Enhancements to Cloud Overlap Radiative Effects for Tropical Cyclone Prediction in HAFS

With support from the Developmental Testbed Center (DTC), NCAR and NOAA, this research is evaluating enhancements to the treatment of cloudy radiative transfer within the RRTMG radiation code in the context of tropical cyclone prediction using the initial operational implementation from 2023 of the Hurricane Analysis and Forecasting System (HAFS) model. A primary goal is to further unify the treatment of subgrid-scale clouds to ensure consistency in performance and encourage ease of maintenance across the spectrum of weather models. In HAFS, the default maximum-random (MR) method in RRTMG for representing the sub-grid radiative coupling, or vertical cloud overlap, of fractional clouds has been replaced with more advanced methods known as exponential (EXP) and exponential-random (ER) to examine their impact and potential benefit to tropical cyclone prediction. In the presence of partial cloudiness, the new methods relax the strict MR assumption of maximum overlap through adjacent cloudy layers by allowing the vertical correlation of clouds to transition exponentially from maximum to random with increasing layer thickness and with increasing distance through cloudy layers. The ER method adds a further level of randomization between non-adjacent blocks of clouds with clear layers in between. The EXP and ER methods include a decorrelation length - either fixed or varying by latitude and day of the year - that affects the rate of exponential transition of cloud overlap through partially cloudy layers. Initial testing in HAFS has demonstrated that EXP and ER significantly alter the profile of radiative heating rates in and around tropical cyclones and, in some instances, modify the atmospheric state sufficiently to change the track and intensity of tropical cyclones. To better understand the range of sensitivities to the choice of cloud overlap technique, we have applied the MR, EXP and ER overlap methods to a selection of 2023 Atlantic and Eastern Pacific basin high-impact hurricanes using the 2023 HAFS-A operational model. This configuration has a storm-following 2-km domain nested inside the 6-km parent and an FV3-GFS physics suite with tropical cyclone-specific modifications. Specific five-day forecast cycles were chosen to represent a variety of storm ages and environmental conditions. We will describe the results in the context of storm maturity, storm translation speeds and overall model predictability.