The goal of this research project is to improve nowcasting and probabilistic forecasting of clouds and precipitation in the weather model Harmonie, which is operational at KNMI in ensemble mode.
This will be achieved by simultaneously improving initial clouds and precipitation in all model members. Clouds may contain ice crystals and/or drops; precipitation is either rain, snow or graupel, or a mixture of these.
It will be made possible to remove initial clouds and/or precipitation in the model if observations indicate clear skies and/or dry weather, and also the other way around: clouds and precipitation can be created if they exist in reality but are absent or at wrong locations in the model initially. Thus better initialisation can also lead to better forecasts.
Observations used include multiple cloud products from the Meteosat satellite, and precipitation products from the two dual-polarization C-band Doppler radars of KNMI.
Better cloud and precipitation forecasts have great potential for improved forecasts of short- and longwave radiation, 2-m temperatures, the height of cloud ceiling, and wind gusts and thunderstorms. These forecasts can help forecasters in the KNMI weather room, and they are useful for aviation, warnings for rail and road traffic, and solar electricity production. The improved forecasts can also contribute to the ‘Early Warning Centre’ of KNMI.
See Fig. 1 for an illustration of the impact of introducing initial satellite clouds on forecast shortwave global radiation in Harmonie ensembles for the location of Cabauw. Analysis time is 6 UTC, and observations are shown as well. It is seen that the ensemble of shortwave radiation forecasts is around the observations in case of satellite cloud initialisation, see Fig. 1b), contrary to the control run, see Fig. 1a).