Lecture — contents 1 Introduction — Activity data

Lecture - contents 1. Introduction - Activity; data processing; approach 2. Radon and earthquakes in the DSR 3. Rn as a proxy of subtle geodynamics - other indicators 4. Conclusions and implications 1

Radon - as a geophysical tracer • Ultra trace gas in geogas (== “air” in subsurface porosity) • Noble gas • Radioactive • Easily measurable with high sensitivity using electronic systems • Extremely large variations in space and in time • A unique combination a unique tool

Local stress/strain, inducing minute changes in rocks (source), enhances release of radon into the geogas environment. T 0 This radon is available for transfer from source to detector DL/L = 0 T 1 Detector s s Advection DL/L= 10 -7 -10 -10 3 Source

Measurement principles U-238 radioactive decay series Decay → recoil → Rn emanation Detection g a a Solid Geogas 3. 82 days 4 1620 years

Earthquakes 1900 -1990 Monitoring sites along Dead Sea Transform 5

Radon monitoring arrays along the Dead Sea Rift (DSR) NW Dead Sea • • Array of stations covering a 20 km sector Next to main western DSR active fault trace 1. 5 m deep in unconsolidated gravel Monitoring since 1994 Intraplate • Depth: 1. 2 m & 90 m • Massive syenite Southern sector of DSR • Array of stations covering a 20 km sector • Precambrian basement rocks of uplifted boundary blocks of DSR 6

NW Dead Sea 19 W – 19 E 17 W 23 W – 23 E Ramon 21 W Arava margin BGO Roded E 1, E 2, E 3 7 IUI

C High Rn zone Monitoring Rn (gamma) sensor Integration time: 15 min 1994 -2005 8

9

E W Rn monitoring at 1. 2 meter in gravel Dead Sea Graben fill U bearing phosphorite 10 u u Rn on carrier gas u u

Alpha and gamma co-registration • Radon ! • Geophysical sensitivity - High 11

Multi-year, seasonal (and multi-day [MD]) variation signatures and signals Days since 1. 1. 1992 Radon time series of (gamma radiation) in geogas in gravel at main monitoring site, NW Dead Sea, DSR. 1. Multi-year decrease (relative to stable background originating from solid gravel) 2. Seasonal variation 3. 12 Multi-day variations (MD) - statistically correlated to earthquakes in the Dead Sea Rift (Steinitz et al. , 2003; see below).

Seasonal, Multi-day (MD) and Diurnal Radon Signals (DRS) Concordance & correlation of signals • Sites 12 km apart • Next to the main western boundary fault of DSR 13

NW Dead Sea -15 km sector Correlation of MD radon signals among three sites • Depth: 2 meters • Lithology: gravel 14

Radon signal at site 17 W • 30 days • varying gamma signal in the geogas. • composed of a multi-day variation (MD) and a superimposed diurnal signal (DRS). A – Measured signal and the smoothed signal representing the multi-day variation (MD) B – Separated diurnal variation 15

Average annual Rn concentration vs. Earthquakes Days since 1. 1. 1992 Multi-year time series of radon (gamma radiation) in geogas in gravel at main monitoring site, NW Dead Sea, DSR. 16

NW Dead Sea, 1995 -2004: • Average annual Rn concentration • Annual number of earthquakes in the DSF Conclusion (1994 -2004) Relationship found between: Annual average Rn level and Annual number of EQ along DSR 17 (IJES 2005)

For (1994 -2004) Relationship found between: Annual average Rn level and Annual number of EQ along DSR The relation between 400 km • MD radon signals (at site 17 W) & 200 km 18 • earthquakes along the Dead Sea Transform

TECTONIC SEGMENTS Earthquake Catalog Seismological Div. , GII 1994 -2002 1075 earthquakes, 4. 2≥ ML ≥ 0 M> 2 M<2 19

Extraction “start” of MD Rn signal, time windows, Earthquakes Smoothing: 25 -hour sliding average Start-time of MD radon signal Threshold: Relative amplitude > 1. 9 Bin = time window (40 Days) minima 20

Correlation between Rn MD signals and EQ in DSR (Geology 2003) For: ML>=2 RA = 1. 9 8 Years: 1995 -2002 No. EQ(ML≥ 2): 165 21

Steinitz et al 2003 Earthquakes are clustered in the 0 -3 days after the starttime of MD Rn signal Number of earthquakes Statistical significance Probability (%) of random occurrence Dead Sea, Kinneret and Hula pull-apart grabens Days after start-time of radon event Timing of 165 earthquakes (ML 2) in the pull-apart grabens of the Dead Sea Rift (Dead Sea, Hula+Kinneret) – relative to the start-time of a radon MD signal. 22 (Steinitz, Begin, Gazit-Yaari, Geology 2003)

Previous approach focused on: Counting earthquakes within multi-day Rn anomalies (Steinitz et al. , 2003, Geology 31: 505 -508) New approach focuses on: Counting days of earthquakes and Rn anomalies (unpublished) 23

MD-Starts and EQ correlation Rn time series at 1 -hour resolution Smoothing: 25 -hour sliding average Residual time series EQ Catalog Regional sets 24 Smoothed time series Extraction of MD starts, Amplitude and RA MD-starts EQ queries

Rn Start-times 25

3 days 26 Rn Start-times for Relative Amplitude > 1. 9 Earthquakes

Flowchart for MD-Starts and EQ correlation Radon time-series Tectonic segment Number of EQ in tectonic segment Extraction of significant “starts” (n ~ 150) Number of measurement days (1995 -2004) 3 & 4 -day time window (bin), Relative to “start” Expected number of EQ per: a) (1 -day) b) 3 & 4 -day time window (bin) All “starts” Count: number of EQ in time window (EQ in time-bin)n starts Histogram: number of EQ in 3 & 4 -day bin 27 All bins

Definition of “Rn anomaly days” for a time bin of n=3 after the start time of Rn anomalies 28

A day is characterized by two attributes: Y N 1) It is a day in which at least one earthquake (of magnitude ≥ML) occurred (or not) 2) It is a day which occurred n days after the start time of a Rn anomaly, with a certain Relative Amplitude (or not) 29 Y N Y N

Are these two attributes independent ? Use the 2 test to determine the probability of random occurrence Did at least one earthquake occur in day? Yes (One degree of freedom) No Observed count day-start Was day within Yes Expected n days after Observed start of No Rn anomaly? Expected Total number of days 30 Total number of days (diff) count EQ (diff) Ntotal

Analyzing the Rn-EQ connection 1. For earthquakes out of the Dead Sea rift valley Rn monitor 31

For earthquakes 1994 -2004, ML >0 Out of the Dead Sea rift valley, Rn anomaly cutoff of Rel. Amp: 2. 0 Did at least one earthquake occur in day? Yes No Was day within Yes 3 days after start of No Rn anomaly? Total number of days 32 Total number of days 342 3094 607 2829 3436

For earthquakes 1994 -2004, ML >0 Out of the Dead Sea rift valley, Rn anomaly cutoff of Rel. Amp: 2. 0 Did at least one earthquake occur in day? Yes Was day within Yes 3 days after start of No Rn anomaly? Observed Expected Total number of days 33 57 60. 4 No Total number of days 285 281. 6 342 550 2544 546. 6 2547. 4 3094 607 3436 2829

For earthquakes 1994 -2004, ML >0 Out of the Dead Sea rift valley, Rn anomaly cutoff of Rel. Amp: 2. 0 Σ [ 2] = 0. 19 * One degree of freedom Was day within Yes 3 days after start of No Rn anomaly? Observed Expected Total number of days 35 Did at least one earthquake occur in day? Yes 57 60. 4 No Total number of days 285 281. 6 342 550 2544 546. 6 2547. 4 3094 607 3436 2829 * Including the Yates continuity correction

For earthquakes 1994 -2004, ML >0 Out of the Dead Sea rift valley, Rn anomaly cutoff of Rel. Amp: 2. 0 Σ [ 2] = 0. 19 Probability of random occurrence = 0. 66 No significant connection Yes Observed Was day within Expected 3 days after start of Observed No Rn anomaly? Expected Total number of days 36 Did at least one earthquake occur in day? Yes 57 60. 4 No Total number of days 285 281. 6 342 550 2544 546. 6 2547. 4 3094 607 3436 2829

Analyzing the Rn-EQ connection 2. For earthquakes within the Dead Sea rift valley Rn monitor 37

For earthquakes 1994 -2004, ML >0 Within the Dead Sea rift valley, Rn anomaly Relative Amplitude > 2. 0 Did at least one earthquake occur in day? Yes Was day within Yes 3 days after start of No Rn anomaly? Observed Expected Total number of days 38 66 49. 2 No Total number of days 276 292. 8 342 428 2666 444. 8 2649. 2 3094 494 3436 2942

For earthquakes 1994 -2004, ML >0 Within the Dead Sea rift valley, Rn anomaly Relative Amplitude > 2. 0 Σ [ 2] = 7. 03 Did at least one earthquake occur in day? Yes Was day within Yes 3 days after start of No Rn anomaly? Observed Expected Total number of days 39 66 49. 2 No Total number of days 276 292. 8 342 428 2666 444. 8 2649. 2 3094 494 3436 2942

For earthquakes 1994 -2004, ML >0 Within the Dead Sea rift valley, Rn anomaly Relative Amplitude > 2. 0 Probability of random occurrence = 0. 008 Significant connection Was day within Yes 3 days after start of No Rn anomaly? Observed Expected Total number of days 40 Did at least one earthquake occur in day? Yes 66 49. 2 No Total number of days 276 292. 8 342 428 2666 444. 8 2649. 2 3094 494 3436 2942

Analyzing the Rn-EQ connection We now test a Rnearthquake connection within the Dead Sea rift valley for 3 days before the start time of Rn anomalies Rn monitor 41

For earthquakes 1994 -2004, ML >0 Within the Dead Sea rift valley, Rn anomaly Relative Amplitude > 2. 0 Σ [ 2] = 0. 98 Did at least one earthquake occur in day? Yes Was day within Yes 3 days before start of No Rn anomaly? Observed Expected Total number of days 42 55 48. 5 No Total number of days 282 288. 5 337 439 2660 445. 5 2653. 5 3099 494 3436 2942

For earthquakes 1994 -2004, ML >0 Within the Dead Sea rift valley, Rn anomaly Relative Amplitude > 2. 0 Probability of random occurrence = 0. 32 No significant connection Was day within Yes 3 days before start of No Rn anomaly? Observed Expected Total number of days 43 Did at least one earthquake occur in day? Yes 55 48. 5 No Total number of days 282 288. 5 337 439 2660 445. 5 2653. 5 3099 494 3436 2942

10 years; 1994 -2004 Earthquakes, 4. 2≥ ML ≥ 0 Twelve 4 -day time bins around “start” TECTONIC SEGMENTS OUT-of. DSR 44 M> 2 M<2 DSR

Testing for correlation between Rn MD signals and EQ in DSR OUT-of-DSR Observed no. of earthquakes & Expected number Enrichment of earthquakes Testing the statistical significance of enrichment Or Probability that correlation is a random one (using the 2 criterion) 10 Years (1995 -2004) 45 RA = 1. 8; 2. 0 bins: 4 days span: -24 to +24 days relative to “start” Earthquakes: ML≥ 0; ML≥ 2

Observed - Expected ML >=2 ML >=0 46 ML >=2 ML >=0

Enrichment 47

Statistical significance 48

Conclusions: Earthquakes within the Dead Sea rift valley (but not out of it) • significantly occur within several days after the start of Radon anomalies, as recorded in the Dead Sea 17 W monitor, (but not before them) [ The daily probability of earthquake occurrence in “Rn-Anomaly days” increases with the increase in the cutoff value of Relative Amplitude of the Rn anomalies ] 49

Preliminary explanation 1. A transient strain causes increase in Rn flux near the 17 W monitor. 2. This strain may cause an earthquake to occur several days later, somewhere within the Dead Sea rift valley. 3. The higher the strain, the higher is the transient Rn flux, and the higher is the probability of an earthquake occurrence in “Rn anomaly” days, relative to other days. 50

Summary of results of 10 years of high-resolution Rn monitoring: 51

1. This study presents a significant statistical relationship between Rn flux and earthquakes that occur within the same tectonic province on an annual basis. 2. This study also presents a significant statistical relationship between Rn anomalies and earthquakes that occur after the start time of the anomalies within the same tectonic province. 52

END Thank You 53