Fig. 1: YKA is located in the Canadian north, on thenorth shore of Great Slave Lake near the City ofYellowknife (top). The array now consists of 18short period sensors in a 25 km long cross patternand two broadband stations (bottom).

The Yellowknife Seismic Array (YKA) was commissioned in 1962 by theUnited Kingdom Atomic Energy Association in cooperation with theCanadian Defense Research Board and Dept. of Mines and TechnicalSurveys (now Natural Resources Canada) as one of four experimentalseismological arrays built to investigate the capabilities of such sensorarrays to detect and characterise underground nuclear explosions.

Located in the Northwest Territories of Canada because of its relativelocation to known nuclear test sites at the time, and its remoteness fromcoastlines, urban areas and other sources of cultural seismic noise, thearray is situated on the stable, aseismic Canadian Shield. It’s design wasoriginally comprised of 19 analog short period seismometers in theshape of an offset cross some 25 km in length. Each sensor is separatedby ~2.5 km (Fig. 1). Long period instruments would later be added in1971, along with independent sensor power from thermal electricgenerators (TEGs) and radio communications which replaced earlierproblematic cables.

At the end of 1986, as computer systems became more powerful andcompact, the array underwent a major upgrade which converted theprevious analog systems to a completely digital one. This configurationof YKA would run continuously until April 2014.

Since its inauguration, the Yellowknife seismic array (YKA) has been innearly continuous operation and has detected and recorded the seismicsignatures of many hundreds of global nuclear explosion tests.Meanwhile, the array has remained in the same configuration withrelatively few changes to the original design making the YKA datasetrelatively unique.

Designated as one of the 50 International Monitoring System’s primaryseismic stations (PS09), YKA continues its original mission after morethan 53 years of operation.

Modernization of the Yellowknife Seismic Array (PS09)Modernization of the Yellowknife Seismic Array (PS09)Wayne N. EdwardsWayne N. EdwardsCanadian Hazards Information Service, Natural Resources Canada, Ottawa, Ontario, Canada, K1A 0Y3Wayne.Edwards@NRCan-RNCan.gc.caCanadian Hazards Information Service, Natural Resources Canada, Ottawa, Ontario, Canada, K1A 0Y3Wayne.Edwards@NRCan-RNCan.gc.ca

1. The Yellowknife Seismic Array (PS09)

2. Former 1989 – 2014 Array

3. PS09 Recapitalization Plan

4. New Hybrid Power Systems

5. Vault & Instrument Changes

6. Comparison of New vs. Old

YKAYKAYKA

YKA Central Facilities BuildingYKA Central Facilities Building

White Broadband StationWhite BroadbandStation

Former Red/BlueShort Period Station

Fig. 2Fig. 2

Former Red/BlueShort Period Station

For nearly 25 years YKA operated successfully as adigitally acquired seismic station using the whatwas state of the art in 1989 (Fig 2).Over time, however, this infrastructure had begunto show its age. The 16 bit digitizers that were tiedto a central GOES clock at the Central FacilitiesBuilding (CFB), had been surpassed by 24 bittechnology, and GPS timing. CFB acquisitioncomputers and data servers had long sincebecome antiquated as computers systems becameincreasingly faster, more powerful, smaller andmore efficient. The propane powered TEGs andfuel tanks also increasingly needed servicing andmanual restarts.In many cases, the instrument vaults had becomecorroded and leaked/flooded in the Spring andFall seasons, sometimes rendering theinstruments operational but unserviceable inblocks of ice in the winter.It was clear a major reinvestment in the overallinfrastructure of the array was needed.

Discussions between Canada and the Provisional TechnicalSecretariat (PTS) of the CTBTO for the renewal of YKA werelargely complete by early 2011. The new array wouldcompletely overhaul the systems and infrastructure of the oldarray to ensure that rigorous IMS standards for data qualityand continuity were met, while minimizing the frequency ofonsite visits and effort required by technicians.Overall, the recapitalization included new hybrid remote powersystems, digitizers, vaults, radio communications and GPStiming for each of the 18 elements of the short period array.

Advances in Sensors, Networks and Processing Design of Sensor Systems

Two new IMS compliant three componentbroadband instruments would replace the STS-1sensors at two sites, and sample rates wouldincrease from 20 Hz to 40 Hz for all channels.At the central facility, new acquisitioncomputers would be installed for real-timecommunications with the field stations andforwarding of data to Ottawa (CNDC) andVienna (IDC) with a live backup system in case ofprimary failure.In total the project would take two completefield seasons and three winters to complete,during which time the original array continuednormal operations with no interruptions. Toinsure that the new array was workingsatisfactorily, both new and old arrays operatedside-by-side for ~6 months for data comparison.

During recapitalization each of the 18 short period vaultswere replaced. Many of these were still the original vaultsfrom the 1960s array, consisting of a steel drum blastedand concreted into the bedrock. However, many sufferedfrom corrosion and leaks causing the vaults to be floodedduring Spring and Fall, at times trapping the instrumentsand digitizers in ice. These vaults were replaced (in mostcases right beside the previous vaults) by the insulatedsurface vaults shown in Fig 4.The instruments used at the YKA short period array areTeledyne-Geotech S13 sensors, which were retained fortheir long established robustness under extremeconditions and stable response, while non-IMS compliantSTS-1 broadband instruments were replaced with twoCMG-3T 360s-50Hz. All digitizers were replaced with CMG-DM24S3EAMs.In both cases, sensitivities of the instruments and digitizerswere adjusted to match that of the former YKA. For theS13s, this required purpose-built low noise CMG-ELP 0110pre-amplifiers to be mated with each S13 and the newCMG-DM24S3EAM digitizers.The result are modest modifications to the YKA (PS09)instrument response (Fig. 5).

Fig. 3: Solar-TEG HybridPower system at B7Fig. 3: Solar-TEG HybridPower system at B7

Fig 4. New vs. Old Vaults @ B6Fig 4. New vs. Old Vaults @ B6

The new power systems incorporate both a solarrecharged deep-cycle battery system, with a propaneTEG system. During most of the year, three banks ofsix deep cycle batteries are used to power the eachsensor element of the array, including its radiocommunications to the CFB. The batteries arecontinually charged by six solar panels orientedsouthward on a 25 foot (7.62 m) articulated tower(Fig. 3). Charge controllers ensure proper powerdistribution amongst the batteries.During cold winter months when sunlight isunavailable to recharge the batteries, the propane-powered TEG system is engaged to provide thenecessary power to run each element. In practice,this occurs only during late January and February.Each power systems is located between 30 – 40 maway from the instrument vault, to minimize thelikelihood that seismic noise generated by the towerduring windy conditions will contaminate the seismicdata (Edwards et al. 2012. SRL, 83, 787-797). Batteryvoltage is routinely monitored remotely, along withseveral other state-of-health indicators at the CFB byonsite technicians.

Both new and old arrays operated simultaneously for ~6 months, making it possibleto cross-validate seismic events between the two systems and determine anysystematic changes in the data after the upgrade (Edwards 2014, GSC OF 7559).Teleseismic observations of earthquakes between mb 4 to 7, were collected duringthis period and beams directed at the P-arrivals were computed for both new and oldYKA arrays. Measurements of period, amplitude and final mb magnitude were thencalculated and compared. In general, a ~20% increase in amplitude was observed inthe short period data for the new array, equating to a magnitude shift of +0.1 m.u.(Fig 6).As well, it was observed that the timing of the former central GOES clock wassystematically fast by 0.096 ± 0.008 seconds vs. GPS time (or ~2 samples, Fig 7).

Old STS-1 + MKINew CMG-3T + DM24

Old S13+RD3

New S13+DM24

New CMG-3T + DM24

Form

er N

yqui

st

New

Nyq

uist

Fig.5

Example: Teleseismic P-phase2013/09/02 02:51:13.240 UTC42.2390°N 133.6357°W 442.49 km 5.6 mbDistance: 62.0° Az: 29.1°P-Arrival: 03:00:48.483 UTC

Time (seconds after P-arrival)

Grou

nd M

otio

n (n

m/s

)Gr

ound

Mot

ion

(nm

/s) YKAB7

YKB7

YKAR2YKR2

3 3.5 4 4.5 5 5.5 6 6.5 73

3.5

4

4.5

5

5.5

6

6.5

7

Old Array Magnitude (mb)

New

Arra

y M

agni

tude

(mb)

Comparison of Magnitude Measurements Between Arrays: 365, Events

Fig. 6

Fig. 7