Thursday, September 13, 2018

AI Hurricane Forecasts


The Deep Learning-based Hurricane Intensity Estimator is a new interactive map developed by NASA which uses artificial intelligence to assess the wind speed intensity of tropical storms. You can use the map at the moment to follow the progress of Hurricane Florence as it approaches the United States and to view the intensity assessments that the model has made for this huge tropical storm.

The model used by the map to assess the intensity of hurricanes is based on the the Dvorak technique. The Dvorak technique looks at patterns in infrared satellite imagery to determine the intensity of tropical cyclones. When these patterns are assessed by humans the intensity estimate is based on a subjective judgement. By instead using deep-learning to recognize the patterns in infrared satellite imagery NASA hopes to develop a more objective way to measure the intensity of tropical storms.

If you click on Hurricane Florence on the interactive map you can view the latest intensity estimate of the storm. You can also view and download the deep learning predictions made by the model since Hurricane Florence began.


The strong arrival of the 2018 hurricane season has led The Guardian to ask the question From Maria to Florence: are hurricanes getting worse? The article uses the mapped tracks of previous hurricanes to visualize the frequency and intensity of hurricanes and answer the question of whether hurricanes are becoming worse.

The Guardian's conclusion is that hurricanes are in general becoming slower, stronger and wetter because of warmer ocean temperatures.

Even though Hurricane Florence has weakened to a category 2 storm its slow speed means that it is likely to cause strong storm surges and bring a lot of rain. The storm is therefore still forecast to cause catastrophic flash flooding and prolonged significant river flooding.
Post a Comment