Drilling at the H2O site provides a unique opportunity to observe drilling-related noise from the JOIDES Resolution on a seafloor seismometer in the frequency band of 0.1-80 Hz. (See "Hawaii-2 Observatory" in the "Leg 200 Summary" chapter for background material on H2O and the retired AT&T oceanic cable that is used to provide continuous, real-time data transmission back to the Makaha cable station on Oahu.)
The University of Hawaii operates an OBSS composed of a Guralp CMG-3T three-component broadband seafloor seismometer and a conventional 4.5-Hz three-axis geophone at H2O (Duennebier et al., 2000, 2002). Data are acquired continuously and are made available to scientists worldwide through the IRIS Data Management Center in Seattle. During the cruise, Jim Jolly and Fred Duennebier at the University of Hawaii relayed sample data files to the JOIDES Resolution by FTP via satellite communication. We were then able to process data and study correlations with on-site activities and weather. The University of Hawaii also maintained a Web site showing seismic data from H2O during the cruise (www.soest.Hawaii.edu/H2O/).
All of the data we show here are from the conventional 4.5-Hz three-axis geophone in digital counts. Sensitivity and transfer functions to get output in ground motion units were not applied. All times in this section are given in UTC, which is equal to local time + 9 hr. Occasionally days are represented by the Julian day, the consecutive number of the day in the year.
The objective of this report is to present an overview of the seismic behavior and some of the natural and man-made noise sources at the site. For more information on seismic ambient noise levels in the ocean see Webb (1998); for an introduction to earthquake seismology see Lay and Wallace (1995).
Figure F89 shows a vertical component spectrogram for the 1-week period from 22 to 28 December (Julian days 356 to 362). A spectrogram is a display of energy levels as a function of frequency vs. time. In this case the frequency range of interest is 0.1-10 Hz. In this band, sea state (the gravity waves on the surface of the ocean) is the dominant source of ambient noise.
It has been shown that the microseism peak, the broad vertical red band at frequencies from 0.2 to 0.3 Hz, is created by nonlinear wave-wave interaction of surface gravity waves (Longuet-Higgins, 1950). This peak is a ubiquitous feature on all terrestrial seismograms and is observed at stations deep within the continents. It is interesting to note that the amplitude of this peak is not dramatically greater for seafloor stations than for some land stations.
This spectrogram also shows two and a half storm cycles. These are the broad red bands that slope upward to the left from ~1 to 0.2 Hz over 1.5 to 2 days. They terminate at the microseism peak. The model for this phenomenon is a steady wind creating local waves. Imagine the wind blowing steadily over a calm sea. Initially small waves with short wavelengths and relatively high frequencies are generated by the wind. As the wind continues to blow the waves get larger, longer in wavelength, and lower in frequency. Often, the intervals when the JOIDES Resolution was WOW correspond to the later times in the evolution of this noise.
The thin, constant-frequency red bands near 1.1 and 2.3 Hz in Figure F89 correspond to resonances in the thin sediment cover at this site (Zeldenrust and Stephen, 2001). These bands are another ubiquitous feature observed on seafloor seismometers either on or in sediment layers. Their frequency will depend on the sediment thickness and velocity structure local to the station, but for a given station the frequencies are constant. The resonances are observed as bands in the ambient noise field and as ringing after impulsive signals. More resonant frequencies are apparent in the horizontal (x) component spectrogram (Fig. F90). A complete explanation for the frequency and relative amplitude of these resonances is still in progress. The major reason for installing broadband seismometers in boreholes on the seafloor is to attenuate the effects of these sediment resonances. Ambient noise spectra from the OSN pilot experiment (Figs. F24, F25) show that these resonances are much more pronounced on the seafloor and shallow buried sensors than on the borehole sensor.
Whales are a biological seismic source. Figure F91 shows a sample of a whale song as we arrived at Site 1224 on 26 December. This figure shows a time history of the vertical component of seafloor acceleration in 30-s segments for 2.5 min near 1550 UTC on 26 December. The largest-amplitude events are whale songs occurring in wave packets of four wavelets about once every 30 s. The four wavelets, separated by 3 to 7 s, correspond to the sound traveling directly from the whale to the seafloor plus multiple bounces (echoes) of the sound in the water column.
Figure F92 is formatted similarly to Figure F91 but covers a 25-min time interval. Water gun arrivals (see "Water Gun") are observed in the first 10 min. The rest of the time series is punctuated with whale calls, except for the two bands of three traces each shown in red. A characteristic feature of the whale songs is that they stop every 15 to 20 min while the whale breathes. In this case, the whale sings for 15 min, takes a breath for 1.5 min, and then repeats the process.
Figure F92 shows the last 10 min of the water gun shooting that we conducted as we approached the site on 26 December. The gun was fired every 10 s. The intermediate-amplitude event ~6 s after the primary event is the water multiple. Note that the repetition rate, amplitude, and frequency content of the water gun and whale song are remarkably similar.
The principal motivation behind drilling at the H2O is to provide a high-quality seismic station for the Global Seismic Network. Some small earthquakes did occur while we were on site. A quick and easy way to scan all of the data continuously is to display root-mean-square (RMS) energy levels in one-octave bands as a function of time. An example for the horizontal (x) component on 7 January is shown in Figure F93. In this example, most of the variability during the day is occurring in the octave centered at 8 Hz. The major event occurring between 10 and 15 hr is a drilling-related effect discussed below. The large peaks near 5 and 20 hr can be identified as earthquakes.
Time series and spectra for the event near 21 hr are shown in Figure F94 The event has a duration of ~20 s and has a broad frequency content. Note that the energy level of the microseism peak near 1 Hz does not increase with the arrival. The energy level of the sediment resonances near 2.8, 4.1, and 5.7 Hz, however, increases by up to 20 dB (a factor of 10 in amplitude).
A second earthquake example is shown in Figure F95. The arrival in this case is spread over a longer time interval, and there is no detectable energy below the microseism peak.
Shipping is a major man-made source of noise in the ocean. Figure F96 shows an RMS summary of the horizontal (x) component for 25 December. The RMS level in the octave centered at 8 Hz increases by 40 dB from 5 to ~12 hr and decreases again at ~15 hr. This event can also be seen in Figure F89 halfway through 25 December (Julian day 359) near 8 Hz. This event is characteristic of a large ship approaching and then leaving the site. The energy is focused near 8 Hz, which is an indication of some type of machinery. This is a very large sound source. If the ship passed directly over the site traveling at ~20 kt, it was affecting noise levels at the station while it was 200 km away. (The large-amplitude peaks in the other frequency bands are an artifact of the processing.) The passage of a container ship bound for Honolulu on 25 December traveling at 17 kt was confirmed by the bridge (P. Mowat, pers. comm., 2001).
In contrast, the JOIDES Resolution is much quieter in this frequency band. While the JOIDES Resolution steamed directly over the site and we fired the water gun at 1500 UTC on 26 December, the RMS level in the 8-Hz octave increased <10 dB.
Without further processing, some drilling-related activities can be identified at the seismic station. In Figure F89, for example, the yellow blotches between 2 and 9 Hz on 26-28 December (Julian days 360 through 362) show some correspondence to drilling activity. The bright yellow band at almost exactly 6 Hz in the second half of 27 December (Julian day 361) corresponds to running pipe and is likely the noise of the drawworks. Drawworks noise is also seen between 1000 and 1500 hr on 7 January (Fig. F93). In Figure F90, the high-amplitude (red) regions from 1 to 9 Hz on 4 and 5 January correspond to drilling with the RCB bit.
On 11 January, drilling progress was slow, and we pulled the pipe to see what was wrong. The bit had fallen off in the hole during drilling. Figure F97 shows the noise record at the H2O seismometer (1.5 km away) during this critical period. From 0 to 4 hr the noise levels are typical for the 14
-in bit. The dropout at 0400 hr occurred when drilling was stopped to circulate a pill. Excessive torque was observed at ~0500 hr, and the pipe was picked up (the weight on bit was zero) at 0620 hr. At 0800 hr there was a pressure drop, but the dramatic change in noise levels between the 4- and 8-Hz octaves occurred at ~0930 hr. We infer that 0930 hr is when the bit broke off the end of the pipe.
Because of the cost of transmitting the data files to the ship, we only looked at a small subset of the available data. The primary data aboard ship were horizontal (x) component only. By comparing the samples in Figures F89 and F90 one can see that there are significant differences between horizontal and vertical components. For most intervals the "day files" had been decimated from 160 to 20 samples per second, so high-frequency (9-70 Hz) studies could only be conducted over very small time intervals such as that of the water gun shooting shown in Figure F92. Furthermore, we only used geophone data. By using data from the Guralp sensor, we can extend the analysis to frequencies below the microseism peak (down to 0.01 Hz).
