Mixture Models for Improved Short-Term Earthquake Forecasting

2009 ◽  
Vol 99 (2A) ◽  
pp. 636-646 ◽  
Author(s):  
D. A. Rhoades ◽  
M. C. Gerstenberger
Keyword(s):  
Mixture Models ◽  
Earthquake Forecasting ◽  
Short Term

scholarly journals Response to Zöller et al.'s critique on “Potential short‐term earthquake forecasting by farm‐animal monitoring”

Ethology ◽  
2021 ◽  
Vol 127 (3) ◽  
pp. 307-308
Author(s):  
Martin Wikelski ◽  
Uschi Mueller ◽  
Paola Scocco ◽  
Andrea Catorci ◽  
Lev V. Desinov ◽  
...  
Keyword(s):  
Farm Animal ◽  
Earthquake Forecasting ◽  
Short Term ◽  
Animal Monitoring

scholarly journals Short-term earthquake forecasting experiment before and during the L’Aquila (central Italy) seismic sequence of April 2009

2015 ◽  
Vol 57 (6) ◽  
Author(s):  
Maura Murru ◽  
Jiancang Zhuang ◽  
Rodolfo Console ◽  
Giuseppe Falcone
Keyword(s):  
Central Italy ◽  
Aftershock Sequence ◽  
Earthquake Forecasting ◽  
Model Parameters ◽  
Short Term ◽  
Productivity Function ◽  
Epidemic Type Aftershock Sequence ◽  
Forecasting Performance ◽  
The One ◽  
Lower Magnitude

<div class="page" title="Page 1"><div class="layoutArea"><div class="column"><p>In this paper, we compare the forecasting performance of several statistical models, which are used to describe the occurrence process of earthquakes in forecasting the short-term earthquake probabilities during the L’Aquila earthquake sequence in central Italy in 2009. These models include the Proximity to Past Earthquakes (PPE) model and two versions of the Epidemic Type Aftershock Sequence (ETAS) model. We used the information gains corresponding to the Poisson and binomial scores to evaluate the performance of these models. It is shown that both ETAS models work better than the PPE model. However, in comparing the two types of ETAS models, the one with the same fixed exponent coefficient (<span>alpha)</span> = 2.3 for both the productivity function and the scaling factor in the spatial response function (ETAS I), performs better in forecasting the active aftershock sequence than the model with different exponent coefficients (ETAS II), when the Poisson score is adopted. ETAS II performs better when a lower magnitude threshold of 2.0 and the binomial score are used. The reason is found to be that the catalog does not have an event of similar magnitude to the L’Aquila mainshock (M<sub>w</sub> 6.3) in the training period (April 16, 2005 to March 15, 2009), and the (<span>alpha)</span>-value is underestimated, thus the forecast seismicity is underestimated when the productivity function is extrapolated to high magnitudes. We also investigate the effect of the inclusion of small events in forecasting larger events. These results suggest that the training catalog used for estimating the model parameters should include earthquakes of magnitudes similar to the mainshock when forecasting seismicity during an aftershock sequence.</p></div></div></div>

scholarly journals mixtur: An R package for designing, analysing, and modelling continuous report visual short-term memory studies

2020 ◽  
Author(s):  
Jim Grange ◽  
Stuart Bryan Moore ◽  
Ed David John Berry
Keyword(s):  
Mixture Models ◽  
Short Term Memory ◽  
R Package ◽  
Memory Representation ◽  
Angular Deviation ◽  
Short Term ◽  
Target Feature ◽  
Component Mixture ◽  
Term Memory ◽  
Visual Short Term Memory

Visual short-term memory (vSTM) is often measured via continuous-report tasks whereby participants are presented with stimuli that vary along a continuous dimension (e.g., colour) with the goal of memorising the stimuli features. At test, participants are probed to recall the feature value of one of the memoranda in a continuous manner (e.g., by clicking on a colour wheel). The angular deviation between the participant response and the true feature value provides an estimate of recall---and hence, vSTM---precision. Two prominent models of performance on such tasks are the two- and three-component mixture models (Bays et al., 2009; Zhang &amp; Luck, 2008). Both models decompose participant responses into probabilistic mixtures of: (1) responses to the true target value based on a noisy memory representation; (2) random guessing when memory fails. In addition, the three-component model proposes (3) responses to a non-target feature value (i.e., binding errors). Here we report the development of mixtur, an open-source package written for the statistical programming language R that facilitates the fitting of the 2- and 3-component mixture models to continuous report data. We also report the results of several simulations conducted to develop recommendations for researchers on trial numbers, set-sizes and memoranda similarity, as well as conducting parameter recovery and model recovery simulations. It is our hope that mixtur will lower the barrier of entry for utilising mixture modelling

Some Challenges of Short-Term Earthquake Forecasting and Possible Solutions

2020 ◽  
Vol 495 (2) ◽  
pp. 910-913
Author(s):  
V. G. Bondur ◽  
M. B. Gokhberg ◽  
I. A. Garagash ◽  
D. A. Alekseev
Keyword(s):  
Earthquake Forecasting ◽  
Short Term

Integration of geological and seismological data in earthquakes occurrence models for Italy: towards a unified model for different forecast perspectives

2019 ◽  
Vol 219 (3) ◽  
pp. 2148-2164
Author(s):  
A M Lombardi
Keyword(s):  
Aftershock Sequence ◽  
Earthquake Forecasting ◽  
Short Term ◽  
Background Rate ◽  
Etas Model ◽  
Term Forecast ◽  
Earthquake Predictability ◽  
Background Seismicity ◽  
Different Types

SUMMARY The operational earthquake forecasting (OEF) is a procedure aimed at informing communities on how seismic hazard changes with time. This can help them live with seismicity and mitigate risk of destructive earthquakes. A successful short-term prediction scheme is not yet produced, but the search for it should not be abandoned. This requires more research on seismogenetic processes and, specifically, inclusion of any information about earthquakes in models, to improve forecast of future events, at any spatio-temporal-magnitude scale. The short- and long-term forecast perspectives of earthquake occurrence followed, up to now, separate paths, involving different data and peculiar models. But actually they are not so different and have common features, being parts of the same physical process. Research on earthquake predictability can help to search for a common path in different forecast perspectives. This study aims to improve the modelling of long-term features of seismicity inside the epidemic type aftershock sequence (ETAS) model, largely used for short-term forecast and OEF procedures. Specifically, a more comprehensive estimation of background seismicity rate inside the ETAS model is attempted, by merging different types of data (seismological instrumental, historical, geological), such that information on faults and on long-term seismicity integrates instrumental data, on which the ETAS models are generally set up. The main finding is that long-term historical seismicity and geological fault data improve the pseudo-prospective forecasts of independent seismicity. The study is divided in three parts. The first consists in models formulation and parameter estimation on recent seismicity of Italy. Specifically, two versions of ETAS model are compared: a ‘standard’, previously published, formulation, only based on instrumental seismicity, and a new version, integrating different types of data for background seismicity estimation. Secondly, a pseudo-prospective test is performed on independent seismicity, both to test the reliability of formulated models and to compare them, in order to identify the best version. Finally, a prospective forecast is made, to point out differences and similarities in predicting future seismicity between two models. This study must be considered in the context of its limitations; anyway, it proves, beyond argument, the usefulness of a more sophisticated estimation of background rate, inside short-term modelling of earthquakes.

scholarly journals Bimodal Dynamics in Short-Term Growth Among Peruvian Infants at Risk of Growth-Faltering

2021 ◽  
Vol 5 (Supplement_2) ◽  
pp. 657-657
Author(s):  
Gwenyth Lee ◽  
Benjamin McCormick ◽  
Pablo Peñataro-Yori ◽  
Maribel Paredes-Olortegui ◽  
Laura Caulfield ◽  
...  
Keyword(s):  
Mixture Models ◽  
Finite Mixture Models ◽  
High Growth ◽  
Growth Patterns ◽  
Finite Mixture ◽  
Short Term ◽  
Lower Growth ◽  
Term Infants ◽  
Normal Distributions ◽  
Growth Faltering

Abstract Objectives Over the short term, infants may grow in a ‘bimodal’ manner, with interspersed periods of greater and lesser growth. Our aim was to describe the modality of high-resolution growth dynamics of infants who experienced cumulative growth faltering in the first year of life. Methods Thrice-weekly measurements of length were recorded for n = 58 children enrolled from birth to one year in the Peru cohort of the Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development’ (MAL-ED) study. Finite mixture models were fitted to the length velocity data to test for evidence of multiphasic patterns. We then tested whether evidence of a multiphasic pattern remained after applying a smoothing algorithm to account for measurement error. Data was smoothed using kernel regression with a monotonic constraint. Finite mixture models were fitted to unsmoothed and smoothed data to examine whether growth patterns varied by age and sex. We also fitted stratified models to compare short-term growth patterns between infants who maintained a consistent length-for-age Z-score (LAZ) from 2–12 months of age, versus those whose LAZ decreased by at least 0.25 over the same period. Results Unsmoothed data were best described by a biphasic finite mixture model. The growth of children less than 3 months old was described by a mixture of two normal distributions with both means significantly greater than zero (0.110 cm/day, 95%CI: 0.105, 0.114 and 0.291 cm/day, 95%CI: 0.235, 0.347). By 6 months, this transitioned to a pattern of two normal distributions, one with a mean near zero (0.019, 95% CI: 0.0178, 0.020), and one with a mean of 0.102 cm/day (95% CI: 0.098, 0.0107). Children who lost 0.25 or more in LAZ from 2 to 12 months had a similar mean growth velocity in both the ‘low’ and ‘high’ growth phase but spent smore time in the former phase (54.9% of days versus 39.2% of days) than children who either maintained or gained LAZ over the same age. Conclusions Consistent with other reports, we find that infant growth appears to follow bimodal dynamics characterized by intervals of greater velocities as well as periods of much lower growth. Funding Sources This work was supported by the National Institutes of Health and the BMGF.

scholarly journals Earthquake trend prediction using long short-term memory RNN

2019 ◽  
Vol 9 (2) ◽  
pp. 1304 ◽  
Author(s):  
Tanvi Bhandarkar ◽  
Vardaan K ◽  
Nikhil Satish ◽  
S. Sridhar ◽  
R. Sivakumar ◽  
...  
Keyword(s):  
Neural Network ◽  
Short Term Memory ◽  
Future Trend ◽  
Earthquake Forecasting ◽  
Short Term ◽  
Feed Forward Neural Network ◽  
Trend Prediction ◽  
Term Memory ◽  
Area Of Interest ◽  
Long Short Term Memory

<p>The prediction of a natural calamity such as earthquakes has been an area of interest for a long time but accurate results in earthquake forecasting have evaded scientists, even leading some to deem it intrinsically impossible to forecast them accurately. In this paper an attempt to forecast earthquakes and trends using a data of a series of past earthquakes. A type of recurrent neural network called Long Short-Term Memory (LSTM) is used to model the sequence of earthquakes. The trained model is then used to predict the future trend of earthquakes. An ordinary Feed Forward Neural Network (FFNN) solution for the same problem was done for comparison. The LSTM neural network was found to outperform the FFNN. The R^2 score of the LSTM is better than the FFNN’s by 59%.</p>

scholarly journals Potential short-term earthquake forecasting by farm-animal monitoring

2020 ◽  
Author(s):  
Martin Wikelski ◽  
Uschi Mueller ◽  
Paola Scocco ◽  
Andrea Catorci ◽  
Lev Desinov ◽  
...  
Keyword(s):  
Seismic Activity ◽  
Earthquake Forecasting ◽  
Farm Animals ◽  
Activity Levels ◽  
Short Term ◽  
Animal Activity ◽  
Long Time ◽  
Earthquake Predictions ◽  
Anticipatory Activity ◽  
Animal Monitoring

AbstractWhether changes in animal behavior allow for short-term earthquake predictions has been debated for a long time. During the 2016/2017 earthquake sequence in Italy, we instrumentally observed the activity of farm animals (cows, dogs, sheep) close to the epicenter of the devastating magnitude M6.6 Norcia earthquake (Oct-Nov 2016) and over a subsequent longer observation period (Jan-Apr 2017). Relating 5304 (in 2016) and 12948 (in 2017) earthquakes with a wide magnitude range (0.4 ≤ M ≤ 6.6) to continuously measured animal activity, we detected how the animals collectively reacted to earthquakes. We also found consistent anticipatory activity prior to earthquakes during times when the animals were in a stable, but not during their time on a pasture. We detect these anticipatory patterns not only in periods with high, but also in periods of low seismic activity. Earthquake anticipation times (1-20hrs) are negatively correlated with the distance between the farm and earthquake hypocenters. Our study suggests that continuous instrumental monitoring of animal collectives has the potential to provide statistically reliable patterns of pre-seismic activity that could allow for short-term earthquake forecasting.One Sentence SummaryA collective of domestic animals repeatedly showed unusually high activity levels before earthquakes, with anticipation times (1-20h) negatively related to distance from epicenters (5-28km).

Sign in / Sign up

Export Citation Format

Share Document