Despite all efforts to improve numerical weather prediction
models and assimilate new data types, forecast uncertainty
remains an unavoidable part of weather forecasting. Uncertainties
arise in all steps of the numerical forecast, including
uncertainties associated with initial conditions, uncertainties
associated with the model itself, and (for limited-area models)
uncertainties associated with lateral boundary conditions from an
outer model.  The preferred way to quantify the forecast
uncertainty is through the use of ensemble forecasts,
i.e. instead of running a single forecast we run several
forecasts with perturbed initial and lateral boundary conditions
and different model configurations, where each perturbed forecast
represents a possible outcome.

At DMI we have for more than five years been running a
short-range ensemble prediction system using the HIRLAM-model.
It is now time to upgrade to a new - and hopefully improved -
ensemble system.  The improvements include
- use of the HARMONIE-AROME model for half the members;
- increased horizontal and vertical resolution;
- rapid updates with new members added to the ensemble every hour;
- more observation data types included;
- more initial and lateral boundary condition perturbations.

In the presentation I will give more details about COMEPS,
including an update on the status of the work towards
operationalization.

Date: 20 April, 2017

Time: 11:00-11:45

Place: Nordlys, DMI

Here is a link to the presentation.