Hurricane Unwinder is building an intensity forecast for hurricanes, typhoons and tropical cyclones. A one-point change on the five-point Saffir-Simpson scale statistically translates to a seven-fold change in damages and intensity information is thus of key importance. In the long run, our goal is to find weather modification opportunities for downgrading the intensity.
Customer segments for the forecast include disaster management and (collateralized re)insurance that can use the intelligence to avoid some of the multi-billion euro physical and financial damages that hurricanes, typhoons and tropical cyclones cause every year.
We provide a user interface with frequent updates, novel analytics and an intuitive presentation of probabilities for different intensities and tracks.
Hurricane intensity, even though crucial, has proven to be particularly challenging for standard weather forecasts. The track forecast error for hurricanes has improved by about 70%* in the last 30 years whereas intensity forecasts have only improved by about 20% (*24-h head-time forecast). Standard weather forecasts solve the equations of physics on a grid with resolution of about 6-10 km, meaning that key features of the low pressure area and the flow in the eye and eyewall get ‘pixelated’ and computing power is not expected to fix this anytime soon. Constant geostationary satellite monitoring is done at 1 km resolution every 30 minutes while low earth orbit satellites obtain snapshots at about 250-meter resolution. Together with data from standard weather forecasts our algorithm produces an improved forecast. We are powered by machine learning on a high-performance cloud server.
Our machine-learning algorithm using neural networks and novel statistical preprocessing can utilize details of high-resolution satellite images. NASA satellites Aqua, Terra, Suomi NPP, NOAA-20 and ESA Sentinel satellites provide high-resolution snapshots while the GOES-16, GOES-17, Himawari-8 and EUMETSAT Indian Ocean provide continuous geostationary satellite observations of all relevant ocean basins.
Our company was spun off from academic research on the possibilities to perform downgrading using weather modification technology. This was attempted in real life by American government agencies during Project Stormfury during 1962-1983 using cloud seeding with ambiguous but promising results. With current technologies, scientific advances of turbulent flows and our intensity forecast, we firmly believe that hurricane downgrading will be possible in the not-so-far future.
Hurricane Unwinder is the 2018 winner of the prestigious Copernicus Master award in the Disaster Management category.
Copyright © 2019 Hurricane Unwinder - All Rights Reserved.