Reproduction of atmospheric radiation by various cloud solvers in the ECRAD model

Comprehensive studies of the quality in reproduction of atmospheric radiation using various configurations and computational schemes in the ECRAD radiative model are important for improving the radiative blocks of numerical weather and climate forecasting models. The paper presents a study of the quality of radiative calculations using the ECRAD autonomous model [2] using various cloud solvers in cloudy conditions based on comparisons with high-precision measurements of the RAD-MSU (BSRN) radiation complex for the Meteorological Observatory of Moscow State University (MO MSU) [1]. The input data for the calculations were prepared for a single vertical grid using 137 levels. The main meteorological variables (including the cloud fraction and cloud water content) were taken from the ERA-5 reanalysis, the aerosol composition of the atmosphere was determined by the CAMS reanalysis, the main greenhouse gases - according to the climatological data of the IFS Cycle 46R1. The time step of calculations and comparisons in cloudy conditions is chosen to be equal to one hour for more accurate time averaging of the spatial structure of clouds and account for ergodicity. The radiative simulations were performed using three algorithms (McICA, Tripleclouds and SPARTACUS) with different settings and parameters.

The results of the comparisons were analyzed for different ranges of cloud cover, which was pre-synchronized according to calculations and visual observations at the MO MSU, as well as separately for optically dense continuous (overcast) cloud cover. In heterogeneous cloud conditions, the error distribution is generally the same for all cloud solvers. At the same time, the maximum errors were observed at cloud amount of 5-7, where there was a significant underestimation of the shortwave net irradiance up to 50-70 W/m². For a higher cloud amount (7-10), the maximum overestimation of the shortwave net irradiance up to 20-40 W/m² was detected. The observed differences are primarily related to the inaccuracy of reproducing direct shortwave radiation in cloud solvers. The longwave net irradiance is restored satisfactorily by all cloud solvers, the error does not exceed 10 W/m² for all step in cloud cover amount. In conditions of optically dense continuous clouds, the errors of shortwave net irradiance are positive for almost all algorithms, and negative for longwave net irradiance. It is necessary to highlight the important problem of non-zero calculation of direct solar radiation in cloud algorithms with continuous cloud cover. According to the analysis data, the best configuration in calculations of dense overcast clouds is the SPARTACUS algorithm with two cloud states ("2reg") and with the inclusion of 3D effects. Additionally, the radiative effects of the three-dimensional cloud structure ("3D"-"1D") and the additional cloud state ("3reg"-"2reg") for the SPARTACUS algorithm were evaluated.

The study was supported by the Russian Science Foundation (grant No. 25-27-00014).

1. Chubarova N. E., Rozental V. A., Zhdanova E. Yu., Poliukhov A. A. New radiation complex at the Moscow State University Meteorological Observatory of the BSRN standard: methodological aspects and first measurement results. // Opt. Atmos. Oc., 2022. – 35 - P. 670–678.
2. Hogan R.J. A flexible and efficient radiation scheme for the ECMWF model / R.J. Hogan, A.A. Bozzo // Journal of Advances in Modeling Earth Systems, 2018. – 10(8) – P. 1907–1929;