A major set of changes was made to the Environment Canada global deterministic prediction system during the fall of 2014, including the replacement of four-dimensional variational data assimilation (4DVar) by four-dimensional ensemble–variational data assimilation (4DEnVar). The new system provides improved forecast accuracy relative to the previous system, based on results from two sets of two-month data assimilation and forecast experiments. The improvements are largest at shorter lead times, but significant improvements are maintained in the 120-h forecasts for most regions and vertical levels. The improvements result from the combined impact of numerous changes, in addition to the use of 4DEnVar. These include an improved treatment of radiosonde and aircraft observations, an improved radiance bias correction procedure, the assimilation of ground-based GPS data, a doubling of the number of assimilated channels from hyperspectral infrared sounders, and an improved approach for initializing model forecasts. Because of the replacement of 4DVar with 4DEnVar, the new system is also more computationally efficient and easier to parallelize, facilitating a doubling of the analysis increment horizontal resolution. Replacement of a full-field digital filter with the 4D incremental analysis update approach, and the recycling of several key variables that are not directly analyzed significantly reduced the model spinup during both the data assimilation cycle and in medium-range forecasts.

A new system that resolves the stratosphere was implemented for operational medium-range weather forecasts at the Canadian Meteorological Centre. The model lid was raised from 10 to 0.1 hPa, parameterization schemes for nonorographic gravity wave tendencies and methane oxidation were introduced, and a new radiation scheme was implemented. Because of the higher lid height of 0.1 hPa, new measurements between 10 and 0.1 hPa were also added. This new high-top system resulted not only in dramatically improved forecasts of the stratosphere, but also in large improvements in medium-range tropospheric forecast skill. Pairs of assimilation experiments reveal that most of the stratospheric and tropospheric forecast improvement is obtained without the extra observations in the upper stratosphere. However, these observations further improve forecasts in the winter hemisphere but not in the summer hemisphere. Pairs of forecast experiments were run in which initial conditions were the same for each experiment but the forecast model differed. The large improvements in stratospheric forecast skill are found to be due to the higher lid height of the new model. The new radiation scheme helps to improve tropospheric forecasts. However, the degree of improvement seen in tropospheric forecast skill could not be entirely explained with these purely forecast experiments. It is hypothesized that the cycling of a better model and assimilation provide improved initial conditions, which result in improved forecasts.

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.

Precipitable water (PW) derived from the GPS zenith tropospheric delay (ZTD) is evaluated (as a first step toward variational data assimilation) through comparison with that of collocated radiosondes (RS_PW), operational analyses, and 6-h forecasts (from the Canadian Global Environmental Multiscale model) of the Canadian Meteorological Centre. Two sources of ZTD data are considered: 1) final ZTD (over Canada), computed by the Geodetic Survey Division (GSD) of Natural Resources Canada, and 2) final ZTD (distributed globally), obtained from the International GPS Service (IGS). The mean GSD GPS–derived PW (GPS_PW) is 14.9 mm (reflecting the relatively cold Canadian climate), whereas that of the IGS dataset is 20.8 mm. Intercomparison statistics [correlation, standard deviation (SD), and bias] between GPS_PW and RS_PW are, respectively, 0.97, 2.04 mm, and 1.35 mm for the GSD data and 0.98, 2.6 mm, and 0.67 mm for the IGS data. Comparisons of GPS_PW with 6-h forecast PW (TRIAL_PW) show slightly lower correlations and a higher SD. The increase in SD is greater for the IGS data, which is not surprising, because in regions such as the Tropics and subtropics, moisture forecasts are of a lower quality and the RS observation network is sparse. From a three-way intercomparison (IGS GPS_PW, RS_PW, and TRIAL_PW) of the SD statistics, it is found that GPS_PW has the lowest estimated PW error (≈1 mm) for PW in the 5–30-mm range. For PW greater than 30 mm, the RS_PW estimated error is ≈2 mm, and that of GPS_PW is ≈2.5 mm. The TRIAL_PW estimated error increases with PW, reaching 5.5 mm in the 40–55-mm PW range. These intercomparison results indicate that GPS_PW should be a useful source of humidity information for NWP applications.

No abstract