Sensitivity Specifications in Broadband Seismometers

Note: This text is a brief web-site only summary. The full article in is available as a PDF from here .

A modern force-feedback broadband seismometer is a complex precision electro-mechanical instrument capable of measuring a wide range of ground motion signals. The best instruments are capable of measuring extraordinarily small ground motions from distant teleseismic events as well as vibrations generated by moderate-to-large regional and small local earthquakes. Most manufacturers will have a number of models distinguished by different specifications for performance, interfaces, operating environment, footprint, and features. In this article we focus on the sometimes misunderstood “sensitivity” specification and show that its significance is about how to interpret the signal, not whether a particular sensor performs better or worse than another. 

At first glance, it would seem that “sensitivity” should be a legitimate measure of performance.  The word seems to suggest something about the ability of the instrument to detect faint signals, which is very desirable in a broadband seismometer. For example one might say, “The Trillium 240 is so sensitive that it can measure the natural vibration modes of the earth.” This suggests that more sensitivity is better and that instruments with higher sensitivity must be better than instruments with lower sensitivity. 

The problem is that “sensitivity” has more than one definition, and this can lead to confusion.   One meaning is “the minimum input signal required to produce a discernable output signal”. By this definition, instruments that are more sensitive can detect weaker signals than less sensitive instruments. However, sensitivity can also mean “input-output gain” or the ratio of the output (volts) to the input (ground motion velocity in meters/second), and it is this input-output sensitivity scaling factor that is used in seismometer specifications. Seismic monitoring systems use this scaling factor to translate the output signal voltage into meaningful units of ground motion.  

With adjustable gain 24-bit digitizers, the choice of a specific input-output sensitivity value is unimportant. However, knowing the accurate sensitivity value is very important. Some manufacturers measure the sensitivity for each individual instrument, or even each separate channel, and provide that information with every instrument sold. Nanometrics takes a different approach, instead calibrating each instrument at the factory so that the sensitivity is consistent from channel to channel and unit to unit. The sensitivity of the Trillium 240, 120 and 40 models are guaranteed to be within +/- 0.5% of their published specifications. Both approaches are equally valid, but it can be more convenient to manage a network of instruments that all have the same sensitivity, simplifying meta-data tracking and data analysis.

If the actual input-output gain is unimportant, are there any disadvantages to specifying a preferred sensitivity value? There are two answers to this. Firstly, sensitivity is an outcome of a design process that optimizes noise floor, clip level, power consumption, voltage ranges and other factors, and although it may be possible for a manufacturer to alter sensitivity, the design changes will affect one or more performance parameters negatively (e.g. lower clip level, or higher noise floor). Secondly, imposing an arbitrary requirement that has no bearing on performance or functionality will limit the choice of suitable instruments. This may result in inadvertently excluding an instrument that is otherwise the best fit for the intended purpose.

If sensitivity isn’t a useful indicator of a seismometer’s ability to measure ground motion, what are the important performance parameters? The two key performance specifications that indicate the range of signals a seismometer is capable of measuring are “self-noise” and “clip level”. The self-noise specification helps address the question: “What is the weakest ground motion the seismometer can measure?” Self-noise is the expected level of signal the seismometer produces within itself even if there is no ground motion at all. This is usually specified as a graph called a “power spectral density” plot showing the typical level of noise as a function of frequency. The clip level specification answers the complementary question: “What is the strongest ground motion the sensor can measure?” Broadband velocity seismometers usually specify the maximum velocity the sensor will accurately measure in millimeters per second. Together, the clip level and self-noise floor put upper and lower limits on the range of signals a seismometer can measure. 

Any emphasis on seismometer input-output sensitivity is a largely historical issue, meaningful primarily for passive sensors coupled to low-resolution digitizers or analog recording devices with limited dynamic range. The advent of active broadband seismometers with their higher output voltage ranges and 24-bit digitizers with built-in adjustable gain have made the sensitivity of a seismometer relatively unimportant. The best specifications for broadband seismology instrumentation systems emphasize clip-level and noise floor, putting sensitivity in its proper place—its exact value is unimportant as long as it is accurately known.

Note: This text is a brief web-site only summary. To read the full article, please click here .