A system that uses artificial intelligence to predict wildfire spread paths in real time and deploy disaster text alerts and civil defense warnings has been developed by Gyeonggi Province.
The Gyeonggi Provincial Government's Emergency Planning Office announced Wednesday that its staff voluntarily formed an AI collaboration team and developed in-house a "Civil Defense Alert Prediction Model," tentatively named G-DAPS.
G-DAPS is a system that provides real-time predictions of expected wildfire paths, estimated arrival times, and projected alert issuance times from the moment a fire breaks out. The system was built using the Korea Meteorological Administration's short-term forecasts, the Korea Forest Service's wildfire risk forecasts, the Ministry of Land, Infrastructure and Transport's (MOLIT) digital twin open API, and audible range data from 589 civil defense alert facilities across the province.
The platform visualizes weather conditions, wildfire history, and alert facility audible ranges on a web-based map. It analyzes wildfire risk at 30-minute intervals and identifies affected local governments down to the eup, myeon, and dong level. Based on this information, on-duty personnel can issue alerts to specific areas or send disaster text messages.
The development process drew attention for the fact that government employees directly used generative AI tools such as Claude and Gemini to code the prediction and analysis system and study relevant cases. It represents a rare instance of civil servants building a disaster response system using AI without outsourcing.
Gyeonggi Province is currently in discussions with its Forestry Division and the Korea Forest Service on adopting the system. A pilot operation is expected to begin as early as this month. The province plans to expand the disaster safety network by sharing the system with all 31 cities and counties within Gyeonggi Province as well as other local governments, covering large-scale natural disasters such as floods and heavy snowfall.
"Starting with wildfires, we plan to continuously improve the prediction model's performance so it can also be applied to responses against large-scale natural disasters such as floods and heavy snow, as well as North Korea's trash and waste balloon provocations," said Cho Kwang-geun, Gyeonggi Province's Emergency Planning Director.
