Himawari-8 had observed with high and stable accuracy since it began operation in 2015. The Advanced Himawari Imager (AHI) onboard Himawari-9 is equivalent to that of Himawari-8 and has been stably observing with high accuracy since the switchover.
By using the high quality AHI Level-1 data, the Fundamental Cloud Product (FCP) has been generated at the Meteorological Satellite Center (MSC) for various downstream applications. FCP is a product that is obtained by calculating the cloud mask, cloud phase, and cloud-top height collectively, which are required as input for many Level-2 products. By obtaining cloud information required for many products from FCP, we can save computing resources and secure the quality of input data. FCP can be broadly divided into two algorithms. One is the cloud mask calculation algorithm created by combining the NWCSAF and GOES-R algorithms at MSC, and the other is the calculation of cloud physical parameters such as cloud phases created with reference to the GOES-R algorithm and cloud top height, which was extended from one to two layers of cloud radiation model at MSC with reference to the GOES-R algorithm.
JMA is developing a new FCP for introduction in FY2024, replacing the cloud mask algorithm with Cloud and Aerosol Unbiased Decision Intellectual Algorithm (CLAUDIA) (Ishida et al. 2018), which uses support vector machine techniques, and the calculation of cloud physics with a method based on a variational principle called Optimal Cloud Analysis (OCA), which was developed at EUMETSAT and introduced to MSC. Current cloud mask is based on threshold technique and uses Numerical Weather Prediction (NWP) data for the thresholds. Therefore, cloud detection is sometimes affected by phenomena that are not well represented by numerical models. On the contrary, since CLAUDIA uses only observation data to calculate the thresholds for cloud detection by machine-learning without any theoretical assumptions or use of the first guesses from NWP, it is expected to improve the accuracy of cloud detection. In addition, users have been requesting cloud optical depth for some time, and the introduction of OCA will allow us to respond to those requests. The new FCP is a method that can efficiently process the dramatically increasing information in the observation bands by using machine-learning cloud masks and cloud physics estimation based on the variational principle.
In this talk, the performance of Level-1 data and Level-2 products for Himawari-9 and the current status of the new FCP development will be presented.

