The most severe wildfire damage in Korea’s recorded history occurred in 2025. According to the Central Disaster and Safety Countermeasures Headquarters, wildfires erupted simultaneously between March 21 and March 30, 2025, resulting in 184 human casualties (28 deaths and 156 injuries) and extensive damage to approximately 104,000 hectares of forest. Damage to private property included 3,848 houses, 6,106 agricultural and fishing facilities, 17,158 agricultural machines, and 3,419 hectares of agricultural and forestry crops. In addition, 769 public facilities were damaged, including national heritage sites, temples, and road infrastructure. Weather conditions during this period were highly conducive to wildfire spread. The highest national average temperature on record (14.2℃) coincided with historically low precipitation levels, producing extremely dry conditions. In Andong, maximum instantaneous wind speeds reached 27.6 m/s. Because adverse weather conditions made it difficult to secure accurate information on fire lines, the timing of evacuations in response to wildfire spread could not be predicted. Limitations of the conventional evacuation system included reduced mobility among elderly residents, situations requiring evacuation across city and county boundaries, and delays in situation reporting caused by power outages and communication failures.

In Korea, both local governments and national public institutions are authorized to transmit EAM. This study investigated the regional transmission patterns of wildfire-related EAM issued in Korea in March 2025 to assess the current status of EAM services provided to residents during wildfire events. Particular emphasis was placed on analyzing the timing and content composition (intended communication objectives) of EAM by region. Figure 1 illustrates the scope and methodology of the analysis.

EAM data were collected from the Disaster Safety Information Sharing Platform (https://www.safetydata.go.kr/disaster-data/disasterNotification). A total of 1,380 wildfire-related disaster text messages transmitted between March 14 and April 1 were collected, encompassing the full period of wildfire occurrence and response.

In Korea, standardized disaster alert texts specify several wildfire-related message types, including wildfire occurrence notifications, evacuation recommendations, evacuation orders, risk forecasting, and warning or emergency alerts. In practice, however, EAM content varies according to situational demands and communication intent. Therefore, this study classified the intended objectives of EAM content into distinct categories, as summarized in Table 1.

As described above, wildfire-related 1,380 EAM messages transmitted between March 14 and April 1 were collected and their intentions were classified. The frequency of these messages was analyzed by region, date, and intended purpose. The results are presented in Figure 2. Regardless of message intent, EAM were transmitted more frequently during the period of concentrated damage from March 21 to March 30. Among all the EAM messages, “wildfire prevention information” accounted for the highest proportion (64.94%). However, there were relatively few EAM transmissions providing wildfire prevention information prior to wildfire outbreaks. “Evacuation information” accounted for 24.62% of all messages, while “wildfire status reporting” comprised 21.34%. By contrast, “traffic control information,” “critical service supply (suspension) information,” and “damage reporting information” accounted for only 2.13%, 0.76%, and 0.53%, of total transmissions, respectively.

As shown in Table 2, the Gyeongbuk region, which experienced the most extensive damage, exhibited the highest frequency of EAM transmission. Unlike the nationwide EAM distribution, messages related to “evacuation information” constituted the largest proportion (51.03%) of EAM in the Gyeongbuk region, followed by “wildfire prevention information” (31.24%). In contrast, in regions that experienced little or no damage, “wildfire prevention information” accounted for the majority of EAM transmissions.

Figure 3 shows the number of EAM transmissions in the Gyeongbuk region. The “evacuation information”-related EAM sent in Uisong, which was the wildfire ignition point, accounted for 31.12% of all “evacuation information”-related EAM transmissions in the Gyeongbuk region. The proportion was 40.82% for Andong, but only 6.63%, 9.69%, and 6.63% for Chongsong, Yeongyang, and Yeongdeok, respectively. In the Gyeongbuk region, the wildfire that broke out in Uisong spread sequentially to Andong, Chongsong, Yeongyang, and Yeongdeok. Considering this order of wildfire spread, a substantial number of human casualties occurred in Chongsong, Yeongyang, and Yeongdeok despite there being sufficient time to provide evacuation guidance (number of deaths: two in Uisong, five in Andong, four in Chongsong, seven in Yeongyang, and ten in Yeongdeok), indicating that EAM failed to play an adequate role in effective emergency response. Andong recorded the highest number of EAM transmissions related to “wildfire status reporting,” accounting for 50.00% of all such messages in the Gyeongbuk region. By contrast, EAM related to “traffic control information” and “critical service supply (suspension) information” were transmitted relatively infrequently. In the Gyeongbuk region, there were only 17 EAM transmissions related to “traffic control information” and ten EAM transmissions related to “critical service supply (suspension) information”.

According to wildfire statistics from the Korea Forest Service, 55.5% of wildfires over the past decade have occurred in spring (March to May). During spring, westerly winds develop over the central region of the Korean Peninsula due to the formation of high pressure in the south and low pressure in the north. These westerly winds pass through a narrow corridor between the inversion layer and the Baekdudaegan mountain range, producing hot and dry conditions accompanied by strong winds. Such wind conditions tilt flames, facilitating heat transfer to surrounding fuel and accelerating fire spread. In addition, airborne embers carried by strong winds can rapidly ignite new fires, making it difficult to deploy firefighting equipment (e.g., firefighting helicopters) and increasing the likelihood of large-scale wildfires [11]. Real-time analysis data from March 2025 provided by the National Forest Fire Risk Forecasting System (http://forestfire.nifos.go.kr/main.action) of the Korea Forest Service indicated that wildfire risk levels were high or extremely high across large portions of the country. In particular, wildfire risk in the Gyeongbuk region was assessed as high during the period from March 19 to 20, prior to the outbreak of simultaneous wildfires. This suggests that “wildfire prevention information” messages should have been transmitted before wildfire ignition. However, the analysis revealed that few “wildfire prevention information” messages were disseminated prior to the outbreak of the simultaneous wildfires. Instead, EAM containing wildfire prevention guidelines were transmitted after wildfires had already occurred. While a relatively large number of evacuation guidance messages were sent in Uisong and Andong, which were the areas initially affected by wildfires, the frequency of evacuation guidance messages was markedly lower in Chongsong, Yeongyang, and Yeongdeok, where wildfires spread rapidly at later stages. Consequently, these later-affected areas experienced higher numbers of human casualties than Uisong, where the wildfire initially ignited. This pattern suggests that spatiotemporal prediction of wildfire spread and corresponding preparedness were insufficient. However, this study did not directly analyze the timing of EAM transmission in conjunction with detailed wildfire spread data. As a result, it is not possible to determine conclusively whether EAM transmissions were appropriately timed relative to wildfire progression. Future research should therefore examine EAM transmission timing in direct relation to wildfire spread dynamics to more accurately assess the effectiveness of alert dissemination.

In addition, further research is needed to examine interactions between public alert messages delivered via mobile devices and other forms of communication, such as written, verbal, and visual alerts [12]. As demonstrated by this case study, questions remain regarding whether EAM alone can adequately support local residents’ responses during wildfire events. EAM are often constrained by character limits, which may result in brief messages that lack sufficient detail [9]. Because the public may not be familiar with the risks associated with large-scale wildfires, it is essential to include clear and actionable protective guidance in alert messages [13]. Alert messages linked to maps or supplemented by detailed action manuals may serve as effective alternatives.

During the large-scale wildfires of March 2025, initial evacuation plans were not adequately established due to the scale and complexity of the disaster. Levels of planning and preparedness can either hinder or facilitate decision making and action during emergencies [10]. Therefore, careful examination is required to develop customized evacuation strategies for communities at risk of wildfires [5]. In Korea, local governments are responsible for issuing EAM related to wildfire evacuation. If local governments lack sufficient manpower or technical infrastructure, they may be unable deliver timely and effective EAM. Accordingly, future research should examine factors that influence EAM transmission capacity, including decision-making structures, institutional authority, and the availability of predictive technologies and supporting infrastructure.

In response to the simultaneous wildfires of March 2025, the Korean government introduced a three-step evacuation instruction framework consisting of preparation, standby, and departure phases. This framework is based on wildfire spread prediction models, in which risk zones are divided into five-hour and eight-hour zones according to an estimated spread speed of 8.2 km/h and a maximum instantaneous wind speed of 27.6 m/s. Residents are instructed to proceed through evacuation preparation, pre-evacuation, and immediate evacuation depending on risk levels. These instructions resemble evacuation policies in the United States (e.g., Ready, Set, Go!) and Australia (e.g., Stay and Defend or Leave Early) in their emphasis on preparedness for resident evacuation. However, they also differ in important respects. In particular, the Korean framework lacks detailed programs related to housing survivability and property protection. Even the relatively well-developed evacuation policies of the United States and Australia have been criticized for shortcomings in countermeasures, technologies, and equipment necessary for effective evacuation [14]. Accordingly, additional guidance for EAM transmission should be developed for each evacuation stage. In particular, clear guidelines related to housing and property protection should be incorporated into alert messaging.

Wildfires disproportionately affect certain populations [3]. For example, at least one-third of all evacuees from wildfires that occur annually in Canada are indigenous people [3]. Similarly, 26 of the 28 deaths resulting from the wildfires in March 2025 involved elderly individuals aged 60 years or older. This study did not investigate whether elderly individuals received, understood, and responded to EAM. In general, the elderly have significantly lower information accessibility than the general adult population due to lower cellphone utilization rate, degraded visual and auditory function, and lower levels of literacy. Previous studies [12,13] have shown that message content and style, receiver characteristics, and contextual factors are critical determinants of public response. Receiver characteristics can be classified into social environment, social ties, socioeconomic status, psychological factors, educational level, and ages. With respect to elderly populations, Park, K. (2023) [15] compared service elements required by elderly users with those required by the general public to improve disaster text messaging services. The analysis revealed that rapid transmission of disaster text messages was important for both groups. The elderly, however, placed greater emphasis on message content, whereas the general public prioritized functional aspects such as system quality and transmission standards. The most effective method to deliver risk-related information through short text-based alert messages on mobile devices remains unclear. Nevertheless, it is evident that message content should be communicated clearly and in terms that are easily understandable for both elderly individuals and the general public.

Meanwhile, further research is required on the optimal frequency of public alert messages delivered through mobile devices. According to Rehman et al. (2020) [9], the effectiveness of text messages decreases when the same sender transmits multiple messages repeatedly, as emergency alerts may be overlooked. In addition, receivers may misunderstand or ignore information conveyed via text messages during emergency situations. By contrast, Bean et al. (2015) [13] emphasized that conveying a sense of urgency is essential to prompting public response and that repetition is a commonly used strategy to achieve this. In Korea, elderly populations are particularly vulnerable to wildfires due to the pronounced aging of rural communities. Although the most effective method regarding the appropriate timing, frequency, and cycle of EAM transmission has yet to be determined, these parameters, as well as message content, should be designed with careful consideration of receiver characteristics. Accordingly, further empirical research is required to identify effective EAM strategies tailored to diverse population groups.