved cloud properties from the inputs, assumptions, and parameterizations can also be assessed with this perturbation method. As an example, we apply this approach to the Atmospheric Radiation Measurement Program baseline retrieval, MICROBASE. Only nonprecipitating single-phase (liquid or ice) clouds have been examined in this study. Results reveal that different influential factors play the dominant contributing role to the uncertainties of different cloud properties. To reduce uncertainties in cloud retrievals, future efforts should be emphasized on the major contributing factors for considered cloud properties. This study also shows high sensitivity of cloud retrieval uncertainties to different cloud types, with the largest uncertainties for deep convective clouds. Limitations and further efforts for this uncertainty quantification method are discussed.