1. A part of a system that discriminates against some of the information entering it. The discrimination is usually on the basis of frequency, although other bases such as wavelength, moveout, coherence, or amplitude may be used. See Figure F-8.
2. Linear filtering is called convolution (q.v.). A linear filter may be characterized by its impulse response or by its frequency-domain transfer characteristics (amplitude and phase response as a function of frequency).
3. Alias filters (q.v.) are very sharp high-cut filters designed to prevent aliasing.
4. Band-pass filters are often specified by listing their low-cut and high-cut component filters. Filter characteristics are often specified by the frequencies at which the amplitude is down by 3 dB (70% or half power) and by the slope of the cutoff. Thus ‘‘14/18–56/36’’ specifies a band-pass filter with a low-cut down 3 dB at 14 Hz with an 18 dB/octave slope and a high-cut down 3 dB at 56 Hz with a 36 dB/octave slope. Typical seismic filter curves are shown in Figure F-9. See also high-cut filter (=low-pass filter) and low-cut filter (=high-pass filter). The order of the specification is sometimes reversed.
5. Notch filters sharply reject a very narrow band of frequencies.
6. Digital filters permit filtering in accord with arbitrarily chosen characteristics that might prove difficult or impossible to achieve with physical circuit components. Filtering can be accomplished by optical methods as well as by electrical and digital methods.
7. Specific types of filters used commonly include the Butterworth filter (q.v.), a band-pass filter design with flat response, and the Chebychev filter (Tchebyscheff filter), a band-pass filter with a steep rolloff, characterized by a uniform ripple in the pass band. 8. See also inverse filter . Stacking is a filtering process.
9. To remove solids from a suspension by passage through a sieve.