2. AN INERTIAL ROTATION METER (TILTMETER) Hutt, C. R., Holcomb, L. G., and Sandoval, L. D. USGS Albuquerque Seismological Laboratory Surface tilt noise has always contaminated the outputs of horizontal broadband seismometers at long periods at most surface installation sites. Currently, the only method for reducing or eliminating tilt noise is to install the sensors at depth either in boreholes or in deep caves or mines. For many years instrumentation personnel have known that, if a pure tilt meter were available, the output of the tilt meter could be used to remove the tilt noise from the horizontal sensors that are installed on the surface. The lack of a sensor that is capable of measuring only pure tilt has prevented the application of this method. In the spring of 2003, a series of preliminary experiments, that was designed to evaluate the feasibility of building a pure tilt meter, was conducted at the Albuquerque Seismological Laboratory (ASL). Parts from two STS-1 horizontal seismometers were combined to create an instrument with a boom with a nearly symmetrical distribution of mass about a central pivot point. The aim of the design was to create an instrument that was relatively insensitive to either horizontal or vertical acceleration and yet could sense pure rotation about an axis in the horizontal plane. An instrument with this design must have an infinite mechanical free period (assuming zero flexure restoring force) if it is to be truly insensitive to horizontal acceleration. The prototype achieved a mechanical free period of about 13 seconds, which is quite a ways from infinity but still a reasonable length. The modified instrument was operated in close proximity with several unmodified STS-1 horizontal sensors in the ASL vault in an effort to determine whether or not the tilt meter could separate pure horizontal tilt from horizontal acceleration. Initial results have been encouraging. The output signal level of the tilt meter increases under windy conditions as one should expect because wind generates localized tilt of the earth's surface. The output of the tilt meter in response to earthquake input signals is significantly lower than the output of the unmodified STS-1 sensors thereby indicating a significant reduction in the tilt meter's sensitivity to horizontal acceleration. There is significant coherence between the outputs of the unmodified STS-1 sensors and the tilt meter at long periods and relatively simple signal processing has yielded a reduction in the power spectral density levels of the output signal from the unmodified STS-1 sensors thereby indicating that some tilt noise can be removed from the standard sensor output at long periods. Work on the tilt meter development was suspended in June 2003 due to time and budgetary constraints.

3. DUAL TILTMETER FABRICATION AND CHARACTERIZATION There are significant questions yet to be answered concerning the tilt meter performance. There are indications that the current sensor is responding in a poorly understood manner to atmospheric pressure variations. Temperature variations may also be producing false tilt output. Air motion may also be a factor. The next step in the project is to build two tilt meters that are mechanically mounted on the same rigid base to ensure that the tilt and acceleration input to both sensors are identical. The two sensors will be operated in conjunction with several (probably 4) unmodified STS-1 horizontal sensors to assist in determining the fidelity of the tilt signal. More elaborate signal processing procedures will be written and applied to the data to determine how much of the tilt signal in the unmodified sensors can be eliminated by using the output of the tilt meters. Total anticipated costs are $17,015. Cost Breakdown: 1. Raw materials a. Base plate assembly $1,000. b. STS-2 substitute bases $150. c. Sensor counter balance assembly $0. d. Sensor brass counter weight $20. 2. Shop fabrication time a. Base plate assembly (40 hours at $35/hour) $1,400. b. STS-2 substitute bases (16 hours at $35/hour) $560. c. Sensor counter balance assembly (24 hours at $35/hour) $840. d. Sensor brass counter weight (30 hours at $35/hour) $1,050. 3. Assembly and adjustment a. Shop mechanical assembly (16 hours at $35/hour + 16 hrs @ $43/hr) $1,248. b. Vault experiment assembly (20 hrs at $48.50/hr + 20 hrs @ $43/hr) $1,860. 4. Data collection and analysis a. Software development (40 hours at $48.50/hour) $1,940 Data analysis and presentation (40 hours at $48.5/hour) $1,940. Total Net Cost: $12,008 USGS Assessments: 41.70% of Net Costs = $5,007 Total Gross Cost: $17,015

4. A “pure tiltmeter” is one that is sensitive only to tilt and is not sensitive to horizontal or vertical acceleration. One way to accomplish this is to use a “beam balance” configuration, as shown in the upper diagram to the right. The lower diagram is equivalent, and this is the configuration we produced when we combined parts from two STS-1 horizontal seismometers to make an inertial rotation meter.

5. Adjusting tiltmeter free period. The large brass counterweight puts lower flexures under tension.

6. Tiltmeter plus two STS-1 horizontal seismometers on single rigid aluminum plate.

7. The Guilty Parties Gary Holcomb (L) and Leo Sandoval (R)

8. This 13.7 hour segment of LH data shows tilt noise (in the center portion of the plot) generated by the wind blowing. As indicated on the first three traces, the tiltmeter and STS-1H seismometers #1 and #2 are on a single plate of aluminum so that they see the same tilt signal. The other three horizontal seismometers are on the concrete pier next to the aluminum plate.

9. This plot of an earthquake shows that the tiltmeter (top trace) has a much lower response to rectilinear horizontal acceleration from the quake than the horizontal seismometers. The very long period waveform superimposed on the tiltmeter trace is its response to ground tilt. When the horizontal seismometer traces are scaled to 6000 counts full scale (same as the tiltmeter trace), one can see this tilt signal in those traces as well. Since the earthquake signal is small and the tilt signal is large in the tiltmeter data (relative to the H seismometers), there is hope of using the tiltmeter data to process the tilt signal out of the horizontal seismometer data. 6000 counts 200,000 counts

10. When we zoom in on a sample 20-minute segment of data from the tiltmeter plus the two H seismometers on the aluminum plate, we see good visual coherence between the tiltmeter signal and either of the H seismometer signals (and between the two H seismometers as well). The coherence between the H seismometer on the pier next to the plate and any instrument on the plate is not as good. This is an indication that the tiltmeter must be operated on a common baseplate with the horizontal seismometer if we expect both instruments to “see” the same tilt signal. Note that the tiltmeter responds better to low frequency signals (which is where most of the tilt signal spectral power lies) and not as well to higher frequency signals as the H seismometers do.

11. Coherence of two seismometers on common baseplate The upper figure is a plot of the power spectral density from two STS-1H seismometers sitting on the same baseplate with the tiltmeter. The lower figure is the coherence between these two seismometers, indicating excellent coherence beyond 1000 seconds period.

12. Coherence between H seismometer and tiltmeter The upper figure is a plot of the power spectral density from one STS-1H seismometer and the tiltmeter on a common baseplate. Note that the tiltmeter signal power is lower at shorter periods and higher at longer periods than the seismometer. The lower figure is the coherence between the tiltmeter and one H seismometer. While the coherence is not as good as between the two STS-1H seismometers in the previous slide, it is high enough to provide encouragement that some of the tilt signal could be removed from the seismometer signal.