During the last two years, Environment Canada in Dorval has been developing a Regional Ensemble Kalman Filter (REnKF-15km) to furnish a more coherent ensemble of initial conditions to feed our operational Regional Ensemble Prediction System (REPS-15km).

To achieve this goal, substantial testing was performed to demonstrate the stability and precision of this system against the sequence of operational Global EnKF-50km systems developed during that period.

Intercomparisons between REnKF-15km and Gloabl EnKF-50km (e.g. same data sets and cycling frequency) show slightly better performance of REnKF-15km in terms of ensemble precision as judged from radiosonde verifications. Otherwise, other objective measures were found to be neutral. In addition, the quality of the downscaled trial fields from the Global EnKF-50km was comparable to that of REnKF-15km test cases. From there starts the next phase of development where substantially more high-resolution data (e.g. METAR, Radars) and faster analysis cycles (e.g. 3h) are considered in order to make a significant distinction with the Global EnKF in terms of improving short-term (<2 days) ensemble forecasting. In order to more adequately support our currently operational Canadian High Resolution Deterministic Prediction System, HRDPS-2.5km, the REnKF-15km will be reconfigured to run over the same horizontal domain as HRDPS-2.5km but will be run at 10km prior to assimilating new types of observations and coupling with the Canadian Land Data Assimilation System (CALDAS). Preliminary results with this new REnKF-10km data assimilation system and future development work will be presented.

---

*email: seung-jong.baek@canada.ca
*Preference: Oral