Development of the muographic tephra deposit monitoring system

Sep 9, 2020

Remote online-monitoring-system

The current real-time tephra monitoring system consisted of four main elements. These were (a) the SMOS located at the observation site including the readout module, (b) the DAQ computer, (c) the server to register, manage and visualize these data, and (d) a web browser installed on a remote computer. The component (c) was the core element of the current monitoring system. It was operated on a dedicated internet cloud service with the following specifications: (a) 6 cores of a central processing unit (CPU), (b) 20 gigabytes (GBs) of a memory, and (c) 800 GB of hard disk drive (HDD) space for processing and visualizing the SMOS data in real time. The operating system used was a 64-bit Canonical Ubuntu Linux 16.04LTS (Xenial Zerus). The database constructed in the server consisted of four kinds of data: (1) event data, (2) experimental condition data, (3) statistical information data, and (4) the Muogram Library. The event data were the compressed data generated by the SMOS readout module. The experimental condition data were configured to the actual setup of the SMOS (the distance between upstream and downstream detectors, number of layers used for analysis, and width of scintillator strips) at the observation site (SMO). Statistical information data that included hourly and daily event rate plots for each layer were automatically generated from the event data by incorporating the experimental condition data. These data could be used for monitoring the PMT’s condition of the detector. The event and experimental condition data were transferred to the ‘Muogram Generator’ to generate the track distribution on a φ-θ plane as a two-dimensional matrix that represents the number of muon tracks as a function of the elevation and azimuth angles) every 10 min automatically in the background. The generated track distribution was subsequently registered to the ‘Muogram Library’. These 10-min muograms could be aggregated later by choosing the specific period and angular region for an offline analysis. The time sequential plots of the muon count rate could be manually generated on this online monitoring system by setting the data aggregation period. In these plots, the volcanic column heights were automatically incorporated and plotted on these time sequential plots so that the comparison between the muon count rate and the volcanic activities could be easily compared. The information pertaining to the volcanic column height was automatically downloaded from the Japan Meteorological Agency (JMA) website28 and stored in the database. The rendering software visualized the muograms and time sequential plots enabled users to download the generated images.

By combining the cosmic muon’s energy spectrum20 and muon’s range-energy relationship21, the transmitted muon flux, Φ (X, θ, ϕ), can be derived. If the average density (ρ) of the target object is given, the densimetric rock thickness, X, can be defined by ρx, where x is the geometrical rock thickness. Once X is given, the minimum energy, Ec, of muons that can escape from the target object can be calculated by using the range-energy relationship. Φ (X, θ, ϕ) can be then derived by integrating the muon’s energy spectrum I(E, θ) over the energy range between Ec and infinity.

$$Phi left( {theta ,phi } right) = mathop smallint limits_{{E_{c} left( {theta ,phi } right)}}^{infty } Ileft( {E, theta } right)dE$$

(3)

Since the observation apparatus has an angular resolution (Δθ, Δϕ), the transmitted muon flux measured in one image pixel is an integration of the flux of the muons that had passed through different regions in the target object. Therefore, the muon’s path length, x, in the target object varies as a function of arriving angles (θ, ϕ), and thus Eq. (3) must be calculated for different θ and ϕ, and averaged over the angle range between θ ± Δθ and ϕ ± Δϕ to derive the averaged transmitted flux, < Φ > , that can be directly compared with the observed flux in the image pixel. Inversely, if < Φ > is given, the muographically averaged densimetric thickness (MADT), < X > , can be uniquely determined by inserting < Φ > into Eq. (3) to derive averaged cutoff energy, < Ec > , that is directly compared with the energy-range relationship. Since the MADT is the column density averaged over the angle range (θ ± Δθ and ϕ ± Δϕ), if the distance between the detector and the target object (D) holds the following conditions:

$$D gg DDelta theta ,$$

(4-1)

$$D gg DDelta phi ,$$

(4-2)

D2ΔθΔϕ < X > gives the total mass of the volume within the angle range, θ ± Δθ and ϕ ± Δϕ, where L is the typical thickness of the target object. The muographically averaged geometric thickness (MAGT) is defined by < X > / ρ. The MAGT is therefore different from the arithmetically averaged rock thickness.

Atmospheric effects in the muographic measurements

(A) Rainfall effects. The annual rainfall in Kagoshima city, Japan is 2.3 × 103 mm. The typical rainfall speed is 10 m/s. Therefore, the density increase in the atmosphere due to rainfalls, being averaged over 1 year is 7 × 10−3 g/m3, which is far below the detection limit (~ 10 kg/ m3) of the current observation. (B) Volcanic ash effects. The deposit area was at least 2 × 104 m2; hence average tephra falls of 2.15 × 105 kg/m2 in 720 days. Ash-fall speed is at an order of 10 m/s. Therefore, the density increase in the atmosphere due to ash falls, being averaged over 720 days is 0.3 g/m3 that is also far below the detection limit of the current observation. This estimate can be deviated by an order of magnitude, but it doesn’t affect the consequence of the current work.

Normalization of the data

In this work, muons arriving from the direction above the ridge of the mountain (open-sky muons) were used for normalizing the count rate of the 5 angular regions (NM, VAB, VSAB, MAR, and sky). The time-sequential plots of muon counts collected from each angular region were respectively divided by the time-sequential plot of the open-sky muons and compared. The seasonal variations of the near-horizontal open-sky muon flux are generally small because the low-energy muons decay before arriving the sea level. However, the muon counts were slightly modulated due to the temporal variations of the detector condition including a gain drift of the photomultiplier tubes during a long period of operation. The current normalization technique cancels this effect. Figure 8 shows the normalization result for the muon data within the angular region that covers the entire mountain. The maximum likelihood fitting result (the horizontal line in Fig. 8) was R = 10−4t + 5.4167, where R is the normalized muon ransmission rate and t is time in units of day, meaning that R at the beginning of the observation and R at the end of the observation varied by less than 0.2%. Also, no significant variations were found near the dormant region (NM) as can be seen in Fig. 4A.