Description and validation of the Japanese algorithm for radiative flux and heating rate products with all four EarthCARE instruments: pre-launch test with A-Train

Abstract

This study developed an algorithm for the Level 2 (L2) atmospheric radiation flux and heating rate product by a Japanese team for Earth Cloud, Aerosol and Radiation Explorer (EarthCARE). This product offers vertical profiles of downward and upward longwave (LW) and shortwave (SW) radiative fluxes and their atmospheric heating rates. This paper describes the algorithm developed for generating products, including the atmospheric radiative transfer model and input datasets, and its validation against measurement data of radiative fluxes. In the testing phase before the EarthCARE launch, we utilized A-Train data that provided input and output variables analogous to EarthCARE, so that the developed algorithm could be directly applied to EarthCARE after its launch. The results include comparisons of radiative fluxes between radiative transfer simulations and satellite and ground-based observations that quantify errors in computed radiative fluxes at the top of the atmosphere against Clouds and the Earth's Radiant Energy System (CERES) observations and their dependence on cloud type with varying thermodynamic phases. For SW fluxes, the bias was 24.4 W m−2, and the root mean square error (RMSE) was 36.3 W m−2 relative to the CERES observations at spatial and temporal scales of 5° and 1 month, respectively. On the other hand, LW exhibits a bias of −10.7 W m−2 and an RMSE of 14.2 W m−2. When considering different cloud phases, the SW water cloud exhibited a bias of −11.7 W m−2 and an RMSE of 46.2 W m−2, while the LW showed a bias of 0.8 W m−2 and an RMSE of 6.0 W m−2. When ice clouds were included, the SW bias ranged from 58.7 to 81.5 W m−2 and the RMSE from 72.8 to 91.6 W m−2 depending on the ice-containing cloud types, while the corresponding LW bias ranged from −8.8 to −28.4 W m−2 and the RMSE from 25.9 to 31.8 W m−2, indicating that the primary source of error was ice-containing clouds. The comparisons were further extended to various spatiotemporal scales to investigate the scale dependency of the flux errors. The SW component of this product exhibited an RMSE of approximately 30 W m−2 at spatial and temporal scales of 40° and 40 d, respectively, whereas the LW component did not show a significant decrease in RMSE with increasing spatiotemporal scale. Radiative transfer simulations were also compared with ground-based observations of the surface downward SW and LW radiative fluxes at selected locations. The results show that the bias and RMSE for SW are −17.6 and 172.0 W m−2, respectively, which are larger than those for LW that are −5.6 and 19.0 W m−2, respectively.