← Back to Sessions

Analyzing the Efficiency of Space-Based Geodesy Data for the Systematic Forecast of Earthquakes

Session: Application of Remote Sensing and Space-Based Earth Observations Data in Earthquake Research I

Type: Oral

Date: 4/21/2021

Presentation Time: 05:30 PM Pacific

Description: 

The efficiency of using space-based geodesy data for earthquake prediction is estimated. Prediction is based on our machine learning technique called the method of minimum area of alarm. The system makes forecast regularly with dt step. At each time t, it converts all types of input data (earthquake catalogs, time series, and raster data) into uniform spatio-temporal grid fields, trains on all data available at time t, and calculates the alarm zone in which the epicenter of the target earthquake is expected in the interval (t, t+dt). A demo version of the system since 2018 is available at https://distcomp.ru/geo/prognosis/.

The estimate was obtained from the results of the forecast of earthquakes in Japan with a magnitude of m ≥ 6.0 and a hypocenter depth of up to 60 km, which occurred from 2016 to 2020. The forecast results from GPS data are compared with the results of a random forecast, a forecast based on the spatial density of earthquake epicenters, a forecast based on spatio-temporal seismic data, and a forecast based on combined GPS and seismological data. It is shown that the coherence of changing the rates of various types of the earth's surface deformation fields in combination with the value of the change in the shear deformation rate can be a precursor to strong earthquakes. It is shown that the probability of forecasting earthquakes with magnitudes m ≥ 6.0 according to GPS data is statistically significantly higher than the probability of forecasting from random data and forecasting from the spatial density of earthquakes. The probability of a successful prediction of earthquakes using seismological data is higher than the probability of forecasting using GPS data. Adding the fields calculated from GPS data to the fields calculated from the earthquake catalog practically does not change the forecast probability. The obtained results are preliminary and require additional research in other seismically active regions.

Presenting Author: Valeri Gitis

Student Presenter: No

Authors