Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high...

63
Towards high-resolution seismic imaging and monitoring Nori Nakata (MIT) Microseismic wavefield imaging Ambient noise imaging Machine learning Ambient noise imaging

Transcript of Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high...

Page 1: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Towards high-resolution seismic imaging and monitoring

Nori Nakata (MIT)

Microseismic wavefield imaging

Ambient noise imaging

Machine learning

Ambient noise imaging

Page 2: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Towards high-resolution seismic imaging and monitoring

using ambient noise, machine learning, and microseismic events

Nori Nakata (MIT)Yuji Kim, Jianhang Yin, Bin Lyu (Univ. Oklahoma)Yuwei Wang (Southwest Petroleum Univ)Rie Nakata Kamei, Zack Spica (Univ. Tokyo)Greg Beroza (Stanford)David Shelly (USGS)

Page 3: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Human-Induced Seismicity

Rubinstein and Babaie, 2015, SRL

• EOR• Geothermal• Mining

Fracking Production Wastewater disposal

Page 4: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Human-Induced SeismicityTasks• Assessing seismic hazard• Understanding seismic sources• Understanding physics of seismicity• Effectively enhancing production• Real-time application• …

Page 5: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Human-Induced Seismicity

source

path

site

Effects of Source + Path + Site

Tasks• Assessing seismic hazard• Understanding seismic sources• Understanding physics of seismicity• Effectively enhancing production• Real-time application

Page 6: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Human-Induced Seismicity

source

path

site

Tasks• Assessing seismic hazard• Understanding seismic sources• Understanding physics of seismicity• Effectively enhancing production• Real-time application

My approachUnique datasets

• Ambient noise• Microseismic events• Repeated active

surveys• Tube waves• …

Advanced techniques• Time-lapse FWI• Machine learning• Stochastic modeling• Wavefield migration• …

Page 7: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Human-Induced Seismicity

source

path

site

My approachUnique datasets

• Ambient noise• Microseismic events• Repeated active

surveys• Tube waves• …

Advanced techniques• Time-lapse FWI• Machine learning• Stochastic modeling• Wavefield migration• …

Tasks• Assessing seismic hazard• Understanding seismic sources• Understanding physics of seismicity• Effectively enhancing production• Real-time application

Page 8: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Human-Induced Seismicity

source

path

site

My approachUnique datasets

• Ambient noise• Microseismic events• Repeated active

surveys• Tube waves• …

Advanced techniques• Time-lapse FWI• Machine learning• Stochastic modeling• Wavefield migration• …

Tasks• Assessing seismic hazard• Understanding seismic sources• Understanding physics of seismicity• Effectively enhancing production• Real-time application

Page 9: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Arrival time is useful if clearly identified

Waldhauser and Ellsworth, BSSA, 2000

Arrival time picks are used to locate source.

Page 10: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Small, weak, and extended events arehard to detect/locate

Wave arrivals?

Migration-based processing

Kao and Shan, GJI, 2004

We can’t pick arrival times accurately

Page 11: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Recording microseismic data

Page 12: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 13: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Numerical example

Page 14: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 15: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 16: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 17: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 18: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 19: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal imaging

Page 20: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

focused!

Time-reversal imaging

Page 21: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

strong artifacts

Time-reversal imaging

Page 22: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

R(x, t)

Time

Time-reversal imaging

Page 23: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal image (conventional method)

New approach (Geometric-mean RTM)

Nakata & Beroza, Geophysics, 2016

Page 24: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-reversal image (conventional method)

Geometric-mean RTM (new approach)

Nakata & Beroza, Geophysics, 2016

Page 25: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

At time-lag = 0…,

Time-reversal imaging= Arithmetic-mean reverse-time imaging

Geometric-mean reverse-time migration

⇥<latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit>

⇥<latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit>

⇥<latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit><latexit sha1_base64="+eCijc6BpjzV9GsPhy5Ua9rnbv0=">AAAB7XicbVDLSgNBEOyNrxhfUY9eBoPgKewGQY8BLx4jmAckS5idzCZjZmeWmV4hhPyDFw+KePV/vPk3TpI9aGJBQ1HVTXdXlEph0fe/vcLG5tb2TnG3tLd/cHhUPj5pWZ0ZxptMS206EbVcCsWbKFDyTmo4TSLJ29H4du63n7ixQqsHnKQ8TOhQiVgwik5q9VAk3PbLFb/qL0DWSZCTCuRo9MtfvYFmWcIVMkmt7QZ+iuGUGhRM8lmpl1meUjamQ951VFG3JJwurp2RC6cMSKyNK4Vkof6emNLE2kkSuc6E4siuenPxP6+bYXwTToVKM+SKLRfFmSSoyfx1MhCGM5QTRygzwt1K2IgaytAFVHIhBKsvr5NWrRo4fn9VqdfyOIpwBudwCQFcQx3uoAFNYPAIz/AKb572Xrx372PZWvDymVP4A+/zB7FdjyQ=</latexit>

Page 26: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Comparison

Page 27: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time reversal GmRTM

Page 28: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

GmRTM is efficient for :- locating with higher spatial resolution- using long-duration signals (fracking, tremors…)- enhancing weak signals (i.e., noisy signals)

Time reversal GmRTM

Page 29: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Field data example (Fracking)An oil field in Japan

• Only 8 receivers available• Challenging geometry

Wang, Nakata, et al., 2019

Page 30: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Field data example (Fracking)

Time reversal GmRTM

Wang, Nakata, et al., 2019

Page 31: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Velocity update with microseismic data

source

path

site

Lyu & Nakata, 2019

Velocity

Microseismic location

GmRTM Passive FWI

R(x) =X

t

Y

i

Wi(x, t)<latexit sha1_base64="K2Sx55V7iAyGW7W4cUxtWuoZkAw=">AAACB3icbZDLSgMxFIYz9VbrbdSlIMEitCBlpgi6EQpuXFaxF2iHIZNJ29BkZkjOSEvpzo2v4saFIm59BXe+jelloa0/BD7+cw4n5w8SwTU4zreVWVldW9/Ibua2tnd29+z9g7qOU0VZjcYiVs2AaCZ4xGrAQbBmohiRgWCNoH89qTcemNI8ju5hmDBPkm7EO5wSMJZvH98VBkV8hds6lT7gdqLi0Oe44fPC4AyKvp13Ss5UeBncOeTRXFXf/mqHMU0li4AKonXLdRLwRkQBp4KNc+1Us4TQPumylsGISKa90fSOMT41Tog7sTIvAjx1f0+MiNR6KAPTKQn09GJtYv5Xa6XQufRGPEpSYBGdLeqkAkOMJ6HgkCtGQQwNEKq4+SumPaIIBRNdzoTgLp68DPVyyTV8e56vlOdxZNEROkEF5KILVEE3qIpqiKJH9Ixe0Zv1ZL1Y79bHrDVjzWcO0R9Znz8jipeE</latexit><latexit sha1_base64="K2Sx55V7iAyGW7W4cUxtWuoZkAw=">AAACB3icbZDLSgMxFIYz9VbrbdSlIMEitCBlpgi6EQpuXFaxF2iHIZNJ29BkZkjOSEvpzo2v4saFIm59BXe+jelloa0/BD7+cw4n5w8SwTU4zreVWVldW9/Ibua2tnd29+z9g7qOU0VZjcYiVs2AaCZ4xGrAQbBmohiRgWCNoH89qTcemNI8ju5hmDBPkm7EO5wSMJZvH98VBkV8hds6lT7gdqLi0Oe44fPC4AyKvp13Ss5UeBncOeTRXFXf/mqHMU0li4AKonXLdRLwRkQBp4KNc+1Us4TQPumylsGISKa90fSOMT41Tog7sTIvAjx1f0+MiNR6KAPTKQn09GJtYv5Xa6XQufRGPEpSYBGdLeqkAkOMJ6HgkCtGQQwNEKq4+SumPaIIBRNdzoTgLp68DPVyyTV8e56vlOdxZNEROkEF5KILVEE3qIpqiKJH9Ixe0Zv1ZL1Y79bHrDVjzWcO0R9Znz8jipeE</latexit><latexit sha1_base64="K2Sx55V7iAyGW7W4cUxtWuoZkAw=">AAACB3icbZDLSgMxFIYz9VbrbdSlIMEitCBlpgi6EQpuXFaxF2iHIZNJ29BkZkjOSEvpzo2v4saFIm59BXe+jelloa0/BD7+cw4n5w8SwTU4zreVWVldW9/Ibua2tnd29+z9g7qOU0VZjcYiVs2AaCZ4xGrAQbBmohiRgWCNoH89qTcemNI8ju5hmDBPkm7EO5wSMJZvH98VBkV8hds6lT7gdqLi0Oe44fPC4AyKvp13Ss5UeBncOeTRXFXf/mqHMU0li4AKonXLdRLwRkQBp4KNc+1Us4TQPumylsGISKa90fSOMT41Tog7sTIvAjx1f0+MiNR6KAPTKQn09GJtYv5Xa6XQufRGPEpSYBGdLeqkAkOMJ6HgkCtGQQwNEKq4+SumPaIIBRNdzoTgLp68DPVyyTV8e56vlOdxZNEROkEF5KILVEE3qIpqiKJH9Ixe0Zv1ZL1Y79bHrDVjzWcO0R9Znz8jipeE</latexit><latexit sha1_base64="K2Sx55V7iAyGW7W4cUxtWuoZkAw=">AAACB3icbZDLSgMxFIYz9VbrbdSlIMEitCBlpgi6EQpuXFaxF2iHIZNJ29BkZkjOSEvpzo2v4saFIm59BXe+jelloa0/BD7+cw4n5w8SwTU4zreVWVldW9/Ibua2tnd29+z9g7qOU0VZjcYiVs2AaCZ4xGrAQbBmohiRgWCNoH89qTcemNI8ju5hmDBPkm7EO5wSMJZvH98VBkV8hds6lT7gdqLi0Oe44fPC4AyKvp13Ss5UeBncOeTRXFXf/mqHMU0li4AKonXLdRLwRkQBp4KNc+1Us4TQPumylsGISKa90fSOMT41Tog7sTIvAjx1f0+MiNR6KAPTKQn09GJtYv5Xa6XQufRGPEpSYBGdLeqkAkOMJ6HgkCtGQQwNEKq4+SumPaIIBRNdzoTgLp68DPVyyTV8e56vlOdxZNEROkEF5KILVEE3qIpqiKJH9Ixe0Zv1ZL1Y79bHrDVjzWcO0R9Znz8jipeE</latexit>

J =1

2||dobss,i � dpres,i ||

22

<latexit sha1_base64="Q7ImEigos/DNnd7vFqV1VNmLf7A=">AAACHXicbVDLSsNAFJ3UV62vqEs3g0VwoSUJBd0IBTfiqoJ9QJuGyWTSDp08mJkIJcmPuPFX3LhQxIUb8W+ctllo64GBc8+5d2bucWNGhTSMb620srq2vlHerGxt7+zu6fsHbRElHJMWjljEuy4ShNGQtCSVjHRjTlDgMtJxx9dTv/NAuKBReC8nMbEDNAypTzGSSnL0+i28gn2fI5yaeWrlWeYN0sgVuZOKM5rDc6hqdWNRZ5ljDSxHrxo1Ywa4TMyCVEGBpqN/9r0IJwEJJWZIiJ5pxNJOEZcUM5JX+okgMcJjNCQ9RUMUEGGns+1yeKIUD/oRVyeUcKb+nkhRIMQkcFVngORILHpT8T+vl0j/0k5pGCeShHj+kJ8wKCM4jQp6lBMs2UQRhDlVf4V4hFRUUgVaUSGYiysvk7ZVMxW/q1cbVhFHGRyBY3AKTHABGuAGNEELYPAInsEreNOetBftXfuYt5a0YuYQ/IH29QMxi6Hw</latexit><latexit sha1_base64="Q7ImEigos/DNnd7vFqV1VNmLf7A=">AAACHXicbVDLSsNAFJ3UV62vqEs3g0VwoSUJBd0IBTfiqoJ9QJuGyWTSDp08mJkIJcmPuPFX3LhQxIUb8W+ctllo64GBc8+5d2bucWNGhTSMb620srq2vlHerGxt7+zu6fsHbRElHJMWjljEuy4ShNGQtCSVjHRjTlDgMtJxx9dTv/NAuKBReC8nMbEDNAypTzGSSnL0+i28gn2fI5yaeWrlWeYN0sgVuZOKM5rDc6hqdWNRZ5ljDSxHrxo1Ywa4TMyCVEGBpqN/9r0IJwEJJWZIiJ5pxNJOEZcUM5JX+okgMcJjNCQ9RUMUEGGns+1yeKIUD/oRVyeUcKb+nkhRIMQkcFVngORILHpT8T+vl0j/0k5pGCeShHj+kJ8wKCM4jQp6lBMs2UQRhDlVf4V4hFRUUgVaUSGYiysvk7ZVMxW/q1cbVhFHGRyBY3AKTHABGuAGNEELYPAInsEreNOetBftXfuYt5a0YuYQ/IH29QMxi6Hw</latexit><latexit sha1_base64="Q7ImEigos/DNnd7vFqV1VNmLf7A=">AAACHXicbVDLSsNAFJ3UV62vqEs3g0VwoSUJBd0IBTfiqoJ9QJuGyWTSDp08mJkIJcmPuPFX3LhQxIUb8W+ctllo64GBc8+5d2bucWNGhTSMb620srq2vlHerGxt7+zu6fsHbRElHJMWjljEuy4ShNGQtCSVjHRjTlDgMtJxx9dTv/NAuKBReC8nMbEDNAypTzGSSnL0+i28gn2fI5yaeWrlWeYN0sgVuZOKM5rDc6hqdWNRZ5ljDSxHrxo1Ywa4TMyCVEGBpqN/9r0IJwEJJWZIiJ5pxNJOEZcUM5JX+okgMcJjNCQ9RUMUEGGns+1yeKIUD/oRVyeUcKb+nkhRIMQkcFVngORILHpT8T+vl0j/0k5pGCeShHj+kJ8wKCM4jQp6lBMs2UQRhDlVf4V4hFRUUgVaUSGYiysvk7ZVMxW/q1cbVhFHGRyBY3AKTHABGuAGNEELYPAInsEreNOetBftXfuYt5a0YuYQ/IH29QMxi6Hw</latexit><latexit sha1_base64="Q7ImEigos/DNnd7vFqV1VNmLf7A=">AAACHXicbVDLSsNAFJ3UV62vqEs3g0VwoSUJBd0IBTfiqoJ9QJuGyWTSDp08mJkIJcmPuPFX3LhQxIUb8W+ctllo64GBc8+5d2bucWNGhTSMb620srq2vlHerGxt7+zu6fsHbRElHJMWjljEuy4ShNGQtCSVjHRjTlDgMtJxx9dTv/NAuKBReC8nMbEDNAypTzGSSnL0+i28gn2fI5yaeWrlWeYN0sgVuZOKM5rDc6hqdWNRZ5ljDSxHrxo1Ywa4TMyCVEGBpqN/9r0IJwEJJWZIiJ5pxNJOEZcUM5JX+okgMcJjNCQ9RUMUEGGns+1yeKIUD/oRVyeUcKb+nkhRIMQkcFVngORILHpT8T+vl0j/0k5pGCeShHj+kJ8wKCM4jQp6lBMs2UQRhDlVf4V4hFRUUgVaUSGYiysvk7ZVMxW/q1cbVhFHGRyBY3AKTHABGuAGNEELYPAInsEreNOetBftXfuYt5a0YuYQ/IH29QMxi6Hw</latexit>

minimize…

Page 32: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Velocity update with microseismic data

Lyu & Nakata, 2019

Initial velocity model

Initial source locationsGmRTM FWI

Page 33: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Velocity update with microseismic data

Lyu & Nakata, 2019

Initial velocity model

Initial source locationsGmRTM FWI

Inverted velocity model

Inverted source locations

True velocity model

Page 34: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Velocity update with microseismic data

Lyu & Nakata, 2019

Initial velocity model

Initial source locationsGmRTM FWI

Inverted velocity model

Inverted source locations

True velocity model

Page 35: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

FWI for source parameter estimation

5

10

15

Dep

th [k

m]

2030405060Distance from deformation front [km]

(b)

−1.5

−1.0

−0.5

0.0

0.5

1.0

1.5km/s

5

10

15

Dep

th [k

m]

2030405060Distance from deformation front [km]

Cost function

Kamei, Lumley, Nakata, 2018

Field data example (crosswell active survey)Vel model Source time func.Radiation

Page 36: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Structure imaging with small earthquakes

Nakata & Shelly, 2018, GRLMagma roof

1 station800 earthquakes

Page 37: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Ambient noise

source

path

site SEISMIC AMBIENT NOISE

E D I T E D B Y

Nori Nakata, Lucia Gualtieri and Andreas Fichtner

A book about Seismic Ambient Noise

Cambridge University Press

370 pp

Published on May 2, 2019

Page 38: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Velocity estimationNear-surface

S velocity 3D P velocity

Nakata et al., 2015

0.0

1.0

Dept

h (k

m)

0.0

3.0

2.0

1.0

4.0

Easting (km)

0.0

8.07.0

6.05.0

4.03.0

2.01.0

Northing (km)

0

10

5

-10

-5 P ve

locit

yflu

ctua

tion

(%)

Page 39: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

High-frequency ground motion prediction

Time (s)0 2 4 6 8 10 12

Nor

mal

ized

pow

er

0

0.2

0.4

0.6

0.8

1 earthquake enveloperadiative transfer (P)radiative transfer (S)

Nakata & Beroza, 2015

8-16 Hz

Page 40: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

source

path

site

Groningen Gas Field60

km

40 km

Spica, Nakata et al, 2018ab, SRL

• 415 stations• 3 components• Receiver spacing: 400 m• One-month continuous record

Page 41: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Groningen Gas Field60

km

40 km

Spica, Nakata et al, 2018ab, SRL

Near-surface velocity model estimated from• H/V• Rayleigh and Love waves

Page 42: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Groningen Gas FieldHourly seismic velocity change

60 k

m

40 km

Nakata, et al., in prep

Page 43: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time-lapse FWI (CO2 monitoring)

−150 −100 −50 0 50DistDnce (m)

900

950

1000

1050

1100

1150

1200

1250

Dep

th (m

)

(a) 011 - 010

−150 −100 −50 0 50DistDnce (m)

(b) 012 - 010

−150 −100 −50 0 50DistDnce (m)

(c) 013 - 010

−150 −100 −50 0 50DistDnce (m)

(d) 014 - 010

−150 −100 −50 0 50DistDnce (m)

(e) 015 - 010

−150 −100 −50 0 50DistDnce (m)

(f) 016 - 010

−150 −100 −50 0 50DistDnce (m)

(e) 017 - 010

−0.5 0.0 0.5dVp (Nm/V)

Baseline

Nakata & Nakata, in prep

Receiver array

Source array

MN1 – Baseline(3000 t injected)

MN2 – Baseline(6000 t injected)

MN3 – Baseline(9000 t injected)

−150−100−500

50DistDnce (m

)

900 950 1000 1050 1100 1150 1200 1250Depth (m)

(a) 011 - 010 −150−100

−50050

DistDnce (m)

(b) 012 - 010 −150−100

−50050

DistDnce (m)

(c) 013 - 010 −150−100

−50050

DistDnce (m)

(d) 014 - 010 −150−100

−50050

DistDnce (m)

(e) 015 - 010 −150−100

−50050

DistDnce (m)(f) 016 - 010 −150−100

−50050

DistDnce (m)

(e) 017 - 010

−0.50.0

0.5dVp (Nm/V)

dVp(km

/s)

-0.5

0.5

Page 44: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Velocity change prediction by Machine learning

Dec 11 Dec 21 Dec 31 Jan 10 Jan 20-1

-0.5

0

0.5

1Ve

loci

ty c

hang

e (%

) observed

Nakata & Kim, in prep

Near-surface velocity changes over 40 days

Page 45: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Dec 11 Dec 21 Dec 31 Jan 10 Jan 20-1

-0.5

0

0.5

1Ve

loci

ty c

hang

e (%

) observedpredicted

Velocity change prediction by Machine learning

Nakata & Kim, in prep

Near-surface velocity changes over 40 days

Prediction using machine-learning regression analysisInput data: first 20 days & environmental parameters

Page 46: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Reflectivity inversion

Reflectivity inversion based on conventional deconvolution (least-squares inversion)

s(t) = h(t) ⇤ r(t)<latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit><latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit><latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit><latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit>

reflectivity

observed datasource wavelet

<latexit sha1_base64="7fCBr4wiH7EWOTWZxVzjcscUfuk=">AAAB7HicbZBNSwMxEIZn61etX1WPXoJFrJeyK4I9Frx4rOC2hXYp2TTbhmazSzIrlNLf4MWDIl79Qd78N6btHrT1hcDDOzNk5g1TKQy67rdT2Njc2t4p7pb29g8Oj8rHJy2TZJpxnyUy0Z2QGi6F4j4KlLyTak7jUPJ2OL6b19tPXBuRqEecpDyI6VCJSDCK1vL1ZRWv+uWKW3MXIuvg5VCBXM1++as3SFgWc4VMUmO6nptiMKUaBZN8VuplhqeUjemQdy0qGnMTTBfLzsiFdQYkSrR9CsnC/T0xpbExkzi0nTHFkVmtzc3/at0Mo3owFSrNkCu2/CjKJMGEzC8nA6E5QzmxQJkWdlfCRlRThjafkg3BWz15HVrXNc/yw02lUc/jKMIZnEMVPLiFBtxDE3xgIOAZXuHNUc6L8+58LFsLTj5zCn/kfP4A1oyOAA==</latexit>

L =1

2||s(t)� h(t) ⇤ r0(t)||22

<latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit><latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit><latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit><latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit>

Kim & Nakata, 2018, TLE

Page 47: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Reflectivity inversion with Machine Learning

Reflectivity inversion based on conventional deconvolution (least-squares inversion)

s(t) = h(t) ⇤ r(t)<latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit><latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit><latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit><latexit sha1_base64="r0gNx3LatYCY6eJ4m5nQTDVrOKI=">AAAB+3icbVDLSgNBEOz1GeNrjUcvg0GIHsKuCHoRgl48RjAPSJYwO5kkQ2YfzPSKYcmvePGgiFd/xJt/42yyB00sGLqo6qZ7yo+l0Og439bK6tr6xmZhq7i9s7u3bx+UmjpKFOMNFslItX2quRQhb6BAydux4jTwJW/549vMbz1ypUUUPuAk5l5Ah6EYCEbRSD27pCt4Sq7JKCtnRJnSs8tO1ZmBLBM3J2XIUe/ZX91+xJKAh8gk1brjOjF6KVUomOTTYjfRPKZsTIe8Y2hIA669dHb7lJwYpU8GkTIvRDJTf0+kNNB6EvimM6A40oteJv7ndRIcXHmpCOMEecjmiwaJJBiRLAjSF4ozlBNDKFPC3ErYiCrK0MRVNCG4i19eJs3zqmv4/UW5dpPHUYAjOIYKuHAJNbiDOjSAwRM8wyu8WVPrxXq3PuatK1Y+cwh/YH3+ABsvkeI=</latexit>

observed datasource wavelet

<latexit sha1_base64="7fCBr4wiH7EWOTWZxVzjcscUfuk=">AAAB7HicbZBNSwMxEIZn61etX1WPXoJFrJeyK4I9Frx4rOC2hXYp2TTbhmazSzIrlNLf4MWDIl79Qd78N6btHrT1hcDDOzNk5g1TKQy67rdT2Njc2t4p7pb29g8Oj8rHJy2TZJpxnyUy0Z2QGi6F4j4KlLyTak7jUPJ2OL6b19tPXBuRqEecpDyI6VCJSDCK1vL1ZRWv+uWKW3MXIuvg5VCBXM1++as3SFgWc4VMUmO6nptiMKUaBZN8VuplhqeUjemQdy0qGnMTTBfLzsiFdQYkSrR9CsnC/T0xpbExkzi0nTHFkVmtzc3/at0Mo3owFSrNkCu2/CjKJMGEzC8nA6E5QzmxQJkWdlfCRlRThjafkg3BWz15HVrXNc/yw02lUc/jKMIZnEMVPLiFBtxDE3xgIOAZXuHNUc6L8+58LFsLTj5zCn/kfP4A1oyOAA==</latexit>

L =1

2||s(t)� h(t) ⇤ r0(t)||22

<latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit><latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit><latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit><latexit sha1_base64="6EL/lAblpR2bu8Il9voErtGIn0E=">AAACFHicbVC7TsMwFHXKq5RXgJHFokIUEFUSIdEFqRILA0ORaIvUhshxndaq85DtIFVpPoKFX2FhACFWBjb+BqfNAC1Hsnx8zr26vseNGBXSML61wsLi0vJKcbW0tr6xuaVv77REGHNMmjhkIb9zkSCMBqQpqWTkLuIE+S4jbXd4mfntB8IFDYNbOYqI7aN+QD2KkVSSo59cwwvY9TjCiZkmVgrHYygq8giewoG6jvlh9hiPHevecvSyUTUmgPPEzEkZ5Gg4+le3F+LYJ4HEDAnRMY1I2gnikmJG0lI3FiRCeIj6pKNogHwi7GSyVAoPlNKDXsjVCSScqL87EuQLMfJdVekjORCzXib+53Vi6dXshAZRLEmAp4O8mEEZwiwh2KOcYMlGiiDMqforxAOkEpIqx5IKwZxdeZ60rKqp+M1ZuV7L4yiCPbAPKsAE56AOrkADNAEGj+AZvII37Ul70d61j2lpQct7dsEfaJ8/VeCbLg==</latexit>

Kim & Nakata, 2018, TLE

L =1

2||rs(t)� h†

⇥(t) ⇤ ss(t)||22

<latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">AAACKHicbZDJSgNBEIZ73I1b1KOXxiCooMwMgrmIghcPHiIYFTJx6OnUJE16FrprhDCZx/Hiq3gRUSRXn8TOcnAraPj4/yqq6w9SKTTa9sCamp6ZnZtfWCwtLa+srpXXN250kikOdZ7IRN0FTIMUMdRRoIS7VAGLAgm3Qfd86N8+gNIiia+xl0IzYu1YhIIzNJJfPr2kJ9QLFeO5U+RuQft9qny9i3v0gHbuc6/F2m1Qhe9ddwCZ0ff12O73fffe9csV+9AeFf0LzgQqZFI1v/zqtRKeRRAjl0zrhmOn2MyZQsElFCUv05Ay3mVtaBiMWQS6mY8OLeiOUVo0TJR5MdKR+n0iZ5HWvSgwnRHDjv7tDcX/vEaGYbWZizjNEGI+XhRmkmJCh6nRllDAUfYMMK6E+SvlHWZSQ5NtyYTg/D75L9y4h47hq6PKWXUSxwLZIttklzjkmJyRC1IjdcLJI3kmb+TderJerA9rMG6dsiYzm+RHWZ9fS4CkQA==</latexit><latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit><latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit><latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit>

Mapping operatorsynthetic data

Page 48: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Synthetic training set

Kim & Nakata, 2018, TLE

L =1

2||rs(t)� h†

⇥(t) ⇤ ss(t)||22

<latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit><latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit><latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit><latexit sha1_base64="uaaUPypqo0nSupLg1TlHgiYe5XQ=">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</latexit>

Page 49: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Resolution test

Kim & Nakata, 2018, TLE

True model

Least squares

Input seismic data

Neural network

Page 50: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Field data example

Kim & Nakata, 2018, TLE

Least squaresInput seismic data Neural network

Page 51: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Studies for human-induced seismicity

source

path

site

TR

GmRTM

Source parameters of micro earthquakes

0.0

1.0

Dept

h (k

m)

0.0

3.0

2.0

1.0

4.0

Easting (km)

0.0

8.07.0

6.05.0

4.03.0

2.01.0

Northing (km)

0

10

5

-10

-5 P ve

locit

yflu

ctua

tion

(%)

3D/4D High-resolution structural imaging

Dec 11 Dec 21 Dec 31 Jan 10 Jan 20-1

-0.5

0

0.5

1

Velo

city

cha

nge

(%) observed

predicted

Near surface velocities and their dynamics

Publications:http://www.mit.edu/~nnakata

Page 52: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic
Page 53: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic
Page 54: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic
Page 55: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Time

Time

Ambient noise correlation

Processing

- crosscorrelation- spectral

whitening- time and/or space

averaging- normalization- etc…

= ambient-fieldcorrelationseismic interferometry2

Page 56: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Small EQs occur more frequently than large EQs

Gutenberg-Richter law (b value)

log10 N = a� bMData

Prediction

N

M

a

b

Page 57: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Amplitude

Ambient noiseObserved data

(snapshot)

Long Beach array (2012)

Data courtesy:Signal Hill Petroleum, NodalSeismic4 km

8 km

2500 receivers

source

path

site

Page 58: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Amplitude

Data mining to extract coherent wavesObserved data

(snapshot)

Amplitude

Amplitude

Rayleigh waves P waves

Correlation analysis

Nakata et al., 2015

Page 59: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Amplitude

Velocity estimationObserved data

(snapshot)

Near-surface S velocity 3D P velocity

Correlation analysis

Nakata et al., 2015

0.0

1.0

Dept

h (k

m)

0.0

3.0

2.0

1.0

4.0

Easting (km)

0.0

8.07.0

6.05.0

4.03.0

2.01.0

Northing (km)

0

10

5

-10

-5 P ve

locit

yflu

ctua

tion

(%)

Page 60: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

High-frequency ground motion prediction

Time (s)0 2 4 6 8 10 12

Nor

mal

ized

pow

er

0

0.2

0.4

0.6

0.8

1 earthquake enveloperadiative transfer (P)radiative transfer (S)

Nakata & Beroza, 2015

8-16 Hz

Page 61: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

source

path

site

Groningen Gas Field60

km

40 km

Spica, Nakata et al, 2018ab, SRL

• 415 stations• 3 components• Receiver spacing: 400 m• One-month continuous record

Page 62: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Groningen Gas Field60

km

40 km

Spica, Nakata et al, 2018ab, SRL

Near-surface velocity model estimated from• H/V• Rayleigh and Love waves

Page 63: Towards high -resolution seismic imaging and monitoring Nori Nakata s.pdf · Towards high -resolution seismic imaging and monitoring using ambient noise, machine learning, and microseismic

Groningen Gas FieldHourly seismic velocity change

60 k

m

40 km

Nakata, et al., in prep