週遭噪訊是因地球上大氣擾動和人類活動交互作用下所產生的。其本質為非線性、隨機的而且非恆定的。我們提出一個新的方法，希望能利用排列熵(permutation entropy, PE)監測這些隨機的噪訊以即時預測火山噴發。排列熵(permutation entropy, PE)是由噪訊中的隨機性所得到的非線性的統計量數。我們研究了從1996年9月29日到 10月13日在冰島Gjálp爆發的火山。波形是以暫時的HOTSPOT震測的網絡系統所記錄。Bárðarbunga地震噴發時的震度為5.6，且造成火山口的環斷層破裂。位於120公里外的監測站，在地震的8.57天前和噴發前10.76天，可從排列熵(PE)可看到噪訊波場的隨機性變化。我們也計算了噪訊的主要頻率(dominant frequency, DF)和中心頻率(centroid frequency, CF)，而其結果顯示噪訊是源於自然。主要頻率大約是2Hz，在Bárðarbunga地震發生的六天前，中心頻率的範圍從0.2~4.8Hz不等，而地震後中心頻率的數值和主要頻率相近。偏振分析指出，超長週期(very long period)的震顫並非隨機性變化的原因。隨機性變化的原因很可能是因高壓地殼在上地幔隆起降低散射的結果。中心頻率和主要頻率的地殼異質性變化一致，而較高頻率的散射也降低。因此，我們認為，在所有火山的環境噪聲中PE的變化，很可能是在特定的高壓地區，導致其地殼散射特性的改變。未來，此方法應用於即時監測與預測火山爆發有很大的潛力。;Ambient seismic noise is generated by the interaction of atmospheric disturbances and human activities with the solid Earth. Its properties are nonlinear, stochastic and non-stationary. We propose a new approach to real – time forecasting of volcanic eruptions by monitoring the stochastic properties of ambient noise using permutation entropy (PE), which is a nonlinear statistical measure of the stochasticity contained within the noise. We studied the 1996 Gjálp eruption in Iceland, which lasted from 29th Sept - 13th Oct 1996. Waveforms recorded by the temporary HOTSPOT seismic network were used. The eruption commenced with the Mw = 5.6 Bárðarbunga earthquake that initiated a ring fault failure along the Bárðarbunga caldera. PE captured changes in the stochasticity of the noise wavefield (at stations located up to 120 km away) 8.57 days before the Bárðarbunga earthquake and 10.76 days before the onset of the eruption. We also calculated the dominant frequency (DF) and centroid frequency (CF) of the ambient noise, whose results suggested that the noise was due to natural sources. DF was ~ 0.2 Hz while the CF ranged between ~ 0.2 to 4.8 Hz up until ~ 6 days before the Bárðarbunga earthquake, thereafter it became similar to DF. Polarization analysis determined that very long period tremor (< 1 Hz) was not the cause of the changes in the stochasticity. Changes in the stochasticity were most likely the result of reduced scattering brought on by the strongly pressurized crust due to doming in the upper mantle. CF coincides with DF as the crustal heterogeneity changes and the higher frequency scattering is reduced. Therefore, we argue that PE variations in the ambient noise, for any given volcanic regime, are likely localized to the areas undergoing strong pressurization, which alters the scattering properties of the crust. This methodology has great potential in future applications of real-time monitoring and forecasting of volcanic eruptions.
