Seismic Noise Change at the Early Stage of COVID-19 Pandemic in Southern California

Baoning Wu, Roby Douilly, Heather Ford, Gareth Funning, Hsin-Yu Lee, Shankho Niyogi, Matt Mendoza, Christodoulos Kyriakopoulos, and David Oglesby

Motivation

Andrew

Gareth

(Gibney, 2020)

Noise drop in Belgium

How about having a look at the nearby stations?

Roby

Shankho

• Dense seismic network, well calibrated
• many stations in urban area, 2 near UCR
• Open-access data

Matt

Methods

Method: principle

High frequency noise

>1 Hz (or <1 s)

low frequency noise

0.1-1 Hz (or 1-10 s)

High frequency (HF) noises are mainly anthropogenic (traffic, construction, factory)

Low frequency noise usually comes from natural sources

(ocean waves, micro-seismicity)

Seismic noise spectrum (global average)

Earthquake signals

Seismic noise

How to monitor the seismic noise level change?

1. The continuous time-series of a station channel is parsed into 30-minute time-series segments, overlapping by 50% (15 minutes) and distributed continuously throughout the time period of interest.
2. Signals power (energy) in the anthropogenic noise frequency band (4-14 Hz) is calculated for each time window, we call it displacement Root Mean Square (RMS).

Displacement RMS (P):

An example

BK.BRK at UC Berkeley

• Anthropogenic noise decay rapidly with distance due to its high frequency nature, usually a few km.

Periodic pattern (diurnal/weekly)

suggests anthropogenic origins

Hsin-yu

Should be ~2km

An example

BK.BRK at UC Berkeley

• Anthropogenic noise decay rapidly with distance due to its high frequency nature, usually a few km.

Hsin-yu

Should be ~2km

Daily median: median noise value of the day.

An example

BK.BRK at UC Berkeley

• Weekdays high, weekend low

Daily median: median noise value of the day.

weekdays, high

weekend, low

2019 Christmas, low

Shelter-in-place

Global change of seismic noise (Lecocq et al., 2020)

• Noise drop in holiday like Christmas, weekly pattern modulated by holiday schedule.

• Noise drop after shelter-in-place, weekly pattern persists

• Consistent with the noise drop observed globally.

Results

CI.RSS in the Bell Tower

Abhi

“Have a look at our own station”

Anthropogenic seismic noise did not show a drop after the shelter-in-place order at CI.RSS in UC Riverside

How about other SoCal seismic stations?

• 19/23 show anthropogenic pattern.
• 5/19 show significant drop
• 14/19 did not show significant drop!

Why do seismic noise drop in some stations but not others?

What is the anthropogenic source that generate these seismic noise?

• All traces show very similar trend before the shelter-in-place order. For both Christmas and normal days.

​

• Differences in trend start from the shelter-in-place.

​

• Considering that noises decay fast (< 2km), sources should be homogeneous in SoCal

Traffic ?

4 pieces of supporting evidence:

1. HF seismic noise trends are homogeneous in SoCal, consistent with the nature of traffic activity in SoCal.
2. Noise direction points perpendicular to the nearby highway.
3. Noise propagates from station near the highway to station away from the highway.
4. HF seismic noise correlates with precipitation: rainfall 🡪 less HF seismic noise, consistent with traffic activity.

Traffic is the major HF seismic noise in SoCal.

Evidence #2: Noise direction points perpendicular to the nearby highway.

CI.RVR

CI.RSS

Seismic noise are mainly Rayleigh waves. It shakes the ground parallel to its propagation direction

• We find the maximum shaking directions of all the 30-minutes time segments (gray lines in (a)), the mode of direction is shown by the thick red line.

(a)

(b)

(c)

• The noise direction rotates clockwise from CI.RVR to CI.RSS along with the slight turn of California 60, indicating a highways traffic origin of seismic noises.

Evidence #3: Noise propagates from near highway station to away station

Station 1

Station 2

Noise propagating direction

CI.RSS (close) x-correlates AM.R2FCF (away)

Gareth

Hsin-yu

Evidence #4: Noise amplitude correlates with precipitation

5 “Drop” Stations VS. precipitation

14 “No Drop” Stations VS. precipitation

Gareth

Jen

Our hypothesis:

• We propose that it is the difference in traffic type that determine whether the seismic noise would drop near a station or not.
• For stations like CI.RSS, where the nearby traffic activities comprise a lot of industrial traffic such as cargo transportation, the displacement RMS would not drop
• For stations like CI.MSJ, where the nearby traffic activities could mostly comprise of academic related traffic such as commuting, the displacement RMS would drop after the shelter-in-place order.

Why do seismic noise drop in some stations but not others?

Heather

This hypothesis can be tested with more data in a future study.

1. We investigate the seismic noise change at 23 SoCal seismic stations at the early stage of COVID-19 pandemic. 19 of the 23 stations record noise with anthropogenic characteristics.
2. To our surprise, 14 of 19 stations did not record a drop of anthropogenic noise after the CA shelter-in-place order, despite a drop is recorded during Christmas.
3. Our analysis suggests that the primary seismic noise source at 4-14 Hz in Southern California might be traffic.
4. We propose that the industrial traffic, such as cargo transportation, is the reason why many stations did not record a noise drop.
5. Our results show that the anthropogenic seismic noise at 4-14 Hz recorded by seismic stations can indicate human activity and that this metric is particularly powerful in measuring how localized communities initially responded to the COVID-19 pandemic.

Summary

David

Thanks!

Christos