See what the weather radar will look like up to 2 hours into the future
Predict where rain will be in the next 1-2 hours
Our radar forecast feature uses sophisticated nowcasting algorithms to predict where rainfall will move over the next couple of hours. Unlike traditional weather forecasts that use large-scale atmospheric models, nowcasting extrapolates the movement of existing radar echoes to predict their future positions.
This approach works particularly well for short-term predictions (up to about 2 hours) because it captures the current state of precipitation systems and projects them forward based on observed motion patterns.
Our radar forecast system uses pysteps, an open-source Python library developed by the meteorological community for probabilistic precipitation nowcasting.
The system:
The result is seamlessly blended into the radar animation, so you see both observed radar data and forecast data in one continuous loop.
Darwin radar showing live data with forecast overlay
Not all rain is equal. Light scattered showers behave differently to organised storm systems. Our system automatically analyses radar activity and selects the most appropriate forecasting method:
This adaptive approach ensures computational resources are used efficiently while providing the best possible forecast for current conditions.
Sydney radar with forecast enabled
The feature is enabled by default for all users
Radar forecasts are automatically shown when viewing any HD radar page. You will see the animation continue past the current time, showing predicted radar returns for the next 1-2 hours.
Forecast frames are visually integrated with observed data, creating a seamless animation from past observations through to future predictions. This makes it easy to see where rain is headed and plan accordingly.
You can adjust the forecast duration in your account settings:
Understanding when nowcasting works best
Radar nowcasting excels at short-term prediction but has inherent limitations:
For longer-range forecasts, traditional NWP (Numerical Weather Prediction) models are more appropriate. Our location forecasts provide 7-day predictions using these methods.