Development and Evaluation of the NEPTUNE Prediction System

The Navy Environmental Prediction System Using a Nonhydrostatic Engine (NEPTUNE) is a next-generation global NWP system being developed at the Naval Research Laboratory (NRL). NEPTUNE makes use of a three-dimensional spectral element dynamical core and represents the first of its kind nonhydrostatic NWP system using a three-dimensional spectral element method. In this presentation, we will discuss the implementation and testing in NEPTUNE of different physical parameterization suites available in the Common Community Physics Package (CCPP). NEPTUNE global atmospheric predictions have been performed with horizontal grid spacings ranging from 9 to 20 km. We will discuss systematic errors that have been identified and address the sensitivity of these biases to changes in physical parameterizations, as well as horizontal and vertical resolution.

We conducted numerous NEPTUNE simulations with different physics configurations and evaluated the forecast skill and numerical stability. We show that using a CCPP physics suite implemented with the Tiedtke convection parameterization performs statistically better than the same suite using the Grell-Freitas and simplified Arakawa and Schubert (SAS) parameterizations through comparisons with ERA-5 reanalyses and the NASA IMERG precipitation dataset. The new physics suite using the Tiedtke scheme is not particularly sensitive to changes in the vertical resolution and improves skill as the horizontal grid spacing is increased to 9 km on the globe. We also investigated the impact of subgrid scale cloud fraction from the planetary boundary layer and convection schemes on radiative heating and the impact of different cloud fraction parameterizations based on Xu-Randall and Thompson microphysics on forecast skill.