Implementation of Deterministic Weather Forecasting Systems Based on Ensemble–Variational Data Assimilation at Environment Canada. Part II: The Regional System

Abstract

The modifications to the data assimilation component of the Regional Deterministic Prediction System (RDPS) implemented at Environment Canada operations during the fall of 2014 are described. The main change is the replacement of the limited-area four-dimensional variational data assimilation (4DVar) algorithm for the limited-area analysis and the associated three-dimensional variational data assimilation (3DVar) scheme for the synchronous global driver analysis by the four-dimensional ensemble–variational data assimilation (4DEnVar) scheme presented in the first part of this study. It is shown that a 4DEnVar scheme using global background-error covariances can provide RDPS forecasts that are slightly improved compared to the previous operational approach, particularly during the first 24 h of the forecasts and in the summertime convective regime. Further forecast improvements were also made possible by upgrades in the assimilated observational data and by introducing the improved global analysis presented in the first part of this study in the RDPS intermittent cycling strategy. The computational savings brought by the 4DEnVar approach are also discussed.

Contributions of natural and anthropogenic sources to ambient ammonia in the Athabasca Oil Sands and north-western Canada

“…The analyses were obtained from ECCC archives (Buehner et al, 2013Caron et al, 2015), in order to prevent chaotic drift of the model meteorology from observations. Consequently, our simulation setup comprises simulations on the North American domain in 30 h cycles starting at 12:00 UTC, and the oil sands domain in 24 h cycles starting at 18:00 UTC (the 6 h lag being required to allow meteorological spinup of the lower resolution model).…”

Methods of investigating forecast error sensitivity to ensemble size in a limited-area convection-permitting ensemble

“…Various centres around the world have pushed the number of ensemble members to very large values (e.g. operationally N =256 at Environment Canada (Buehner et al, 2015;Caron et al, 2015), and non-operationally N = 256 (Yashiro et al, 2016) and even 30 N = 10240 (Kondo and Miyoshi, 2016) at RIKEN in Japan). The models used in these systems do not approach convectivescales though, so there is still much progress to be made to allow large ensembles to be used routinely with convective-scale models.…”

Impact of Lateral Boundary Conditions on Regional Analyses

“…This is the essential motivation of the work presented here. Taking advantage of the fact that the CRCM5 is very close to the limited-area regional model of MSC, GEM-LAM (Mailhot et al 2006), used to produce regional analyses (Caron et al 2015), it was technically possible to use the CRCM5 instead of the GEM-LAM in data assimilation and therefore, to benefit from the immense work done to validate the system for the large volume of assimilated data. Validating a model in this context is a long process, and this approach avoided many difficulties that arise when building a data assimilation system from scratch (e.g., quality control of the observations, detailed study of each observation operator, tuning of the error statistics).…”

“…In collaboration with Environment and Climate Change Canada (ECCC), the MSC regional ensemblevariational data assimilation system (EnVar) Caron et al 2015) was adapted to use the CRCM5 in the assimilation of all observations currently used at MSC. However, producing regional analyses in a fully cycled assimilation system gives rise to difficulties associated with the way the regional model is driven at its lateral boundaries.…”

“…The assimilation system is based on the fourdimensional ensemble-variational data assimilation (4DEnVar) recently implemented at MSC for deterministic weather prediction Caron et al 2015). The 4DEnVar is an incremental variational assimilation system based on the use of hybrid background error covariances with a weighted average of the static covariances B nmc formulation, used in the previous 4DVar and 3DVar systems, and a flowdependent 4D ensemble covariances B ens obtained from the ensemble Kalman filter assimilation system also used at MSC for its ensemble prediction system (Houtekamer et al 2014).…”