Very short term forecasting data (generally up to 3-6 hours) is obtained through specific weather nowcasting techniques characterized by the use of real-time data inputs (e.g. radar data, lightning, observations in general) that are then integrated and reanalysed with high-resolution modelling simulations. Recently, these techniques have seen a strong boost in development thanks to the use of specific artificial intelligence algorithms that improve their performance, both in terms of prediction and reduced processing time.

Nowcasting weather

Nowcasting weather forecasts can involve all the main meteorological parameters — precipitation, temperature, solar radiation, cloud cover, wind — and are characterised by a high update frequency (up to 5 minutes), a frequency that can vary and is customisable depending on the parameter of interest. In the service operational time window — i.e. for the next 3 to 6 hours — this type of supply is much more reliable and performs better than the weather forecast produced using conventional methods.
Nowcasting techniques are often used in hydrology for very short-term weather forecasting of precipitation and, in particular, of storms for which the space-time scale is very short and thus require continuous updating with observed data. However, such techniques are also increasingly used in other areas, such as in the energy sector (e.g. to forecast wind farm or photovoltaic plant production) or the mobility/transportation sector (e.g. to provide travellers with real-time information regarding adverse weather events that could make driving difficult or lengthen travel times).