Weather forecasts using Artificial Intelligence and Large Language Models

Weathermap begins using AI LLM for day 1 and 2 weather descriptions.

Figure 1: The new Artificial Intelligence and Large Language Model automated summary forecast and the feedback loop on www.weathermap.co.nz

In conjunction with Neil Gordon, MetOcean and Weathermap.co.nz have released an update to Weathermap that uses a specific weather AI LLM developed by Neil Gordon for Day 1 and 2 written weather briefings/forecasts for a select subset of sites (Auckland, Hamilton, Raglan, Wellington, Christchurch and Dunedin).   

Dr. Neil Gordon has a fantastic legacy in New Zealand meteorology - he worked for the Meteorological Service of New Zealand from 1968 to 2011.  The first half of his career including observing, bench forecasting, and research - specialising in the application of computers and models in forecasting. His responsibilities as a senior manager in the second half of his career included managing the entire forecasting function from 1988 to 1992, and the National Weather Services division from 1992 to 2005.  His LLM API can be accessed here:

https://chatgpt.com/g/g-ydDNVGsoT-global-weather-forecaster 

Brett Beamsley – GM Research and Innovation Hub says, “We’re beginning with a subset of sites in order to understand how the fully automated AI LLM can work alongside other automated services and gather feedback from end users. To this end we’ve included a link to an online Survey and will be encouraging people to fill in and feedback on how they perceive the forecasts.” 

The use of a fully automated weather specific AI LLM is in keeping with Weathermap’s non-intervened forecasts and is used to supplement the automated graphical forecasts on the site. 

Neil has developed the API so that it references official watches and warnings issued by MetService when a specific forecast location falls within one of the associated warning/watch polygons – that way the API keeps consistent with official MetService, and ensures there is a single ‘voice of truth’ with respect to severe weather conditions. 

Figure 2: Auckland’s AI LLM Summary and feedback loop for Tuesday 16 September