Fostering Vibrant Development of Physical Parameterizations via the CCPP

In order to improve numerical weather prediction, physical parameterizations need to be continually developed, either through improvements to existing schemes or through replacement of existing schemes with new ones. This development is typically done by groups of scientists, often distributed geographically and over multiple institutions with diverse goals. For example, developers of a single scheme may be U.S or internationally-based, be located in academic institutions, national research laboratories, or operational modeling centers, and be interested in applications ranging from short to seasonal time scales. Parameterization developers often work with more than one host model and are therefore taxed with getting their schemes working in multiple code bases. Ascertaining that the innovations necessary for a given application do not inadvertently affect others requires mechanisms to synchronize the code as multiple developers contribute to it. Additionally, aspects of interest to operational applications, such as computational performance, need to be considered in the development process and regularly assessed.

The Common Community Physics Package (CCPP) was designed by the Developmental Testbed Center (DTC) to address several of these challenges. It encompasses a library of physical parametrizations (CCPP Physics) and a Framework that can connect the library to practically any host model. The CCPP version 6 released in June 2022 contains 23 primary parameterizations representing a wide range of meteorological and surface processes. The CCPP supports a community of developers via open-source code, a well-established governance structure and a robust set of code management practices that include community reviews and testing for all participating host models.

The CCPP Framework supports configurations ranging from process-oriented diagnostic research studies to performance-oriented NWP prediction as it enables host models to assemble the parameterizations in flexible suites. Framework capabilities include flexibility with respect to the order in which schemes are called, ability to group parameterizations for calls in different parts of the host model (including the dynamical core), and ability to call some parameterizations more often than others with a shorter time step or in an iterative process.

The CCPP public releases include the CCPP Single-Column Model (SCM), a simple host that can be used to conduct hierarchical studies in which physics and dynamics are decoupled to isolate processes and more easily identify systematic model biases. The CCPP SCM can also be employed for rapid prototyping and testing because it is computationally affordable. The CCPP SCM can be executed over a range of weather scenarios using cases derived from observational field campaigns and is being widely used for a number of hierarchical system development activities.

The CCPP has demonstrated reliability and performance for use in an operational setting and is a cornerstone of the Unified Forecast System (UFS) infrastructure. It is on track for inclusion in all upcoming NOAA operational implementations of the UFS Weather Model and its applications, such as the Hurricane Analysis and Forecast System (HAFS), the Rapid Refresh Forecast System (RRFS), and the Global Forecast System (GFS). It is also at different stages of adoption by selected models developed by the Naval Research Laboratory (NRL) and under the auspices of the National Center for Atmospheric Research (NCAR).

In this presentation, we will provide an update on CCPP development and plans, as well as review existing resources for users and developers, such as the public releases, documentation, tutorial, and support.