Abstract
Whether 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.
Introduction
Seismology has long excelled at long-term hazard assessment, yet short-term prediction remains an elusive goal. Current technical warning systems typically provide only seconds of notice—the time it takes for seismic waves to travel from the epicenter to a sensor. While anecdotal reports of unusual animal behavior prior to disasters date back to antiquity, these accounts are often retrospective and lack quantitative rigor. This study addresses the gap between folklore and hard science by instrumentally measuring animal behavior to evaluate its potential for a statistically reliable earthquake forecasting system.
Methods
The researchers employed a "bio-logging" approach, equipping livestock with high-tech sensors. 3D-accelerometers sampling at 54Hz were attached to six cows, five sheep, and two dogs on a farm in Capriglia, Italy. To quantify movement intensity, the team calculated Overall Dynamic Body Acceleration (ODBA).
To isolate abnormal pre-seismic behavior from daily noise, Vector Autoregressive (VAR) models were used to subtract circadian rhythms and inter-species influences. Unusual activity was defined as movements exceeding 2.5 standard deviations above the norm within a 20-hour window preceding a seismic event.
Key Results and Insights
The study utilized a massive dataset from the 2016/2017 Central Italy earthquake sequence, including over 18,000 earthquakes.
The Power of the Collective
While individual animal signals were often noisy, aggregating data into a "collective" signal across all three species significantly enhanced the statistical clarity of pre-seismic activity. Dogs were identified as the most sensitive to precursors, followed by cows, while sheep were the least reactive.
The Distance-Time Inverse Relationship
A striking finding was the significant negative correlation between anticipation time and hypocentral distance. As shown in the data, the further the farm was from the earthquake's origin, the shorter the anticipation time. This suggests that animals may be sensing a physical precursor, such as air ionization or gas release, that behaves like a diffusive process originating from the earthquake's nucleation point.
Environmental Influence
Anticipatory behavior was only detectable when animals were housed in a stable. When animals were on open pasture, the signal vanished. This discrepancy may be due to higher behavioral noise outdoors or a reduced sense of vulnerability in open spaces.
Limitations and Future Directions
This research was a "single-station" study; while it could detect an impending earthquake, it could not determine the direction of the strike. Future work aims to develop an "animal-sensor triangulation system" by monitoring multiple farms across a landscape. By analyzing different anticipation times at various sites, researchers could theoretically pinpoint a likely epicenter. Further investigation is also required to identify the exact physical stimulus the animals perceive and to refine the system to reduce false positives.
Summary updated