Retrieved precipitable water vapor from diffuse radiances measured by a sky-radiometer

Earth radiation budget is affected by aerosols, clouds, and gases, e.g. water vapor and ozone. Since these factors have large temporal and spatial variations, it is necessary to observe their long-term variations all over the world. Ground-based remote sensing by the sky-radiometer achieves the aim. The sky-radiometer measures the solar direct irradiances and the angular distributions of diffuse radiances at eleven wavelengths including aerosol (340, 380, 400, 500, 675, 870, 1020 nm), cloud (1627, 2200 nm), water vapor (940 nm), and ozone (315 nm) channels. This instrument is suitable for a long-term and automatic observation since the on-site self-calibration method, Improved Langley (IL) method, is developed. However, the IL method can be applied to only the aerosol channels. In this study, we developed the on-site self-calibration method for water vapor channel. This method consists of the 4 steps: (1) Aerosol channels are calibrated by the IL method. (2) Aerosol optical and microphysical properties are retrieved using direct solar irradiances and diffuse radiances at aerosol channels. The aerosol optical properties at water vapor channel are interpolated from those at aerosol channels, (3) The precipitable water vapor (PWV) is retrieved using the ratios of diffuse radiances to the direct irradiance at water vapor channel. (4) The calibration constant at water vapor channel is estimated from the PWV and aerosol optical properties. We conducted the intensive sensitivity tests of our method using the simulated data of the sky-radiometer. We could retrieve the PWV and the calibration constant exactly in the case that the PWV is less than 3 cm. We applied our method to the actual measurements in dry season at several sites of the SKYNET which is the international network of sky-radiometer for aerosol-cloud-solar radiation interaction researches. After determining the calibration constants at aerosol and water vapor channels, we retrieved PWV from direct solar irradiance in all seasons and compared them with those derived from GPS or microwave radiometer. The results showed good agreements at every site.