4.2 The Future of the Operational Environment

4.2.2 New Analysis Techniques

Deficiencies of observations in areas prone to polar low development hampers the ability to detect the atmospheric conditions necessary for polar low formation. This limits the numerical model’s ability to trigger polar low development.

Recently, the use of variational assimilation schemes in numerical models has allowed direct assimilation of satellite radiances. Improvements in satellite coverage and resolution will add to this in years to come. The full benefits of such a system are realized where models incorporate a complete representation of the stratosphere, as is the case with leading models today.

An extension of the 3D variational analysis scheme consists of applying a time dimension to the process at the heart of the assimilation cycle. The 4D variational assimilation (4DVAR) is capable of assimilating data at their exact time of observation. It is a computationally intensive system and is in development at a number of modeling centres. The observations are assimilated in a continuous way through a 12-hour model period in order to achieve an optimum fit of the model to the observation dimensions. In this way, the analysis at a particular point can be influenced by the effects of a real observation made at a previous time and different position in a realistic manner. This helps to improve the analysis over data sparse areas, such as the arctic ice fields.

4.2.3 Non-hydrostatic Models

Operational forecast models around the world make use of hydrostatic approximation. Hydrostatic approximation essentially states that there is an equilibrium between the gravitational and the vertical pressure forces. Vertical velocity accelerations are therefore neglected. This approximation is widely used in models because it holds true down to resolutions as low as 10 km.

The next generation of models will certainly have resolutions where the hydrostatic assumption breaks down. These non-hydrostatic models will be run at resolutions so fine (~2 km) as to allow direct simulation of convection. This considerably reduces the errors associated with convective parameterization and other processes impacted by vertical velocity.

It is worth noting that some modeling centres have begun using non-hydrostatic models operationally for both global and regional forecasting.

4.2.4 Regional Ensemble Forecasting

Many factors limit the extent to which deterministic models can by improved. Uncertainties about the state of the atmosphere will always limit the accuracy of initial conditions, despite improvements to analysis systems. An alternative to a purely deterministic forecast is the use of ensemble forecasting.

Ensemble forecasting provides a practical way of addressing variability in the initial conditions and uncertainties in a model. An ensemble forecast compares the outcome from many different forecasts for the same domain and time period using different models, parameterizations, or initial conditions. This comparison can make up for inaccuracies in initial conditions and lends itself to forecasting probabilities.

So far the approach has mainly been used for medium-range forecasts using ensembles of numerical models of relatively coarse resolution. This method also is of interest for mesoscale forecasting. Among the challenges posed by the new approach is the selection of appropriate perturbations to the initial conditions to adequately capture the uncertainties of the initial conditions.

Given the vast, data-sparse region over which polar lows occur, ensemble forecasts may eventually prove a valuable tool for the operational forecaster.

References

Parker, Neil, 1997. Cold Air Vortices and Polar Low Handbook for Canadian Meteorologists. Environment Canada