An overview of operational rainfall algorithm for AMI onboard GEO-KOMPSAT-2A satellite

An operational rainfall rate (RR) algorithm has been developed for the Advanced Meteorological Imager (AMI) onboard the Geo-KOMPSAT-2A (GK-2A) satellite launched on December 5, 2018. The algorithm uses the a-priori information including the microwave rainfall data from the low-earth orbiting satellites and infrared (IR) brightness temperatures from geostationary satellites. The algorithm can better perform with a variety of a-priori information describing all possible precipitating systems. In addition, separation of physically different precipitating systems in the retrieval process likely to improve the accuracy of retrieved rainfall data. However, it has been well known that such the classification of precipitating clouds can be hardly achieved based on the measurements of cloud top temperatures. For the classification, this algorithm utilizes the brightness temperature differences (BTDs) between IR channels to include the various radiative characteristics due to various distributions of hydrometeors and cloud thickness. The algorithm thus discriminates five types of precipitating clouds: one shallow and four non-shallow types. In addition to the classification of cloud types in a-priori databases, the algorithm also uses databases classified by latitudinal bands. The bands are separated with four latitudinal zones. The separation of databases based on latitudes may have an effect of distinguishing the cloud types that can occur regionally. The a-priori databases are thus classified with 20 different categories. Once the a-priori databases are constructed, the algorithm inverts the AMI IR brightness temperatures to the surface rainfall rate based on a Bayesian approach. The Bayesian approach has advantages on using multi-channel brightness temperatures simultaneously and utilizing the probability of rainfall reserved in the a-priori databases. This algorithm is applied to the Advanced Himawari Imager (AHI) and AMI data, respectively. Retrieval statistics and ongoing works for the algorithm improvements will be discussed.