No matter how the weather changes, radar vision can Pierce the clouds, emergency rescue flood

2022-05-09 0 By

Floods are usually accompanied by a variety of adverse weather conditions, such as dark clouds, heavy rain and strong winds.Gpu-based data science systems can now help researchers and flood response teams “see through” it all.John Murray, visiting professor at the Geographic Data Science Laboratory at the University of Liverpool, has developed cuSAR, which uses Radar data from the European Space Agency to monitor conditions on the ground.CuSAR uses satellite data to create images that depict accurate geographic information about ground conditions in severe weather conditions.To create the radar vision platform, Murray used the NVIDIA RAPIDS software library suite and CUDA parallel computing platform as well as the NVIDIA GPU.Flood Emergency Response The platform was originally designed for the property insurance sector where mortgage and insurance providers needed to assess risk factors affecting property, including flooding.Using satellite data in this way requires clear ground visuals, but getting analyzable images can mean waiting for weeks in overcast British weather.With cuSAR, users can gain insights in near real time.The application case of radar vision platform has been extended to the field of security.The Geographic Data Science Laboratory was first contacted by the North Wales Regional Emergency Planning Authority a few years ago with a request to address severe flooding in the Dee Valley.Low, dense clouds enveloped the valley, making it impossible for the team to fly helicopters and drones to get a full picture of the flood plains along the river.Using an NVIDIA GPU-based image analysis platform, Murray delivers high-quality renderings of disaster areas in real time.The Emergency Planning Agency can use this information to allocate limited resources to critical areas and adjust its efforts as the floods evolve.Last year, the lab provided radar data to monitor a vaccine plant threatened by rising water levels.Emergency teams were able to get helicopters into the best position to fight the floods when weather conditions allowed.Correcting distorted view Creating analyzable images from radar data is no easy task.The curvature of the Earth, the perspective distortion of satellite images, requires the use of “rubber stretching” techniques, which are corrected mathematically and superimposed with position data to achieve accurate geolocation.A typical radar inventory contains billions of data points, presented as a grid.”We can’t image from radar data alone,” Murray says.There’s a lot of processing and math involved, and that’s where the GPU comes in.”Murray coded cuSAR using NVIDIA RAPIDS and Python Numba CUDA to seamlessly match radar and location data.Traditional Java or Python code typically takes about 40 minutes to output.With an NVIDIA GPU, it takes just 4 seconds.Once the data is processed, the platform outputs an image with precise geographic information that matches orDNance Survey grid coordinates.Within 15 minutes of receiving satellite data, it can be handed to emergency teams, giving them the information to respond quickly to rapidly changing conditions on the ground.Over the past decade, Britain has experienced some of its wettest months on record.Notably, 2020 was the first year on record to make the top 10 in all three key weather rankings — the warmest, wettest and sunniest.The weather service predicts severe flash floods will be nearly five times more likely in 50 years.Technologies like cuSAR enable researchers and first responders to monitor and respond to disasters in a timely manner, protecting homes and businesses that are most vulnerable to severe weather conditions.