Retrieval of cirrus cloud radiative properties from brightness temperatures in infrared window bands

The present understanding of cirrus microphysical property climatology is limited, which is an important key for better understanding the earth radiation budget and climate. An algorithm using three infrared window bands of Moderate Resolution Imaging Spectroradiometer (MODIS) (bands 29, 31 and 32) has been developed for retrieval of cirrus radiative and microphysical properties. We have developed a semi-analytical formula of the brightness temperature, which represents the dependence of the infrared signal on atmospheric and surface parameters. The accuracy of the approximation is about 0.33 K in band 29 and 0.17 K in bands 31 and 32, with significant correlations between the errors in each band. The solution for the inverse problem is from an optimal estimation based on the maximum a posteori, where prior information, measurement noise and modeling error are taken into account. As known from previous studies, the infrared method is sensitive to the surface temperature and cloud top temperature, which should be given with high accuracy. Sensitivity tests for the brightness temperature and retrieval error analysis showed that compared to the two-band split-window method, the three-band retrieval is capable of reducing the retrieval errors in optical thickness (tau) for optically thin cirrus (tau < 1) and in effective particle radius (r(e)) for very small particle sizes (r(e) < 5 mu m). In general cases, the three-band retrieval is better to stably obtain the cirrus cloud properties with higher accuracy.