The impact of varying the radiation parameterization and adding a partial cloudiness scheme to Hurricane WRF

Testing conducted by both the NOAA Environmental Modeling Center (EMC) Hurricane team, and the Developmental Testbed Center (DTC), has shown that the radiation parmeterization used in the operational configuration of the Hurricane WRF (HWRF), the Geophysical Fluid Dynamics Laboratory (GFDL) radiation scheme, has its shortcomings for hurricane applications.

In 2013 the EMC Hurricane team tested the Rapid Radiation Transfer Model for Global Circulation Models (RRTMG) as an alternate, more sophisticated radiation parameterization for HWRF. Ultimately, RRTMG was not adopted at EMC for operations because it degraded intensity and track forecasts when combined with several other physics-related upgrades. During the same time period, DTC tested HWRF with an alternate physics suite containing the Thompson microphysics and a version of the RRTMG scheme that had been coupled with the Thompson microphysics, ensuring consistency of hydrometeor parameters between the packages. Although case studies indicated hurricane track and intensity forecast improvement when using the Thompson/RRTMG package, larger tests at the DTC mostly revealed statistically neutral-to-negative impacts on track and intensity forecasts especially in the northern Eastern Pacific basin.

Further analysis of the DTC large-scale test resulted in two notable findings. First, through funding from both the NOAA Hurricane Forecast Improvement Project (HFIP) and the DTC Visitor Program, Dr. Robert Fovell and his graduate student, Peggy Bu, showed that the RRTMG radiation scheme represented cloud radiative forcing more realistically than its GFDL counterpart. This GFDL deficiency was especially apparent for long wave tendencies at cloud top, and had a large impact on storm structure, intensity, and motion. Second, there was an overabundance of shortwave radiation reaching the ground for the Thompson/RRTMG experiment. DTC discovered that this was due to two reasons: a) only explicit clouds from the microphysics parameterization interact with the radiation scheme while the sub-grid scale clouds produced by the Simplified Arakawa Schubert (SAS) deep- and shallow-convection parameterization used in HWRF are transparent to the RRTMG scheme; and b) the coarse horizontal and vertical grid spacing in the HWRF parent domain, (much like other models) does not produce as many stratus clouds as observed.

The radiation imbalance finding led Greg Thompson of DTC to implement a scale-aware partial cloudiness scheme for RRTMG, which acts to simulate liquid- and ice-water content based on humidity and temperature thresholds to represent a “cloud” with radiative properties. The DTC performed a multi-storm HWRF test in which the RRTMG with partial cloudiness scheme replaced the operational GFDL radiation, but the operational Ferrier microphysics scheme was left unchanged. The results of the test show more realistic cloud and radiation distribution, as well as a neutral-to-positive impact in both the Atlantic and East Pacific basins for track and intensity when RRTMG with partial cloudiness is used. These results, along with synoptic- and storm-scale analyses and their interactions in cases will be discussed further.