A unified approach to cirrus microphysics, remote sensing and climate prediction, and its impact in a GCM

The most recent report of the Intergovernmental Panel on Climate Change (IPCC) published in 2013 concluded that the coupling between clouds and the Earth's atmosphere is still one of the largest uncertainties in predicting climate change. One such cloud type that compounds this uncertainty is cirrus, due to in-part, its large range of ice crystal habits, which has propagated many area and mass-dimensional power laws in the literature, and the choice of power laws within global models affects the evolution of cirrus within those models. However, the consistent coupling of cirrus microphysics and remote sensing to the radiation scheme in operational global models, that are used to inform the IPCC, is still non-existent. This means that internal inconsistency within a physical model may give rise to systematic error cancellation when predicting the short-wave and long-wave fluxes at top-of-the-atmosphere (TOA). This inconsistency has arisen because cirrus microphysics and optical property parametrizations have developed independently of each other.

In this paper, we demonstrate that a more consistent approach between assumed cirrus area and mass-dimensional power laws, and an ice crystal optical model, where the ice crystal optical model has been optimized through the use of remote sensing, can lead to desirable radiative impacts at TOA in a GCM, as well as on the temperature structure of the atmosphere. The ice crystal optical model consists of an ensemble of cirrus ice crystals, comprising of six ice crystal members. The first member is the hexagonal ice column of aspect ratio unity, followed by six-branched bullet-rosettes, thereafter; hexagonal elements are randomly attached to each other, forming three, five, eight and finally ten monomer ice aggregates. Each member is assigned to a size bin of a particle size distribution (PSD), and the bulk extinction of the ensemble can be varied by changing the weighting applied to each member. It is demonstrated that the ensemble model predicts areas and masses of ice crystals that are to within the current observational uncertainty.

The PSDs applied to the ensemble, and used in the GCM cloud physics scheme, are generated from a moment estimation parametrization of many thousands of in-situ measured PSDs, where the effects of ice crystal shattering have been largely removed. In this parametrization, the moments of the PSD are related to the second moment (i.e., the ice water content) and in-cloud temperature via polynomial fits to the measured moments. The ice water content (IWC) is an important GCM prognostic variable, and in the cirrus bulk optical property parametrization of the mass extinction coefficient, single-scattering albedo and asymmetry parameter, it is the prognostic IWC that is the couple between the GCM cloud physics and radiation schemes. Therefore, the evolving cirrus GCM predicted IWC is directly linked to the predicted cirrus radiative properties, without diagnosing other properties within the radiation scheme, such as the ill-posed ice crystal effective dimension, which itself relies on unphysical deterministic relationships between temperature and/or IWC.

To optimize the cirrus bulk optical property parametrization applied to the GCM novel use is made of space-based radiometric observations. The active remote sensing observations used in this paper are taken from the DARDAR and 2C-ICE cloud profile products, both of which are based on the A-train 94 GHz cloud profiling radar and CALIOP lidar observations. These products are used to simulate the cirrus brightness temperatures at 8, 11 and 12 µm, and the simulations are compared against the radiometric measurements of IIR at 8, 11 and 12 µm ( IIR has a 1 km footprint, and it is therefore, well co-located with CALIOP), covering a total of 26791 globally distributed cirrus observations. We show that with well chosen microphysics, based on state-of-the-art microphysics observations applied to the ensemble model, it is possible to minimise radiometric differences between the simulations and IIR measurements, where the global mean bias < 1 K, thereby reducing the latitude dependence of the bulk optical properties to a very weak dependency, in contrast to other parametrizations, which may be applied to current climate models.

The remotely sensed globally optimized ensemble model bulk optical properties are parametrized into the Met Office Global atmosphere model version 6. In the GCM, the assumptions about cirrus PSDs, mass-dimensional relationship, fall speeds, and capacitance are consistent with the parametrized bulk optical properties, where the two schemes are directly linked via the GCM prognostic variable IWC. In this paper, the impact in the GCM of this unified parametrization on the twenty-year averaged-annual short-wave and long-wave fluxes at TOA, and on the temperature structure of the atmosphere will be presented in the form of comparisons against the current version of the Met Office Global Atmosphere model and observations. Using the unified parametrization, the GCM predicted global area-weighted averaged-annual short-wave and long-wave fluxes at TOA are directly compared against CERES observations, and the temperature structure of the atmosphere is compared against the ERA-Interim analysis of global atmospheric temperatures. The paper demonstrates that with an internally consistent approach, it is possible to improve a climate model, without reliance on systematic error cancellation between the cloud and radiation schemes, the latter being propagated in current climate models used to inform the IPCC.