Recent and Planned Upgrades to Operational HWRF Data Assimilation

The NOAA/NCEP HWRF model has undergone significant upgrades over the past few years, including major improvements to its data assimilation (DA) system. Past versions of HWRF used error covariance provided by the NCEP global model, which is suboptimal for hurricane applications, for the flow-dependent portion in a GSI hybrid scheme. The 2017 version of HWRF introduced a fully cycled EnKF to provide error covariance for the hybrid, which is more accurate for tropical cyclones. Concomitant with the DA system upgrades, HWRF now makes much greater use of data than in the past, particularly in the inner core. A number of data types are now being assimilated that previously were not, including several classes of atmospheric motion vectors as well as flight-level reconnaissance data. Data impact tests revealed that this data significantly improves HWRF forecasts and is at least partly responsible for much improved HWRF performance in 2017. The improvements in data usage, particularly for the inner core, mean that inner core reconnaissance is becoming increasingly important for improving intensity forecasts.

Additional possible upgrades for 2018 and beyond will be discussed. Ongoing experiments suggest that improving dropsonde assimilation can improve intensity forecasts by at least 15% at 48 h, and tuning the error covariance in GSI can offer additional improvements of 15%. Other data types, including SFMR, will also be tested for possible 2018 assimilation. One potentially significant change would be to discontinue use of the vortex modification scheme, which modifies the 6-h forecast to provide a first-guess for assimilation, when inner core data is available. Further desired upgrades to the DA system and data being assimilated will be discussed.