The representation of Arctic mixed-phase clouds and their radiative properties in ECMWF during ACLOUD.

The representation of Arctic clouds in numerical weather prediction models is challenging, especially for mixed-phase clouds with both a liquid and ice phase present. This cloud type is frequently observed in the Arctic with a pronounced longevity. In this paper, measurements conducted during the Arctic CLoud Observations Using airborne measurements during polar Day (ACLOUD) campaign, which took place in May/June 2017 northwest of Svalbard, Norway, are compared with the operational 'Atmospheric Model high resolution' configuration (HRES) of the Integrated Forecasting System (IFS), operated by the European Centre for Medium-Range Weather Forecasts (ECMWF). Instead of using cloud retrieval products from airborne remote sensing, the comparison is performed in the observational space of spectral solar irradiances reflected by the clouds. To allow such an analysis along the flight track at flight level, the operational ecRad radiation scheme of the IFS is used in offline mode. Besides the HRES model output, vertical profiles of concentrations of trace and greenhouse gases provided by the ECMWF Atmospheric Composition Reanalysis 4 serve as the input for ecRad. The ability of the IFS to realistically represent the airborne radiation measurements collected during ACLOUD is evaluated for flight sections above sea ice and open ocean. Based on the good agreement of the HRES output with common sea-ice concentration satellite products, the radiation measurements are not significantly biased by sea-ice fraction. A case study with boundary-layer clouds of changing cloud phase is investigated in detail and the impact of the ice cloud mass mixing ratio on the spectral irradiance is investigated. For the cloud dominated by ice crystals, the common ice optics parameterization applied in the ecRad simulations is tested against more advanced scattering databases. The ice optics parameterization from Baran et al. (2016) agrees closest to ACLOUD observations in the 778−1242 nm and the 1626−1942 nm band, the parameterization from Yi et al. (2013) agrees closest in the 1299−1626 nm band. [ABSTRACT FROM AUTHOR]