Science news

FMI's researchers publish about 350 peer-reviewed articles annually.

In Science News we publish current information about FMI's studies on the weather, the sea and the climate.

New methods for short-range rainfall forecasting in urban environments

New methods for short-range rainfall forecasting in urban environments

Reliable rainfall and flood forecasts are important for densely populated urban areas. The study presents new approaches to using radar measurements for short-term rainfall forecasting. The reliability of the forecasts was tested using a radar that covers an urban metroplex in the United States.

The term nowcast is used for short-range forecasts (0-3 h) based on machine vision, learning or statistical models. Rapidly updating high-resolution nowcasts are particularly important for densely populated urban areas, where severe rainfall can trigger flash floods. Numerical weather prediction models, that describe atmospheric physics, are impractical for this purpose.

Weather radars collect measurements from 5-20 elevation angles, which allows determining the vertical profile of precipitation. This is particularly important for predicting convective storms. The traditional nowcasting methods are limited to using the lowest angles. A key result of the study is that the reliability of nowcasts can be significantly improved by using vertically integrated liquid computed from the whole radar volume. In addition, a new statistical prediction model was developed. Whereas the traditional models are predominantly predicting the motion of precipitation, the proposed model can also predict growth and decay.

The proposed model was validated by using the US National Weather Service NEXRAD radar that covers the Dallas-Fort Worth urban metroplex with over 7 million people. The validation was done in the near range of the radar (under 80 km), which allows a high vertical resolution. It was shown that the new model can produce up to 25% more reliable nowcasts than the existing ones. The model is also computationally efficient: for instance, it can produce one-hour nowcasts in a 1 km2 grid with 5-minute time resolution.

Further information:

researcher Seppo Pulkkinen, Finnish Meteorological Institute,

Pulkkinen, S., Chandrasekar, V., von Lerber, A. and Harri, A.-M., Nowcasting of Convective Rainfall using Volumetric Radar Observations, IEEE Transactions on Geoscience and Remote Sensing, 2020,

Science news archive