MetOcean Solutions has received research funding for a project applying innovative technology to weather forecasting. The research, entitled ‘Machine learning for convective weather analysis and forecasting’, was funded in the Ministry of Business, Innovation and Employment (MBIE) 2017 Endeavour Round.
Led by MetOcean Solutions, the project is a collaboration with the Knowledge, Engineering & Discovery Research Institute (KEDRI) of Auckland University of Technology, and the New Zealand MetService.
“We are very pleased to receive funding for this exciting project,” states project lead Dr David Johnson.
The research will use machine learning to predict convective weather events. Convective weather produces heavy rain, lightning and strong winds, significantly impacting the safety, efficiency and well-being of New Zealanders, and recent severe events have caused loss of life and extensive damage to property.
“Convective weather is localised in time and space and can develop quite quickly and sometimes without much warning,” continues Dr Johnson. “These events are always associated with cloud masses - like big thunderstorm clouds - which can be seen from Earth-observing satellites. Currently, human forecasters detect these events by looking at satellite imagery, numerical model guidance, and rain radar. Trained forecasters are good at doing this, but humans cannot look everywhere all the time. Recent advances in machine learning mean that computers are now exceptionally good at identifying and labelling features in images. Our research will train a machine-learning algorithm to analyse satellite imagery, possibly combined with other inputs such as numerical model guidance or rain radar, and predict where and when heavy rain, lightning or wind squalls might occur. As this is a machine process it can potentially be fully automated and then used to send alerts on a phone app.
“There is a good chance that the algorithm will be better than a human at consistently making the correct predictions, which means that all New Zealanders will benefit from better weather forecasts. An automated forecasting system also allows for greater customisation and localisation for individual needs. If successful, the research could lead to apps that makes it possible for you to request a phone alert if heavy rain is likely at your location - your phone already knows where you are - allowing you to take in your washing, seek shelter or postpone your drive home. Many industries are weather-dependent, and accurate local forecasts of strong winds or heavy rainfall will help anyone working or organising events outdoors, including the forestry, fisheries, construction and transport industries, all of whom have different weather thresholds to ensure safety and efficiency. Human forecasters could never manage to serve all the myriad of end-user needs at different locations and times.
“The key to success is the collaboration with KEDRI, who are world-leading machine learning experts, and MetService who carry out day-to-day severe weather forecasts for New Zealand. MetOcean Solutions brings our track-record of applying state of the art science and technology to provide end-user tools and services.”
MBIE's 2017 Endeavour Round funded 68 new scientific research projects from 408 applications.
Click here for the MBIE press release on the funding.