A
Title: Volcanic warnings based on appropriate evaluations of volcanic activities by the Japan Meteorological AgencyAkimichiTakagi1,2✉Emailtakagi@met.kishou.go.jp
HitoshiYamasato2Emailhyamasat@nifty.com
1Seismology and Volcanology DepartmentMeteorological Agency3-6-9 Toranomon, Minato-ku105-8431TokyoJapan, Japan
2CTI Engineering Co., LtdLtd, 3-21-1 Nihombashi Hamacho, Chuo-Ku103-8430TokyoJapan
Author #1: Akimichi Takagi, Seismology and Volcanology Department, Japan Meteorological Agency, 3-6-9 Toranomon, Minato-ku, Tokyo 105–8431, Japan, takagi@met.kishou.go.jp
Author #2: Hitoshi Yamasato, CTI Engineering Co., Ltd, 3-21-1 Nihombashi Hamacho, Chuo-Ku, Tokyo 103–8430 Japan, hyamasat@nifty.com
Indicate the corresponding author: Akimichi Takagi
Abstract
Volcanic alert levels (VALs) are used around the world to inform people of volcanic activity. To mitigate volcanic disasters, the Japan Meteorological Agency (JMA) has gradually introduced Volcanic Warning (VW) schemes for 111 active volcanoes and VAL schemes for 49 volcanoes under continuous monitoring since 2007. VALs are devised to ensure that VWs can be smoothly utilized in disaster prevention measures. VALs describe "target areas", i.e., areas that pose a danger to life, and "recommended action to be taken for disaster management organizations and residents" in five levels depending on volcanic activity. The JMA sets individual VALs for each volcano in advance, judges the VAL based on current/ongoing volcanic activity, then raises, maintains, or lowers the VAL before climbers and residents can be affected by an eruption. However, the criteria for judging the current VAL are based on observations and analyses of volcanic activity, and are updated for each volcano based on evolving knowledge of its activity. At Asamayama, one of the most active volcanoes in Japan, criteria were set based on knowledge of precursory phenomena observed during historical eruptive activity, but small eruptions in 2019 occurred without any of the typical precursors, and the VAL was not raised from 1 to 2 before the eruption. We analyzed observational data and determined that eruptions occurred in association with the volcanic conduit becoming partially choked, and the VAL criteria were updated in 2024 to evaluate such phenomena. We also statistically analyzed the occurrence of eruptions following the issuance of warnings by calculating a threat score (TS), an index of forecast reliability, for each volcano with a VAL scheme in Japan. Asamayama had relatively high TS values (0.40), and the average of TS for all volcanoes was calculated to be 0.20, consistent with prior results for volcanoes around the world.
Keywords
Volcanic warnings
Volcanic alert level
Japan Meteorological Agency
Asamayama
Conceptual model of volcanic activity
Threat score
Main Text
1 Introduction
To mitigate volcanic disasters, information on the state of volcanic activity must be disseminated to residents and agencies in a timely and accurate manner. To achieve this, precursory volcanic phenomena must be understood and constantly quantified through routine observation. In Japan, the Japan Meteorological Agency (JMA) issues Volcanic Warnings (VWs) and Volcanic Alert Levels (VALs; 1–5 from lowest to highest alert level in Japan), and local governments take disaster prevention measures such as evacuating residents accordingly.
Similar systems are used around the world to inform the public of volcanic activity. For example, the 1991 eruption of Pinatubo (Philippines) prompted the organization of VAL schemes (Winson et al., 2014), which are currently in operation for six Philippine volcanoes. Currently, 11 nations inform residents of volcanic activity using VALs or Color Codes (Winson et al., 2014). In other countries such as Russia (Kamchatka), the Aviation Color Code, which issues volcanic activity alerts to aircrafts, is used as a substitute for informing the public. In Japan, the VW/VAL scheme was introduced in 2007 (Yamasato et al., 2013). Table 1 shows major nations that have implemented VAL schemes for residents.
Although VALs are in operation at many volcanoes in many nations, the decision to raise, maintain, or lower levels is always based on historical observational data. In Japan, following the 2014 Ontakesan eruption, the criteria for deciding VALs have been carefully reviewed and published*.
*VAL criteria (Japanese only): https://www.data.jma.go.jp/vois/data/filing/level_kijunn/keikailevelkijunn.html
Even if VAL criteria are based solely on historical observational data, it is desirable that the evolving volcanic phenomena have been modeled, even if only by qualitatively. This qualitatively modeling refers to a model of the progression of volcanic activity based on empirical rules rather than by precise numerical calculations based on physical models. A truly objective issuance of a VAL must refer to criteria that are based on empirical relationships defined based on observational data, on the current status of volcanic activity, and on the progression of activity predicted numerically, as is the case for weather forecasting. However, currently we are doing our best to create qualitatively modeling of the volcanic progression.
Ideally, a VAL could be issued with all necessary modeling completed, but available historical observational data are not sufficient to provide inputs for such modeling at all volcanoes. Furthermore, volcanic activity does not necessarily progress according to assumed hypotheses or progression models. Nonetheless, VAL criteria should be improved as much as possible based on modeling of the volcanic progression.
In this paper, we first introduce Japan's VW/VAL system and provide an overview of VAL criteria. We then provide an example of setting VAL criteria at Asamayama volcano, for which we have a strong understanding of volcanic phenomena based on abundant historical observational data. Finally, we conclude with a statistical overview of VW issuances and eruption occurrences in Japan.
2 Volcanic Warnings and Volcanic Alert Levels
The VW scheme was introduced in 2007 and is currently put into operation for all 111 active volcanoes in Japan, and VAL schemes have been gradually introduced since then. As of March 2025, VALs have been implemented for 53 craters at 49 volcanoes (Fig. 1).
Fig. 1
The 111 active volcanoes and 49 volcanoes for which Volcanic Alert Levels have been implemented in Japan.
The JMA issues VWs and Volcanic Forecasts (VFs) based on observational data, monitoring and evaluation to mitigate the effects of potential disasters for 111 active volcanoes in Japan (Yamasato et al., 2013) (Fig. 1). The issuance of VWs and VFs, like Weather Warnings, is based on the legal basis of the Meteorological Services Act*. They are also transmitted by the JMA to local governments and other relevant agencies using the same means as Weather Warnings. When extremely hazardous volcanic phenomena (e.g., ballistic projectiles, pyroclastic flows, mudflows caused by snowmelt) or the expansion of affected areas are expected, VWs are issued for areas where extremely hazardous volcanic phenomena are expected as “target areas” (Japan Meteorological Agency, 2025).
* The Meteorological Services Act (Japanese only): https://laws.e-gov.go.jp/law/327AC0000000165
For residential areas, “Volcanic Warnings (for residential areas)” (a.k.a. Residential Area Warnings) are issued, which are classified as Emergency Warnings. For non-residential areas near or around the crater, “Volcanic Warnings (for areas near the crater)” (a.k.a. Near-crater Warnings) are issued.
The designation of residential areas and areas near the crater is described in literature such as the Regional Disaster Prevention Plans formulated by local governments. VWs are issued to inform residents: current volcanic activity, such as the number of volcanic tremors, is described in a Details of Volcanic Activity that is issued subsequently (Japan Meteorological Agency, 2025a). VFs are issued when VWs are lifted.
VALs were devised to ensure that the issuance of VWs and VFs could be smoothly utilized in disaster prevention measures. VALs describe “target areas” that pose a risk to life and provide “recommended action to be taken for disaster management organizations and residents” in five levels according to the current and expected volcanic activity (Japan Meteorological Agency, 2025a) (Fig. 2).
Fig. 2
Volcanic Warnings and Volcanic Forecasts for Japanese volcanoes for which VALs have been implemented (Japan Meteorological Agency, 2025b).
The VAL System is run by Volcanic Disaster Management Councils (VDMCs), which include representatives of prefectures, municipalities, meteorological observatories, erosion and sediment control departments, volcanologists, and others, as part of the Regional Disaster Prevention Plan. Under normal circumstances, a VDMC produces eruption scenarios comprising expected progressions of volcanic activity, creates hazard maps, and decides evacuation advisory areas and the timing of evacuation notices in advance. Based on these, even in the event of an eruption crisis, regional authorities responsible for evacuations, the JMA as the monitoring agency, and volcanologists will work closely together to share a common understanding regarding local disaster mitigation measures. This framework allows regional authorities to smoothly implement emergency measures and residents to take swift and appropriate action according to the VAL. This type of framework for creating VALs in strong collaboration with municipalities and other organizations is not widely used globally, but is a major feature of regional volcanic disaster prevention measures in Japan.
For example, at Level 3, the target area includes non-residential areas near the crater, the keyword is set as “restriction on proximity to the volcano”, and residents are instructed to “stand by and pay attention to changes in volcanic activity” and prepare “for the evacuation of the elderly and other persons requiring special care, depending on the situation” (Fig. 2). Based on interviews and observations conducted at the five U.S. Geological Survey volcano observatories, Fearnley and Beaven (2018) argued that supplementary communication techniques are crucial to facilitate specific responses of key users of the VAL. In Japan, VDMCs carefully discusses and decides on eruption scenarios, hazard maps, and VAL in advance. Therefore, this issue has been largely resolved by adding keywords and “recommended actions to be taken for disaster management organizations and residents” to the VALs.
In Japan, in the event of a small eruption or the possibility of an eruption, a VW will be issued and the VAL will be raised to Level 2 or higher. In other words, if it is judged that an eruption is not imminent, the VAL remains at Level 1, and Level 1 activity is not further subdivided. In contrast, in alert level schemes beyond Japan, the issuance of an alert level above the minimum level does not necessarily mean that an eruption is imminent. For example, in the Philippines, the Philippine Institute of Volcanology and Seismology (PHIVOLCS) operates alert levels for six volcanoes*; for three of those (Hibok-Hibok, Taal, and Pinatubo), the alert level may be raised from the lowest level of 0 to 1 (on a scale from 0 to 5) when increased volcanic activity is not considered to be related to an imminent eruption. Therefore, although a change in alert level is not necessarily linked to the issuance of VWs in other systems, the VAL scheme used in Japan (Levels 1–5) requires that VWs indicating an imminent eruption be issued at Level 2 and above for all active volcanoes.
In many countries around the world, volcanologists decide to change VALs practically (Papale, 2017), whereas in Japan, monitoring officials decide and issue VALs using the criteria for deciding VALs, described in Section 3.
“Target areas” of VALs are not determined simply by the scale of the potential eruption, but instead vary among volcanoes depending on the location of the crater relative to residential areas and the volcanic hazards assumed in the eruption scenario. Nonetheless, if an eruption emits ejecta locally around the crater, it could still put the lives of climbers at risk; accordingly, the target area at VAL 2 is set “around the crater”. Indeed, in Japan, the areas around the craters of active volcanoes are often tourist spots and hiking trails, so it is essential to issue VWs, even in the case of small eruptions where the target area is only around the crater. Even if the eruption is small, the increased possibility of eruption must be reliably judged and a VW must be issued appropriately in the case that the VAL is raised above 1. So far, many more VALs of 2 have been issued than VALs exceeding 2, which would affect residential areas (VAL2: 95 times, VAL3: 37 times, VAL4: 2 times, VAL5: 2 times between December 2007 and February 2025).
* Volcano Alert Levels in operation by PHIVOLCS:
https://www.phivolcs.dost.gov.ph/volcano-alert-levels/
3 Criteria for deciding Volcanic Alert Levels
The JMA sets individual VALs for each volcano in advance, judges the VAL based on current volcanic activity, and raises, maintains, or lowers the VAL accordingly before climbers and residents are affected. Criteria for judging the current VAL are based on observations and analyses of volcanic activity.
In Japan, VAL criteria have only been established for volcanoes where VALs have been implemented: 53 craters at 49 active volcanoes as of March 2025. During the 2014 Ontakesan eruption that claimed 63 lives, the scale of the precursor phenomena was smaller than during the previous eruption in 2007, so the VAL was not raised in that case. Based on this experience and recommendations from the government's Council of Experts, all VAL criteria were carefully re-examined and were gradually revised and published from 2016 onwards. VALs are raised and lowered strictly in accordance with these criteria.
For volcanoes that have erupted numerous times historically, such as Sakurajima, VAL criteria are set quantitatively based on the results of volcanic observations, such as an increase in earthquakes or ground deformation before an eruption. In contrast, many volcanoes have not yet been observed erupting, and their VAL criteria cannot be established in the same manner.
This difficulty is not unique to VAL designations in Japan but is the same for alert level or color code designations at volcanoes around the world. For example, as of March 2025, the PHIVOLCS alert level scheme has been implemented for six volcanoes*: Pinatubo, Taal, Mayon, Kanlaon, Hibok-Hibok, and Bulusan. Those alert levels are determined based on seismic activity (number of occurrences of each type of waveforms and hypocenter), volcanic gas emissions (emission rates, concentration ratios), ground deformation, water temperature, pH, gravity anomaly changes, fumarolic activity, thermal activity (intensity of volcanic glow), and eruptive phenomena (e.g., heights of eruptive plumes). However, quantitative criteria have only been set for Pinatubo and Taal, where numerous eruptions have been observed; alert levels at the other four volcanoes are based only on qualitative criteria.
Hereafter in this section, we present how JMA configures specific VAL criteria for various cases of active volcanoes.
For some volcanoes, we have a strong understanding of volcanic activity stemming from numerous research results, including qualitative conceptual models of volcanic activity (see Section 4). In such cases, it may be possible to interpret the observed activity, which is reassuring when setting VAL criteria (Fig. 3a).
Fig. 3
Rationales for setting the criteria to decide the VALs used by the JMA at the 53 craters at 49 volcanoes for which VAL schemes have been implemented.
In contrast, at approximately half of the volcanoes where VAL schemes have been implemented, no historical eruptions have been observed (Fig. 3b). For volcanoes that have erupted at least once historically, empirical rules are estimated based on past observational data, and the VAL criteria are configured quantitatively such that a similar eruption cannot occur without a VW being issued (Fig. 3c). Depending on the observations available, criteria are configured using quantitative data such as numbers of earthquakes, ground deformation, and volcanic gas emission rates, as well as qualitative observations of unusual phenomena such as volcanic glow, volcanic tremor, very-long-period earthquakes, and felt earthquakes. For example, if a volcano has had three historical eruptive episodes that correspond to VAL 2, and the maximum daily number of earthquakes occurring immediately prior to those eruptions was 60, 70, and 30, the criteria for implementing VAL 2 for that volcano would be set at 30 earthquakes in a day for the sake of safety. Notably, though, if there had also been a case at the same volcano where 40 earthquakes had occurred in a single day without an accompanying eruption, the VAL would be raised to 2 if ≥ 30 earthquakes occurred in a single day, yet it remains possible that an eruption will not occur.
For volcanoes with no historically observed eruptions, VAL criteria are sometimes configured by reference to similar other volcanoes for which eruptions have been observed (Fig. 3e). For example, “similar” volcanoes are considered to undergo similar eruptive activities in terms of volcanic geology. At Midagahara and Nikko-Shiranesan, where no eruptions have been observed in the past, VAL criteria are set based on the observation results of Ontakesan, which is also an andesitic volcano.
If it is difficult to configure VAL criteria using any of the above methods, the most abnormal value observed during historical periods of volcanic unrest is used as the criterion. For example, if the maximum daily number of earthquakes ever observed at a certain volcano was 100, that value would be set as the criterion for raising the alert level to VAL 2 (Fig. 3d). Again, it remains possible that an eruption will not occur. Criteria exceeding VAL 2 are set by reference to other similar volcanoes.
A
Table 2 shows the relatively simple example of the VALs criteria for Adatarayama volcano in northeastern Japan; no eruptions have been observed since the geophysical monitoring began at this volcano.* Volcano Alert Levels in operation by PHIVOLCS: https://www.phivolcs.dost.gov.ph/volcano-alert-levels/
4 Setting Criteria for deciding the Volcanic Alert Level for Asamayama
Asamayama volcano (central Japan) is one of the most active andesitic volcanoes in Japan. The volcano experienced two large eruptions in 1108 and 1783, causing devastating damage (Yasui and Koyaguchi, 2004). Japan's first volcanic observatory was established at Asamayama in 1911, and a wealth of observational data have been accumulated. Since then, it has been observed that Vulcanian explosive eruptions are typically preceded by a gradual increase in B-type volcanic earthquakes that occur relatively shallow (Minakami, 1960), increasing volcanic gas emissions (Ohwada et al., 2013), and smoke rising from the crater. Recent smaller eruptions have occurred without such typical precursory phenomena being observed. Nonetheless, based on data obtained using the dense, multi-component observation network around the crater, the detailed mechanisms leading to eruptions are becoming clearer (e.g., Takeo et al., 2022; Yoshigai et al., 2023). Therefore, for the safety of climbers and tourists, it has become necessary to elaborate the criteria for raising the VAL in advance, even for such small eruptions.
In this section, we introduce the method for setting VAL criteria at Asamayama based on qualitative conceptual models for these two eruption scenarios, with and without typical precursory phenomena. Although VALs range from 1 to 5, we here discuss only the criteria for which the first VW would be issued, i.e., raising the VAL from 1 to 2 as a period of quiescence ends.
4.1 Setting criteria in the case of Typical Volcanic Unrest
Based on data accumulated over the years, various precursor phenomena are known to correspond to increased volcanic activity at Asamayama. Based on geodetic observations over the past few decades, inflation of the ground surface on the western foot of Asamayama has been correlated with the occurrence of A-type (volcano-tectonic) earthquakes several months prior to eruptive activity. Increased steam plume volumes, volcanic gas emissions, and crater-bottom temperatures have also been observed to precede eruptive activity. Most eruptions since 2000 were preceded by this typical volcanic unrest (TVU) (Fig. 4). Of these, the eruptions in 2004 and 2009 were Vulcanian eruptions, producing tens of thousands of tons of ejecta. There have also been cases when similar TVU was observed but did not result in an eruption (2020, 2023). Nonetheless, this trend of activity is a typical pattern preceding Vulcanian eruptions caused by magmatic activity at Asamayama volcano (Fig. 5a).
Fig. 4
Time series (2000–2024) of (a) eruption occurrences, volcanic glow occurrences, and plume heights; (b) volcanic gas emissions; (c) seismicity; and (d) ground deformation at Asamayama volcano. Vertical hatched rectangles indicate periods of typical volcanic unrest (TVU).
Fig. 5
Volcanic phenomena leading to eruptions equivalent to VAL 2 at Asamayama: (a) typical volcanic unrest (TVU) and (b) choked conduit unrest (CCU).
The occurrence of this TVU is explained as follows. An artificial seismic survey detected the presence of a dike intrusion chamber (DIC) > 1 km below sea level and approximately 5 km west of Asamayama (Aoki et al., 2009), and seismic interferometry results have indicated that a magma chamber exists directly below the DIC at 5–10 km below sea level (Nagaoka et al., 2012). As magma begins to recharge the DIC, ground deformation is observed at the surface due to the inflation of the DIC. Additionally, A-type earthquakes associated with the rupture begin to occur frequently around the DIC and within the volcanic conduit (Nakada et al., 2005). Degassing of the magma within the DIC results in increased sulfur dioxide emissions (Kazahaya et al., 2015). Furthermore, as magma rises beneath the volcano, volcanic earthquakes are caused not only by rupture but also by exsolved volatile, such as B-type volcanic earthquakes, tremors, long-period vibrations, and T-type earthquakes that is waveform with low attenuation (e.g., Minakami 1960; Maeda et al. 2019; Takeo et al. 2022). In addition, volcanic glow, indicating high temperatures at bottom of the crater, is frequently observed.
Based on the above research findings and observations, a schematic diagram of this TVU is shown in Fig. 6a. An eruption scenario based on this TVU was used to set the criteria for issuing VAL 2 (Fig. 7, left). Furthermore, if activity increases within this eruption scenario, the VAL will be increased further.
Fig. 6
Schematic conceptual models of (a) typical volcanic unrest (TVU) and (b) choked conduit unrest (CCU) eruption scenarios at Asamayama.
Fig. 7
Criteria for deciding to raise the VAL to 2 and or lower it to 1 for cases of TVU and CCU.
4.2 Setting criteria in the case of Choked Conduit Unrest
From 2000 to 2018, eruptions were observed at Asamayama on 42 days. During this period, five episodes of TVU (2003, 2004, 2009, 2015, and 2016; Figs. 4, 5a) were all accompanied by eruptions. However, on 7 and 25 August 2019, smaller eruptions occurred without any TVU. Therefore, it was not possible to raise the VAL from 1 to 2 (i.e., to issue a VW) in advance using the conventional criteria.
Although there was no obvious increase in the number of earthquakes preceding these eruptions, more detailed seismic analysis revealed clear precursor activities. JMA classifies B-type earthquakes occurring at shallow depths within Asamayama volcano into lower-frequency (predominant frequency < 3 Hz; BL-type), and higher-frequency types (predominant frequency > 3 Hz; BH-type). In July 2019, about a month before the first smaller eruption, BL-type earthquakes occurred around 30 times per day. The occurrence of BL-type earthquakes then rapidly decreased to 0–6 times per day five days before the eruption on 7 August and remained low until the day before the eruption. In contrast, BH-type earthquakes occurred infrequently (around 6 times per day) in July 2019, then more frequently (> 10 times per day) seven days before the eruption (“I” in Fig. 8). After the eruption, the number of BL-type earthquakes returned to previous levels and the number of BH-type earthquakes gradually decreased. Similar seismic activity was observed prior to the second eruption on 25 August (“II” in Fig. 8). These precursor phenomena were observed approximately one week before both eruptions. However, similar seismic activity in September did not result in an eruption.
Fig. 8
Daily number of earthquakes before and after the August 2019 Asamayama eruptions. (Top) BL-type earthquakes. (Bottom) BH-type earthquakes. BL-type earthquakes rapidly decreased, and BH-type earthquakes rapidly increased within a week before the eruptions on 7 and 25 August, and in September (orange shaded rectangles). Red triangles indicate eruption dates.
In addition, infrared thermal observations by the University of Tokyo showed that the temperature at the bottom of the crater dropped rapidly 11 days before the eruption. The temperature remained low until just before the eruption, and then rose rapidly during the eruption (Yoshigai et al., 2023).
Sakurajima, an andesitic volcano similar to Asamayama that also undergoes repeated Vulcanian eruptions, also experiences BH-type and BL-type earthquakes. By comparing ground deformation and surface phenomena, Ishihara and Iguchi (1989) showed that BH-type earthquakes occur frequently due to increased internal pressure when the volcanic conduit becomes choked, whereas BL-type earthquakes occur frequently when pressure decreases due to the release of volcanic ash and gas. They also revealed that BL-type earthquakes occur at shallower depths than BH-type earthquakes. In the case of Asamayama, the depth difference between the hypocenters of these two B-type earthquakes is unclear, but if we assume that a similar phenomenon is occurring at Asamayama, the phenomena preceding the two eruptions in 2019 could be explained by choking of the shallow conduit.
In other words, when even small amounts of magma are supplied to the DIC, volatiles are released from the DIC, move through the conduit, and reach the crater, producing BL-type earthquakes, which are thought to be caused by volatiles and occur around shallow volcanic conduits. If part of the shallow conduit becomes choked, the reduced supply of volatiles is expected to suppress the occurrence of BL-type earthquakes at shallow depths. At the same time, increasing internal pressure around the conduit due to small amounts of volatiles rising from the DIC increases the occurrence of BH-type earthquakes. Furthermore, when the conduit is choked, high-temperature volatiles cannot reach the crater bottom, and the temperature at the crater bottom drops. Figure 6b shows a schematic qualitative conceptual model of these phenomena. We note that no change in the sulfur dioxide emission rate was observed due to the overall small emissions during this period; if observations had been made at higher sensitivity, a decrease in the emission rate might have been detected. This series of activities differs from the TVU at Asamayama, and is considered to be a rare phenomenon that occurs only when the magma supply is small. Hereafter, to contrast with TVU, we refer to this activity as choked conduit unrest (CCU).
Based on the above qualitative conceptual model, we set new criteria for designating the case of a choked conduit based on seismic activity (a sudden decrease in BL-type earthquakes and an increase in BH-type earthquakes) using the following procedure. Because the background levels of earthquake occurrences change over the long term, we define the “current” seismic activity as the average number of earthquakes per day over a three-day (72-h) period (AND3), and the background seismic activity as the average number of earthquakes per day over a 30-day period (AND30). Based on these measures of seismic activity, we define the rapid change index (RCI) as:
where HNj is the hourly number of earthquakes j hours before the present, and DNi is the daily number of earthquakes i days before the present.
We define RCI separately for BL- and BH-type earthquakes as RCIBL and RCIBH, respectively. By using these two RCIs, we attempt to detect sudden changes in seismic activity that may suggest that the conduit is choked. We note that a sufficient supply of volatiles is necessary to generate a pressure increase when the conduit is choked; i.e., background BL-type seismicity must be relatively high. To capture this, we set the prerequisite that the average number of BL-type earthquakes per day over a 30-day period must be at least 25.
To detect the precursory activity preceding the eruptions on 7 and 25 August, appropriate RCIBL and RCIBH thresholds were determined by trial and error. We found that precursory activity could be detected before both eruptions when RCIBL decreases to at most 0.1 and RCIBH increases to at least 1.7 (Fig. 9). In summary, the criteria for detecting precursory eruption activity in the case of CCU are that the following three sub-criteria are met simultaneously:
Fig. 9
Criteria for raising the VAL to 2 based on the 2019 seismic activity at Asamayama. (a) The period when each of the three sub-criteria (AND30BL, RCIBL, RCIBH) were met (horizontal black lines), the days when all three sub-criteria were met simultaneously (vertical blue lines, meeting criteria for VAL 2), the 30-day validity period after issuance of VAL 2 (light blue horizontal lines), and the dates of the eruptions (red triangles) during July to November 2019. (b, c) Daily number (gray bars), 30-day averaged daily number (blue line), 3-day averaged daily number (light blue line), and rapid change index (RCI, dark orange line) for (b) BL-type and (c) BH-type earthquakes. Periods in which a sudden decrease in BL-type earthquakes were accompanied by a sudden increase in BH-type earthquakes are marked by vertical orange shaded rectangles. The dashed horizontal lines delimit the thresholds for the RCI criteria for BL-type (RCIBL ≤ 0.1) and BH-type earthquakes (RCIBH ≥ 1.7).
The criteria for lowering the VAL from 2 to 1 was set approximately one month after the last observation of any phenomenon that met the criteria for raising the VAL to 2. The reason for this extended alert period is the same as for other eruption scenarios: to avoid frequent raising and lowering of VALs in a short period of time. This alert period was also set so that no eruptions since the 1960s would have occurred at VAL 1. However, if an eruption does not occur after raising the VAL to 2 based on these new criteria, the level will be lowered from 2 to 1 around two weeks after it is assumed that the choked conduit has been resolved (RCIBH ≤ 1). These values were set as the criteria for VAL 2 for Asamayama during CCU (Fig. 7, right).
According to these new criteria, the VAL could have been raised to 2 at 1900 JST on 4 August 2019, 75 h before the eruption. In addition, volcanic activity fell below the criteria for VAL 2 at 0500 JST on 8 August, but in accordance with the stand-down procedures (Fig. 7), VAL 2 would have been maintained for the next 30 days. Due to the stand-down procedures alone, the 25 August eruption would have occurred during the ongoing VAL 2 period. For reference, the RCIBL and RCIBH sub-criteria for raising the level were met 18 h before the eruption on 25 August, whereas the AND30BL sub-criterion was not. Moreover, in September (“III” in Fig. 8), when there was a similar sudden change but no accompanying eruption, the AND30BL sub-criterion was again not met.
This process of volcanic activity when the magma supply is small and the conduit is choked has been added as one of the eruption scenarios and the criteria for VAL 2 have been revised (Fig. 7, right).
5 Statistical surveys of Volcanic Warnings and eruption occurrence
Setting the criteria for raising the VAL for Asamayama from 1 to 2 (i.e., issuing a Volcanic Warning), as described in the previous section, is a quantitative evaluation of volcanic activity. The 2009 Asamayama magmatic eruption occurred after the VW was issued (hereafter called a “Hit”). This successful issuance resulted from accumulated knowledge of past volcanic activities. However, not all eruptions will occur after a VW is issued; there may be cases where no eruption occurs after a VW is issued (hereafter, a “Fault”) or, conversely, where eruptions occur before a VW is issued (hereafter, a “Miss”). In the case of volcanoes with little historical volcanic activity, the accuracy of VW issuance is likely to be lower.
In this section, we review past issuances of VWs, which raise Level from 1 to 2 or higher, from the implementation of the VAL scheme on 1 December 2007 until 31 March 2025 and statistically survey the results of the issuance (Hits, Faults, and Misses). This review includes the 53 craters at 49 volcanoes for which VALs are currently maintained, covering the period from December 1, 2007, when the scheme of VAL was first implemented, to March 31, 2025. The dataset includes 69 volcanic unrest periods for 21 craters at 18 volcanoes.
To examine the results (Hit, Fault, or Miss) for all periods of unrest, we calculated a Threat Score (TS) to evaluate the appropriateness of VW issuances for each volcano. An index of forecast reliability that is also used to verify the results of weather forecasts, TS is expressed as follows based on the numbers of Hits, Faults, and Misses:
When applied to periods of unrest at Asamayama (Table 3), we obtain TS = 0.4 (contingency table shown in Fig. 10). Individual TS values for all volcanic unrest periods during the span of the study for each volcano are shown in Table 4. Of the 69 VWs issued, 14 were Hits (20%), 45 were Faults (65%), and 10 were Misses (15%). Among the volcanoes that have experienced five or more unrest periods since December 2007, Asamayama (0.40) and Asosan (0.27) had relatively high TS values. The average TS value among all the volcanoes is 0.20.
Fig. 10
Contingency table of the issuance of VWs and eruption occurrence at Asamayama based on Table 3.
Winson et al. (2014) examined the timings of the issuance of volcanic alert levels for 194 eruptions at 60 volcanoes around the world between 1990 and 2013. They similarly found that 19% of the VALs issued for events that ended in eruption accurately reflected the hazards before the eruption.
This analysis method based on the number of Hits, Faults, and Misses per VW issuance is appropriate for evaluating eruptions at certain volcanoes, but a different approach is required to evaluate more active volcanoes. For example, since the VW at Sakurajima was issued, the VAL has not been decreased to 1 because eruption activity has continued for a long period. Therefore, we also calculated the percentage of days of Hits, Faults, and Misses relative to days for which VWs were issued (Table 5). The combined TS value for all volcanoes during the study period was 0.14, markedly lower than the average TS calculated per VW issuance. For more active volcanoes, TS values are relatively higher, such as 0.155 (980/6331) for Suwanosejima, 0.296 (584/1973) for Asosan, and 0.594 (3761/6331) for Sakurajima. In contrast, for less active volcanoes with infrequent eruptions, TS values are relatively lower, such as 0.002 (2/1006) for Kirishimayama (Ioyama) and 0.002 (6/3933) for Satsuma-Iojima. Thus, there is a positive correlation between number of eruption days and daily average TS (Fig. 11). This correlation may reflect that VWs are issued appropriately for active volcanoes with a wealth of eruptions, and that active volcanoes are more likely to enter another unrest period before the previous VW is lifted. In contrast, for volcanoes that have not been very active historically, even if activity decreases, the decision to lift the VW is made cautiously and may be overly delayed.
Fig. 11
Relationship between the number of days an eruption occurred and Threat Scores at various volcanoes in Japan. The figure is a log–log plot based on Table 5. There is a positive correlation between the two.
6 Discussion
Quantitative evaluation of all combined aspects of volcanic activity is not yet possible; it is even more difficult to quantitatively indicate the possibility of an eruption from the results of any evaluation attempting to do so. In recent years, the Volcanic Unrest Index (VUI) was developed to compile multiple volcanic phenomena into a worksheet and semi-quantitatively evaluate the intensity of activity (Potter et al., 2015). However, like the VAL schemes presented herein, the VUI is derived by comparison with a given volcano's past activity levels or those at similar volcanoes. An alternative approach to quantitatively and automatically indicate the states of volcanoes used multivariate observational data based on Bayesian statistics (e.g., Cannavò et al., 2017).
Although the criteria for deciding VALs have been gradually upgraded around the world based on accumulated knowledge, such upgrades should include not only observational data (including numbers of eruptions and unrest periods experienced), but also models that utilize the observational data. Such models, even if only qualitative, should be prepared for many volcanoes and updated as more knowledge comes to light. Future research should develop more sophisticated quantitative indicators for evaluating volcanic activity.
7 Summary
In Japan, the Volcanic Warning (VW)/Volcanic Alert Level (VAL) scheme was put into operation for active volcanoes in 2007. VALs describe target areas and recommended action to be taken for disaster management organizations and residents in five levels depending on volcanic activity, such that the VAL can be smoothly utilized in disaster prevention measures. When the VAL is raised above 1, it is accompanied by the issuance of a VW due to the possibility of an eruption.
The JMA sets individual VAL schemes for each volcano in advance, judges the current VAL based on current/ongoing activity, and raises, maintains, or lowers the VAL as appropriate. Criteria for deciding the VAL are set based on the results of observations and analyses of volcanic activity. For volcanoes that have erupted numerous times historically, these criteria are set quantitatively based on the results of volcanic observations. For the most active volcanoes of which we have extensive knowledge and numerous research results have been produced, there are even qualitative conceptual models of volcanic activity.
As an example of how the criteria for VALs are set, we examined the case of Asamayama volcano. Most eruptions at Asamayama since 2000 have been preceded by typical volcanic phenomena. A conceptual model has been developed based on this knowledge and VAL criteria have been set accordingly. However, the conventional criteria could not be applied to smaller eruptions of Asamayama in 2019. Therefore, we developed a new conceptual model hypothesizing that the cause of these smaller eruptions was choked conduit conditions, and we described new VAL criteria for raising the level to 2 (and, therefore, issuing a VW) based on seismicity.
For context, we surveyed results (Hits, VWs issued prior to an eruption; Faults, VWs issued but no eruption ensued; or Misses, eruptions occurring without a VW being issued) of all volcanic unrest periods in Japan since the VAL scheme was implemented in December 2007. We calculated a Threat Score (TS) to evaluate the appropriateness of raising the VAL from 1 to 2 or higher for each volcano for which a VAL scheme has been developed. Of 69 unrest periods, 14 VWs were Hits (20%), 45 were Faults (65%), and 10 were Misses (15%). Among the volcanoes that have experienced five or more unrest periods since December 2007, Asamayama and Asosan had relatively high TS values. The average TS for all volcanoes was calculated to be 0.20, consistent with the results of an earlier analysis of the issuance of volcanic alert levels and eruptions at 60 volcanoes around the world between 1990 and 2013 (Winson et al., 2014).
For more active volcanoes in Japan, we calculated the percentages of days of Hits, Faults, and Misses relative to the number of days for which VWs were issued. Across all volcanoes, the average TS in this case was lower at 0.14. However, the TS values were positively correlated with the number of eruption days, and the most active volcanoes scored the best.
Criteria for deciding VALs have been gradually upgraded around the world based on accumulated knowledge. We hope that future resurveys would improve the TS value. Furthermore, it will be necessary to compare the TS values of volcanoes around the world using this method.
Abbreviations
CCU Choked conduit unrest
DIC Dike intrusion chamber
JMA Japan Meteorological Agency
PHIVOLCS Philippine Institute of Volcanology and Seismology
RCI Rapid change index
TS Threat score
TVU Typical volcanic unrest
VAL Volcanic alert level
VDMC Volcanic Disaster Management Councils
VF Volcanic forecast
VUI Volcanic Unrest Index
VW Volcanic warning
Supplementary Information
Additional file containing earthquake frequency data and analysis results for setting new VAL criteria to apply to the 2019 eruption of Mt. Asama.
Acknowledgements
We are grateful to Y. Yoshigai for discussion of the qualitative conceptual model of Asamayama. We thank the Japan Meteorological Agency for allowing us to use various observational data and issuance documents for Volcanic Warnings and Volcanic Alert Levels.
A
Author contributions
TA designed this study, collected volcanic information documents and drafted the manuscript with significant contributions from YH. YH conducted the data analysis to set criteria. All the authors have read and approved the final manuscript.
A
Funding
This study was supported by JSPS KAKENHI Grant Number JP23K03512.
A
Availability of data and materials
Almost data needed to evaluate setting criteria (section 4.2) in the current study are present in the Supplementary Materials. Volcanic information documents, which written by Japanese only, are available here: https://www.data.jma.go.jp/vois/data/report/volinfo/volinfo.php
A
DeclarationsEthics approval and consent to participate
Not applicable
Consent for publication
Not applicable
Competing interests
The authors declare that they have no competing interests.
Author details
1 Seismology and Volcanology Department, Japan Meteorological Agency, 3-6-9 Toranomon Minato-ku, Tokyo 105–8431, Japan. 2 CTI Engineering Co., Ltd, 3-21-1 Nihombashi Hamacho, Chuo-Ku, Tokyo 103–8430, Japan.
Tables and Figures
A
Table 1 Major nations that have implemented VAL schemes for residents. The number of volcanoes for which VALs are implemented, the number of levels in each VAL, the organization with the authority to issue VALs, and their websites as of 31 March 2025 are provided.Table 2 An example of criteria for deciding VALs at Adatarayama volcano, revised on March 14, 2024
https://www.data.jma.go.jp/vois/data/filing/level_kijunn/214_level_kijunn.pdf (only Japanese)
A
Table 3 Volcanic unrest periods at Asamayama between December 2007 and March 2025 and the success of issued VWs.A
Table 4 Threat scores (TS) for all volcanic unrest periods at all continuously monitored volcanoes in Japan for which VAL schemes have been implemented between December 2007 and March 2025.A
Table 5 Threat scores (TS) calculated based on the number of days with Hits, Faults, and Misses at all continuously monitored volcanoes in Japan for which VAL schemes have been implemented between December 2007 and March 2025.Electronic Supplementary Material
Below is the link to the electronic supplementary material
Supplementary Material 1
References
Aoki Y, Takeo M, Aoyama H, Fujimatsu J, Matsumoto S, Miyamachi H, Nakamichi H, Ohkura T, Ohminato T, Oikawa J, Tanada R, Tsutsui T, Yamamoto K, Yamamoto M, Yamasato H, Yamawaki T (2009) P-wave velocity structure beneath Asama Volcano, Japan, inferred from active source seismic experiment. J Volcanol Geotherm Res 187:272–277. https://doi.org/10.1016/j.jvolgeores.2009.09.004
Cannavò F, Cannata A, Cassisi C, Giuseppe DG, Montalto P, Prestifilippo M, Privitera E, Coltelli M, Gambino S (2017) A multivariate probabilistic graphical model for real-time volcano monitoring on Mount Etna. J Geophys Res Solid Earth 122:3480–3496. https://doi.org/10.1002/2016JB013512
Fearnley CJ, Beaven S (2018) Volcano alert level systems: managing the challenges of effective volcanic crisis communication. Bull Volcanol 80:46. https://doi.org/10.1007/s00445-018-1219-z
Ishihara K, Iguchi M (1989) THE RELATIONSHIP BETWEEN MICRO-EARTHQUAKE SWARM AND VOLCANIC ACTIVITY AT SAKURAJIMA VOLCANO. Annuals Disas Prev Res Inst Kyoto Univ 32B-1:1–11. https://repository.kulib.kyoto-u.ac.jp/dspace/handle/2433/72177 (in Japanese, with English abstract and captions)
Japan Meteorological Agency (2025a) JMA’s Volcanic Disaster Mitigation Efforts. VOLCANOES 6. https://www.jma.go.jp/jma/kishou/books/kazan/kazan.pdf
Japan Meteorological Agency (2025b) Volcanic Warnings and Volcanic Forecasts for volcanoes where Volcanic Alert Levels are in effect. VOLCANOES 7. https://www.jma.go.jp/jma/kishou/books/kazan/kazan.pdf
Kazahaya R, Aoki Y, Shinohara H (2015) Budget of shallow magma plumbing system at Asama Volcano, Japan, revealed by ground deformation and volcanic gas studies. J Geophys Res Solid Earth 120:2961–2973. https://doi.org/10.1002/2014JB011715
Maeda Y, Takeo M, Kazahaya R (2019) Comparison of high- and low-frequency signal sources for very-long-period seismic events at Asama volcano, Japan. Geophys J Int 217:389–404. https://doi.org/10.1093/gji/ggz021
Minakami T (1960) Fundamental research for predicting volcanic eruptions: Part 1. Bull Earth Res Inst 38:497–544. https://doi.org/10.15083/0000033840
Nagaoka Y, Nishida K, Aoki Y, Takeo M, Ohminato T (2012) Seismic imaging of magma chamber beneath an active volcano. Earth Planet Sci Lett 333–334:1–8. https://doi.org/10.1016/j.epsl.2012.03.034
Nakada S, Yoshimoto M, Koyama E, Tsuji H, Urabe T (2005) Comparative Study of the 2004 Eruption with Old Eruptions at Asama Volcano and the Activity Evaluation. Bull Volcanol Soc Japan 50:303–313. https://doi.org/10.18940/kazan.50.5_303(in Japanese, with English abstract and captions)
Ohwada M, Kazahaya K, Mori T, Kazahaya R, Hirabayashi J, Miyashita M, Onizawa S, Mori T (2013) Sulfur dioxide emissions related to volcanic activity at Asama volcano, Japan. Bull Volcanol 75:775. https://doi.org/10.1007/s00445-013-0775-5
Papale P (2017) Rational volcanic hazard forecasts and the use of volcanic alert levels. J Appl Volcanol 6:13. https://doi.org/10.1186/s13617-017-0064-7
Potter SH, Scott BJ, Jolly GE, Neall VR, Johnston DM (2015) Introducing the Volcanic Unrest Index (VUI): a tool to quantify and communicate the intensity of volcanic unrest. Bull Volcanol 77:77. https://doi.org/10.1007/s00445-015-0957-4
Takeo M, Aoki Y, Koyama T (2022) Recent volcanic activity at the Asama volcano and long-period seismic signals. Proc Jpn Acad Ser B 98:416–438. https://doi.org/10.2183/pjab.98.022
Winson A, Costa F, Newhall C, Woo G (2014) An analysis of the issuance of volcanic alert levels during volcanic crises. J Appl Volcanol 3:14. http://www.appliedvolc.com/content/3/1/14
Yamasato H, Funasaki J, Takagi Y, Technical Note of the National Research Institute for Earth (2013) The Japan Meteorological Agency’s Volcanic Disaster Mitigation Initiatives. Sci Disaster Prev 380:101–107. https://dil-opac.bosai.go.jp/publication/nied_tech_note/pdf/n380_E.pdf
Yasui M, Koyaguchi K (2004) Sequence and eruptive style of the 1783 eruption of Asama Volcano, central Japan: a case study of an andesitic explosive eruption generating fountain-fed lava flow, pumice fall, scoria flow and forming a cone. Bull Volcanol 66:243–262. https://doi.org/10.1007/s00445-003-0308-8
Yoshigai Y, Morita Y, Kobayashi T, Ozawa D, Matsusue S, Kawaguchi R, Ohminato T (2023) Reexamination of Volcanic Seismicity Before and After the 2019 Eruption of Asama Volcano, Detected by the Matched Filter Method. Bull Volcanol Soc Japan 68:229–244. https://doi.org/10.18940/kazan.68.4_229(in Japanese, with English abstract and captions)
Nation | volcanoes | Volcanic alert levels | Organization with the authority to issue volcanic alert levels | Citation |
|---|---|---|---|---|
Colombia | Green/Yellow/Orange/Red | Instituto Colombiano de Minería y Geología | https://www2.sgc.gov.co/Noticias/Paginas/Asi-son-los-colores-de-los-volcanes-colombianos.aspx | |
Iceland | Green/Yellow/Orange/Red | Icelandic Meteorological Office | https://en.vedur.is/volcanoes/fagradalsfjall-eruption/hazard-map/ | |
Indonesia | 68 | I ~ IV | Center for Volcanology & Geological Hazard Mitigation (PVMBG) | https://magma.esdm.go.id/v1/edukasi/tingkat-aktivitas-gunung-api |
Italy | 5 | Green/Yellow/Orange/Red | Civil Protection Department | https://rischi.protezionecivile.gov.it/en/volcanic/activities/ |
Japan | 49 | 1 ~ 5 | Japan Meteorological Agency (JMA) | https://www.jma.go.jp/jma/kishou/know/kazan/English/level.html |
Mexico | Green/Yellow/Red | Centro Nacional de Prevencion de Desastres (CENAPRED) | https://www.gob.mx/cenapred | |
New Zealand | 12 | 0 ~ 5 | Institute of Geological and Nuclear Sciences Limited (GNS Science) | https://www.geonet.org.nz/about/volcano/val |
Philippines | 6 | 0 ~ 5 | Philippine Institute of Volcanology and Seismology (PHIVOLCS) | https://www.phivolcs.dost.gov.ph/index.php/volcano-hazard/volcano-alert-level |
Russian Federation (Kamchatskaya) | 66 | Green/Yellow/Orange/Red | Kamchatka Volcanic Eruption Response Team (KVERT) | http://kvert.febras.net/ |
Spain (Canary Islands) | ≥ 10 | Green/Yellow/Orange/Red | Special Civil Protection and Emergency Response Plan for Volcanic Risk (PEVOLCA) | https://www.gobiernodecanarias.org/infovolcanlapalma/semaforo/ |
United States of America | 57 | Green/Yellow/Orange/Red | US Geological Survey | https://volcanoes.usgs.gov/activity/alertsystem/index.php |
Vanuatu | 7 | 0 ~ 4 | Vanuatu Meteorology and Geohazards Department | https://www.vmgd.gov.vu/geohazards/volcanoes |
Level | Step-up Criteria | Step-down Criteria |
|---|---|---|
5 | An eruption that would cause serious damage to residential areas has occurred or is imminent Volcanic blocks or mudflows caused by melting snow (during snow accumulation) have reached residential areas or are imminent. | If the phenomenon on the left is no longer observed and a decrease in volcanic activity is confirmed |
4 | Possibility of an eruption causing serious damage to residential areas Amid increased eruptive activity, phenomena suggesting the rising of magma, such as an increase in large earthquakes or ground deformation, have been observed. | If the phenomenon on the left is no longer observed and a decrease in volcanic activity is confirmed |
3 | An eruption occurs that would have a significant impact on nearby residential areas (within around 2.5 km of the crater) Large volcanic blocks were observed flying from more than 1 km to up to 2.5 km from the crater. Possibility of an eruption causing a significant impact on nearby residential areas (within around 2.5 km of the crater) If any of the following phenomena are observed in addition to the occurrence of a phenomenon that meets the criteria for Level 2: • Further rapidly increase in volcanic earthquakes or tremors, and increase in their scale (amplitude) • Significant ground deformations indicating inflation of the volcanic edifice (greater than the criteria for VAL 2) • Increased eruption activity (above level 2) • pyroclastic flows or surges | If the phenomenon on the left is no longer observed and there is no trend of increased volcanic activity |
2 | An eruption occurs that would affect the area around the crater (within around 1 km of the crater) Large volcanic blocks were observed flying within around 1 km of the crater. Possibility of an eruption that would affect the area around the crater (within around 1 km of the crater) If multiple of the following phenomena are observed: • Increase in volcanic earthquakes (≥ 30 per 24 hours) However, epicenter and depth must be taken into consideration • Increase in low-frequency earthquakes or occurrence of volcanic tremors (excluding minor ones) • Certain ground indicating inflation of the volcanic edifice • Increased thermal activity, including active fumarolic activity, expansion of geothermal areas, and a significant rise in ground temperature | If none of the phenomena on the left are observed, or if there is a certain tendency for seismic activity to return to the state before the unrest, and there is no tendency for ground deformation, fumarolic activity, or thermal activity to increase. In addition, even if the level is lowered to 1 because there is a certain tendency for volcanic activity to return to the state before the unrest, if volcanic activity subsequently becomes active again, the level will be raised back to 2 even if the left criteria are not met. |
Unrest NO. | Date when Level was raised to 2 or higher | Date when Level was lowered to 1 | Eruption | Date of the first eruption or missed eruption |
|---|---|---|---|---|
1 | August 8, 2008 | April 15, 2010 | Hit | August 10, 2008 |
2 | June 11, 2015 | August 30, 2018 | Hit | June 16, 2015 |
3 | August 7, 2019, 22:30 | November 6, 2019 | Miss | August 7, 2019, 22:08 |
4 | June 25, 2020 | February 5, 2021 | Fault | (No eruption) |
5 | March 23, 2021 | August 6, 2021 | Fault | (No eruption) |
Volcano | Hit | Fault | Miss | TS |
|---|---|---|---|---|
Meakandake | 1 | 3 | 0.250 | |
Tokachidake | 1 | 0.000 | ||
Zaozan | 2 | 0.000 | ||
Azumayama | 3 | 0.000 | ||
Kusatsu-Shiranesan (Yugama) | 3 | 0.000 | ||
Kusatsu-Shiranesan (MotoShirane) | 1 | 0.000 | ||
Asamayama | 2 | 2 | 1 | 0.400 |
Yakedake | 1 | 0.000 | ||
Ontakesan | 1 | 1 | 0.000 | |
Hakoneyama | 1 | 1 | 0.500 | |
Miyakejima | 1 | 1.000 | ||
Ioto | 1 | 1.000 | ||
Tsurumidake and Garandake | 1 | 0.000 | ||
Asozan | 3 | 6 | 2 | 0.273 |
Kirishimayama (Ioyama) | 1 | 5 | 0.167 | |
Kirishimayama (Shinmoedake) | 2 | 5 | 2 | 0.222 |
Kirishimayama (Ohachi) | 1 | 0.000 | ||
Sakurajima | 1 | 1.000 | ||
Satsuma-Iojima | 3 | 2 | 0.000 | |
Kuchinoerabujima | 7 | 1 | 0.000 | |
Suwanosejima | 1 | 1.000 | ||
Total / Average | 14 | 45 | 10 | 0.203 |
Volcano | Hit (days) | Fault (days) | Miss (days) | TS |
|---|---|---|---|---|
Meakandake | 3 | 282 | 0 | 0.011 |
Tokachidake | 0 | 71 | 0 | 0.000 |
Zaozan | 0 | 101 | 0 | 0.000 |
Azumayama | 0 | 937 | 0 | 0.000 |
Kusatsu-Shiranesan (Yugama) | 0 | 2,162 | 0 | 0.000 |
Kusatsu-Shiranesan (MotoShirane) | 0 | 438 | 1 | 0.000 |
Asamayama | 16 | 2,972 | 1 | 0.005 |
Yakedake | 0 | 96 | 0 | 0.000 |
Ontakesan | 13 | 1,243 | 1 | 0.010 |
Hakoneyama | 3 | 338 | 0 | 0.009 |
Miyakejima | 10 | 2,734 | 0 | 0.004 |
Ioto | 183 | 2,561 | 0.067 | |
Tsurumidake and Garandake | 0 | 20 | 0 | 0.000 |
Asozan | 584 | 1,387 | 2 | 0.296 |
Kirishimayama (Ioyama) | 2 | 1,004 | 0 | 0.002 |
Kirishimayama (Shinmoedake) | 79 | 3,798 | 2 | 0.020 |
Kirishimayama (Ohachi) | 0 | 35 | 0 | 0.000 |
Sakurajima | 3,761 | 2,570 | 0.594 | |
Satsuma-Iojima | 6 | 3,925 | 2 | 0.002 |
Kuchinoerabujima | 106 | 3,827 | 1 | 0.027 |
Suwanosejima | 980 | 5,351 | 0.155 | |
Total / Average | 5,746 | 35,852 | 10 | 0.138 |