Hierarchical classification of snowmelt episodes in the Pyrenees using seismic data

In recent years the analysis of the variations of seismic background signal recorded in temporal deployments of seismic stations near river channels has proved to be a useful tool to monitor river flow, even for modest discharges. The objective of this work is to apply seismic methods to the characterization of the snowmelt process in the Pyrenees, by developing an innovative approach based on the hierarchical classification of the daily spectrograms. The CANF seismic broad-band station, part of the Geodyn facility in the Laboratorio Subterráneo de Canfranc (LSC), is located in an underground tunnel in the Central Pyrenees, at about 400 m of the Aragón River channel, hence providing an excellent opportunity to explore the possibilities of the seismic monitoring of hydrological events at long term scale. We focus here on the identification and analysis of seismic signals generated by variations in river discharge due to snow melting during a period of six years (2011–2016). During snowmelt episodes, the temporal variations of the discharge at the drainage river result in seismic signals with specific characteristics allowing their discrimination from other sources of background vibrations. We have developed a methodology that use seismic data to monitor the time occurrence and properties of the thawing stages. The proposed method is based on the use of hierarchical clustering techniques to classify the daily seismic spectra according to their similarity. This allows us to discriminate up to four different types of episodes, evidencing changes in the duration and intensity of the melting process which in turn depends on variations in the meteorological and hydrological conditions. The analysis of six years of continuous seismic data from this innovative procedure shows that seismic data can be used to monitor snowmelt on long-term time scale and hence contribute to climate change studies.近年来，对河道附近临时布设地震台站所记录的地震背景信号变化进行的分析，已被证明是监测河道流量的有效工具，即便在流量较小时亦能发挥作用。本研究的目标是将地震方法应用于比利牛斯山脉的融雪过程特征刻画，通过构建一种基于每日频谱分层分类的创新方法。隶属于坎弗朗克地下实验室（Laboratorio Subterráneo de Canfranc, LSC）Geodyn设施的CANF宽频地震台站，坐落于比利牛斯山脉中部的一条地下隧道内，距离阿拉贡河河道约400米，这为长期尺度下的水文事件地震监测研究提供了绝佳契机。本文聚焦于2011至2016年这六年期间，由融雪导致的河道流量变化所引发的地震信号的识别与分析。在融雪事件期间，河道流量的时间变化会产生具备特定特征的地震信号，可将其与其他背景振动源区分开来。我们已开发出一套利用地震数据监测解冻阶段发生时间与特征的方法。所提出的方法基于层次聚类技术，依据相似性对每日地震频谱进行分类。该方法可区分多达四种不同类型的事件，揭示融雪过程的持续时长与强度变化，而这些变化又取决于气象与水文条件的波动。对六年连续地震数据采用该创新流程进行分析的结果表明，地震数据可用于长期尺度下的融雪监测，从而为气候变化研究提供助力。