Biggs Juliet, Anantrasirichai Nantheera, Albino Fabien, Lazecky Milan, Maghsoudi Yasser
COMET, School of Earth Sciences, University of Bristol, Bristol, UK.
Visual Information Laboratory, Department of Computer Sciences, University of Bristol, Bristol, UK.
Bull Volcanol. 2022;84(12):100. doi: 10.1007/s00445-022-01608-x. Epub 2022 Nov 3.
Radar (SAR) satellites systematically acquire imagery that can be used for volcano monitoring, characterising magmatic systems and potentially forecasting eruptions on a global scale. However, exploiting the large dataset is limited by the need for manual inspection, meaning timely dissemination of information is challenging. Here we automatically process ~ 600,000 images of > 1000 volcanoes acquired by the Sentinel-1 satellite in a 5-year period (2015-2020) and use the dataset to demonstrate the applicability and limitations of machine learning for detecting deformation signals. Of the 16 volcanoes flagged most often, 5 experienced eruptions, 6 showed slow deformation, 2 had non-volcanic deformation and 3 had atmospheric artefacts. The detection threshold for the whole dataset is 5.9 cm, equivalent to a rate of 1.2 cm/year over the 5-year study period. We then use the large testing dataset to explore the effects of atmospheric conditions, land cover and signal characteristics on detectability and find that the performance of the machine learning algorithm is primarily limited by the quality of the available data, with poor coherence and slow signals being particularly challenging. The expanding dataset of systematically acquired, processed and flagged images will enable the quantitative analysis of volcanic monitoring signals on an unprecedented scale, but tailored processing will be needed for routine monitoring applications.
The online version contains supplementary material available at 10.1007/s00445-022-01608-x.
合成孔径雷达（SAR）卫星系统地获取图像，这些图像可用于火山监测、岩浆系统特征描述以及在全球范围内潜在的火山喷发预测。然而，对大量数据集的利用受到人工检查需求的限制，这意味着信息的及时传播具有挑战性。在此，我们自动处理了哨兵 -1 卫星在 5 年期间（2015 - 2020 年）获取的超过 1000 座火山的约 600,000 张图像，并使用该数据集来证明机器学习在检测变形信号方面的适用性和局限性。在标记最为频繁的 16 座火山中，5 座经历了火山喷发，6 座显示出缓慢变形，2 座有非火山变形，3 座存在大气伪影。整个数据集的检测阈值为 5.9 厘米，相当于在 5 年研究期内每年 1.2 厘米的速率。然后，我们使用大型测试数据集来探索大气条件、土地覆盖和信号特征对可检测性的影响，并发现机器学习算法的性能主要受可用数据质量的限制，相干性差和信号缓慢尤其具有挑战性。系统获取、处理和标记的图像数据集不断扩大，将能够以前所未有的规模对火山监测信号进行定量分析，但常规监测应用需要进行定制处理。
在线版本包含可在 10.1007/s00445-022-01608-x 获取的补充材料。
