Long-term variations in thermal data and stage segmentation of activity
Long-term plots of eruptive activity constructed using 6177 nighttime AHI images acquired between 1 June and 31 August are shown in Fig. 4a–c. Although most of the images examined here had no cloud over the volcano (June to August is in the middle of the dry season of Java), a few minor cloudy periods, suggested by low T11Mx values being below the background level, can be recognized (Fig. 4c). Here, we identify long-term variations in this eruptive activity by focusing on the maximum values measured during each day in order to minimize or avoid the influence of the plume or the subpixel size clouds located over the eruptive center, as well as cases where the eruptive center is located on the boundary between two pixels (or occasionally four pixels).
This period of eruptive activity was divided into four stages based on this time-series analysis (Fig. 4d). Two major pulses were recognized in this pattern; the first pulse occurred from 20 June to 1 August (Pulse 1) and the second pulse occurred from 1 to 13 August (Pulse 2). Based on their trends, both pulses were subdivided into the Increasing, Peak and Decreasing Periods. Furthermore, in R2.3Mx (Fig. 4b), a low-level thermal anomaly was observed from 14 to 19 June (Precursory Stage), which occurred prior to the start of Pulse 1, and another low-level thermal anomaly was observed from 14 to 20 August (Terminal Stage), which occurred immediately after the end of Pulse 2.
In R1.6Mx, Pulses 1 and 2 exhibited uniformly higher levels of thermal anomalies than the Precursory and Terminal stages; during the latter stages, their levels remained at background values. This indicates that Pulses 1 and 2 were caused by the introduction of a persistent high-temperature heat source, as is suggested by the relationship shown in Fig. 3; thus, these pulses are interpreted to have resulted from the lava that represents the main product of this activity. The effusion rate of lava is believed to have changed to exhibit a consecutive two-pulse pattern (Fig. 4e), as is suggested by the variations in its thermal anomalies. Thus, the formations of the lower and upper lava groups of the lava bed (Fig. 2d) most likely correspond to these two pulses.
We assume that the thermal anomaly at 1.6 μm basically reflects variations in the effusion rate of lava. This is consistent with the results of seismic observations reported by the Pusat Vulkanologi dan Mitigasi Bencana Geologi (2016), which are based on real-time seismic amplitude measurement (RSAM) data (Fig. 4g). The RSAM counts are believed to be related to the transport of magma to the surface (Endo and Murray 1991). The RSAM counts here exhibit a consecutive two-pulse pattern that is similar to the variations observed in R1.6Mx (Fig. 4a). The good agreement between these two types of observations supports the supposition presented above.
Variations during each stage
Here we describe the variations and characteristics of each stage based on the combined analysis performed using both long- and short-term variations.
Precursory stage/14–19 June (6 days)
In R2.3Mx (Fig. 4b), a low-level thermal anomaly was recorded from 14 to 19 June. During this period, no distinct activity, such as lava effusion or ash emission, was reported. Prior to magma effusion, magma may have risen to a shallow level in the conduit and released hot gas, or heated the surrounding groundwater, which may have caused this slightly increased thermal anomaly, as was observed before previous eruptions, such as the 1986 Izu-Oshima (Kagiyama and Tsuji 1987), 1999 Shishaldin (Dehn et al. 2002) and 2004 Mt Asama (Kaneko et al. 2006) eruptions. This thermal anomaly can thus be interpreted to represent a precursor to the major eruptive phase, which occurred during Pulse 1.
Pulse 1/20 June–1 August (43 days)
Pulse 1-increasing period/20–28 June (9 days) On 20 June, the levels of thermal anomalies in R1.6Mx, R2.3Mx and T3.9Mx rose discontinuously from low or background levels and then increased gradually until 28 June (Fig. 4a–c). This gradual increase is interpreted to reflect an increase in the rate of lava effusion.
The time of the onset of lava effusion was not identified using ground-based observations. Figure 5a shows the short-term variations in thermal anomalies from 18 to 20 June using both daytime and nighttime data. During the daytime, each index exhibited an arch-shaped pattern reflecting the reflected sun light from 18 to 19 June. On 20 June, the patterns of R1.6Mx and R2.3Mx were the same as they had been on the previous two days, but that of T3.9Mx where the effect of the reflected sunlight was relatively small, began to increase from the middle of the day at approximately 6:50 UTC (13:50 local time), thus reflecting the start of lava effusion. The effusion continued thereafter, as observed in the nighttime of the same day.
Pulse 1-peak period/29 June–24 July (26 days) High levels of thermal anomalies, including those in R1.6Mx, were continuously recorded from 29 June to 24 July (Fig. 4a–c). This suggests that high levels of lava effusion continued, which likely formed the majority of the lava bed (Fig. 2b, d). Long-term minor fluctuations in these variations are believed to reflect long-term fluctuations in the effusion rate of this lava. In the short-term plots, indexes of thermal anomalies were constant and nearly flat, which suggests that the effusion of lava was stationary (e.g., the nighttime of 20 June in Fig. 5a).
Pulse 1-decreasing period/25 July–1 August (8 days) Between 25 July and 1 August, the levels of these indexes decreased and finally reached very low levels, which suggests that the effusion rate of lava also decreased to a low level (Fig. 4a–c). During the last day of this stage (1 August), a small thermal pulse occurred several hours before the start of Pulse 2 (Fig. 5b), which can be interpreted to represent a precursory event to the start of Pulse 2 (a small explosive event?).
Pulse 2/1–13 August (~ 12 days)
Pulse 2-increasing period/1–2 August (~ 1 day) The thermal anomaly rapidly increased on 1 August (Fig. 4a–c). On the short-term plot (Fig. 5b), this rapid increase started at 20:20 UTC and continued until approximately 14:00–15:00 UTC on the next day.
Pulse 2-peak period/2–3 August (~ 2 days) After reaching a climax at approximately 15:00–16:00 UTC on 2 August (Fig. 5b), the anomalies in R1.6Mx, R2.3Mx and T3.9Mx quickly began to decrease, such that the Increasing and Peak Periods of Pulse 2 were shorter than those of Pulse 1 (Fig. 4a–c).
Pulse 2-decreasing period/4–13 August (10 days) The Decreasing Period of Pulse 2 continued for 10 days. The upper line of the hatched region during the middle of this period shows the interpolated level during the small depression that was caused by cloud cover (Fig. 4a–c). During the last few days of this period, these indexes decreased to low levels and finally discontinuously dropped on 14 August, which marks the point when the effusion of lava finally ceased.
Terminal stage/14–20 August (7 days)
This stage was characterized by a low-level thermal anomaly that was only clearly visible on R2.3Mx (Fig. 4b). As lava effusion cannot be recognized in the photographs taken on 14 and 15 August (Pfeiffer 2015), this thermal anomaly likely reflected the small-scale gas emissions at the cone, from which repeated small-scale eruptions occurred. Finally, the anomaly in R2.3Mx decreased to its background levels on 21 August, which signaled the end of the June–August 2015 Mt Raung activity.