Statistical and Non-linear Dynamics Methods of Earthquake Forecast: Application in the Caucasus

Chélidzé, T.; Melikadze, Giorgi; Kiria, Tengiz; Jimsheladze, Tamar; Kobzev, Gennady

doi:10.3389/feart.2020.00194

Cited by 7 publications

(6 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a result, hidden non-linear structures were discovered in seismic data. These characteristics vary with time, which is in contradiction with the memory-less purely Poissonian approach (Chelidze et al, 2020). The analysis of temporal variations in the complexity of seismic measures, namely, the phase space portrait, can be used for forecasting strong earthquakes (Chelidze et al, 2018).…”

Section: Complexity Analysis and Machinementioning

confidence: 99%

“…The ML is often used in such diverse fields, such as medicine, geophysics, disaster management, and business (Witten et al, 2017). Applications of ML techniques in the last years have become extremely widespread in solving various problems of seismology, from signal recognition and analysis to forecasting future acoustic/ seismic activity on the laboratory and regional scales (Rouet-Leduc et al, 2017;Rouet-Leduc et al, 2018;Chelidze et al, 2020;Ren et al, 2020;Johnson et al, 2021). In our analysis, we consider the problem of forecasting EQs in a given magnitude range and a given time interval using a supervised classification approach, namely, forecasting the probability of occurrence/absence of the seismic event using a previous day geophysical observations' training dataset.…”

Section: The Basic Of ML Metrics For Forecastingmentioning

confidence: 99%

“…Moreover, we analyzed geomagnetic time series recorded by the Dusheti Geophysical Observatory. In the previous work (Chelidze et al, 2020), we analyzed the precursory data of the water level and geomagnetic field variations collected one day before events of magnitudes M ≥ 3.5 in 2017-2019 on the territory of Georgia. Here, we present the results of the next-day forecast using multiparametric (hydrodynamic, magnetic, and tidal) monitoring data collected one day before events of magnitude M ≥ 3.5 carried out in 2020 on the territory of Georgia.…”

Section: Network Devices and Datamentioning

confidence: 99%

“…of the input data-the daily time series of geophysical data (attributes/features), namely, the input data were as follows: water level in the network of boreholes, tidal variations, geomagnetic field intensity, and atmospheric pressure in five circular (overlapping) regions of Georgia centered on the following borehole locations: Ajameti, Akhalkalaki, Kobuleti, Marneuli, and Nokalakevi during 2020 (see Figure 3). We compare these daily values with the data of daily occurrence of EQs of magnitude M ≥ 3.5 using the machine learning (ML) classification approach (Chelidze et al, 2020): to the days with events of M ≥ 3.5 , we assign value one and to "quiet" days, value 0.…”

Section: Geophysical Datasets Preparation Stagementioning

confidence: 99%

“…In this article (Chelidze et al, 2020), we considered the problem of earthquake (EQ) forecast using a machine learning (ML) approach, namely, the package ADAM (Kingma and Ba 2014), based on experimental (training) data on monitoring water-level variations in deep wells as well as geomagnetic and tidal time series in Georgia (Caucasus). In the 2020 article, we used the low EQ threshold value of magnitude M ≥ 3 as a forecast object.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Earthquake Forecast as a Machine Learning Problem for Imbalanced Datasets: Example of Georgia, Caucasus

et al. 2022

Self Cite

View full text Add to dashboard Cite

In this article, we considered the problem of M≥3 earthquake (EQ) forecasting (hindcasting) using a machine learning (ML) approach, using experimental (training) time series on monitoring water-level variations in deep wells as well as geomagnetic and tidal time series in Georgia (Caucasus). For such magnitudes’, the number of “seismic” to “aseismic” days in Georgia is approximately 1:5 and the dataset is close to the balanced one. However, the problem of forecast is practically important for stronger events—say, events of M≥3.5—which means that the learning dataset of Georgia became more imbalanced: the ratio of seismic to aseismic days for in Georgia reaches the values of the order of 1:20 and more. In this case, some accepted ML classification measures, such as accuracy leads to wrong predictions due to a large number of true negative cases. As a result, the minority class, here—seismically active periods—is ignored at all. We applied specific measures to avoid the imbalance effect and exclude the overfitting possibility. After regularization (balancing) of the training data, we build the confusion matrix and performed receiver operating classification in order to forecast the next day probability of M≥3.5 earthquake occurrence. We found that the Matthews’ correlation coefficient (MCC) is the measure, which gives good results even if the negative and positive classes are of very different sizes. Application of MCC to observed geophysical data gives a good forecast of the next day M≥3.5 seismic event probability of the order of 0.8. After randomization of EQ dates in the training dataset, the Matthews’ coefficient efficiency decreases to 0.17.

show abstract

Section: Complexity Analysis and Machinementioning

confidence: 99%