In this paper, the application of an algorithm for precipitation retrieval is studied based on the statistical analysis of the changes of brightness temperature gradient in different infrared spectra of Advanced Himawari Imager(AHI) of H8 in the field of view of "precipitation" and "non-precipitation". Taking Anhui region as an example, when precipitation occurs, there is some change in brightness temperature gradient of AHI channel 7-16. Furthermore, dictionary learning and regularization constraints are used on precipitation retrieval. Firstly, based on the 118/AHI spectral brightness temperature data and GPM precipitation, spectral "brightness temperature" and "precipitation" dictionary are matched as historical sample databases. Secondly, K-nearest neighbor ( KNN ) method is used to identify "precipitation" and "non-precipitation" signals on the brightness temperature of the infrared spectrum based on the "dictionary". Finally, precipitation retrieval for infrared data is carried out in the precipitation signal "subspace" with regularization constraints. The preliminary experimental results show that precipitation structure based on brightness temperature for H8/AHI , which was retrieved by using the Bayesian model averaging-gamma probability distribution model, has a good similarity with GPM, as well as low relative error, and the critical success index is higher than others. Furthermore, the algorithm is extended and applied to the AHI brightness temperature retrieval of typhoon "Maria" precipitation, and the spiral rain belt can be obtained.
