Abstract. An algorithm based on CO 2 slicing, which has been used for cirrus cloud detection using thermal infrared data, was developed for high-resolution radiance spectra from satellites. The channels were reconstructed based on sensitivity height information of the original spectral channels to reduce the effects of measurement errors. Selection of the reconstructed channel pairs was optimized for several atmospheric profile patterns using simultaneous studies assuming a cloudy sky. That algorithm was applied to data by the Greenhouse gases Observing SATellite (GOSAT). Results were compared with those obtained from the space-borne lidar instrument on-board CloudAerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO). Monthly mean cloud amounts from the slicing generally agreed with those from CALIPSO observations despite some differences caused by surface temperature biases, optically very thin cirrus, multilayer structures of clouds, extremely low cloud tops, and specific atmospheric conditions. Comparison of coincident data showed good agreement, except for some cases, and revealed that the improved slicing method is more accurate than the traditional slicing method. Results also imply that improved slicing can detect lowlevel clouds with cloud top heights as low as approximately 1.5 km.