No abstract

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.

The transient temperature component was determined from long-term highly resolved temperature records in a borehole at Ajameti near Kutaisi in Western Georgia during the period between July 2017 and September 2018. Temperatures were recorded at depths of 100, 175 and 250 m with a resolution below the Millikelvin range and a recorded measurement frequency of 3 per hour, resulting in 72 individual measurements daily. At the depth of 100 m, a linear temperature increase of 0.0036 K/year was observed during those 15 months of measurement. At both larger depths, a precise linear trend could not be estimated. Additional impacts of water flow in the subsurface, penetrating from the surface or ascending from deeper layers during the time of measurements, superpose the transient component. The linear trend at 100 m can be understood as an increasing surface temperature at a rate of 0.015 K/year since 80 to 90 years. Two models agree with the data, i.e. a fast rise of the temperature as well as a continuous increase at a temperature diffusivity of κ= 0.9*10-6m²/s of the subsurface. The result coincides with the history of the Ajameti village which was founded in 1935, a period in which the settlement trees were probably cut to obtain a larger area of cultivated land, continuously increasing the surface temperature.

<p>In terms of geodynamic life, territory of Georgia is one of the most active region. The macro structural factor here is represented by the contact with the Arabian and Eurasian tectonic plates, which in addition to the geological diversity of the area conditions the high seismicity of mentioned region. </p><p>More the 20 year was operating a special network of hydro-geodynamical (water level, Atmosphere pressure and air temperature) observation on the territory of Georgia. Ten deep boreholes located basically on the main geo-plate and open deep aquifers. These wells as sensitive strain-meters recorded all kinds of deformation caused by exogenous (atmospheric pressure, tidal variations and season variation), as well as endogenous processes.</p><p>During observation on the territory of Georgia has observed various anomalies by water level before seismic events. Revealing of the mechanism of interrelation between the deformation processes, forestall strong earthquakes, and a hydrodynamic variation of underground waters, would allow to explain such preliminary behavior of hydrodynamic effects and to develop scientifically proven methods of the forecast of earthquakes.</p>