WO2015170116A1

WO2015170116A1 - Improvements in fibre optic distributed sensing

Info

Publication number: WO2015170116A1
Application number: PCT/GB2015/051360
Authority: WO
Inventors: Roger Crickmore; Emery KU
Original assignee: Optasense Holdings Limited
Priority date: 2014-05-08
Filing date: 2015-05-08
Publication date: 2015-11-12
Also published as: GB201408132D0

Abstract

This application describes methods and apparatus for distributed fibre optic sensing that allows for relatively rapid detection of temperature changes or other low frequency effects and/or calibration. The method involves performing a series of interrogations of an optical fibre(101). Each interrogation involves launching at least one interrogating pulse into the optical fibre; and detecting optical radiation which is backscattered from within the fibre –where the light from successive interrogations does not interfere. The method involves applying a predefined variation in frequency (Δf) between the pulses of different interrogations and determining any variation in backscatter intensity arising from such predefined variation in frequency. By varying the frequency of the interrogating pulses the operating point of a given channel is deliberately modulated. In the absence of any environmental stimulus this will lead to a repeated pattern of intensity variation, which could be used to determine the operating characteristics of the relevant channel. However any low frequency change will be detectable as a modulation on this intensity variation which can be detected.

Description

Improvements in Fibre Optic Distributed Sensing

This application relates to fibre optic distributed sensing, and especially to methods and apparatus for detecting low frequency variations such as temperature or strain variations and/or for improving the consistency of response.

Fibre optic sensing is a known technique where an optical fibre, deployed as a sensing fibre, is interrogated with interrogating radiation and radiation which emerges from the fibre is detected and analysed to determine environmental changes acting on the optical fibre. Some fibre optic sensors rely on deliberately introduced features within the fibre, e.g. fibre Bragg gratings or the like, to induce reflection from a point in the fibre. In a fibre optic distributed sensor however the radiation which is backscattered from inherent scattering sites within the fibre is detected. The sensing function is thus distributed throughout the fibre and the spatial resolution and arrangement of the various sensing portions depends on the characteristics of the interrogating radiation and the processing applied.

Fibre optic sensors for distributed temperature sensing (DTS) are known which rely on detecting light which has been subjected to Brillouin and/or Raman scattering. By looking at the characteristics of the Brillouin frequency shift and/or the amplitudes of the Stokes/anti Stokes components the absolute temperature of a given sensing portion of fibre can be determined. DTS is a useful technique with a range of applications but most DTS systems require relatively long time averages to provide the desired accuracy, meaning such DTS systems are less useful for detecting relatively rapid changes in temperature.

Fibre optic sensors for distributed acoustic sensing (DAS) are also known. Various types of DAS sensor have been proposed including sensors based on Rayleigh scattering of light from the sensing fibre. Light transmitted into an optical fibre will be scattered from the various inherent scattering sites within an optical fibre. A

mechanical vibration of the fibre, such as caused by an incident acoustic wave, will alter the distribution of scattering sites resulting in a detectable change in the properties of the Raleigh backscattered light. Analysing such changes allows relatively high frequency vibrations/acoustic stimuli acting on sensing portions of the optical fibre to be detected. Some DAS sensors analyse the intensity of backscattered light from each of a number of sensing portions of the sensing fibre, also called channels of the DAS sensor, and monitor the change in intensity to determine any acoustic stimulus acting on the fibre. Such DAS sensors have also been employed in a wide range of applications. However the sensing portions or channels of some types of DAS sensor will typically exhibit a non-linear response to a given input acoustic stimulus. Also the relative intensity change in response to a given input stimulus will vary from channel to channel and can also vary for a given channel over time. In other words the gain of the channels is variable.

Embodiments of the present invention aim to provide methods and apparatus for distributed fibre optic sensing which at least mitigate at least some of the above mentioned disadvantages. Thus according to the present invention there is provided a method of distributed fibre optic sensing comprising:

performing a series of interrogations of an optical fibre, each interrogation comprising: launching at least one interrogating pulse of optical radiation into the optical fibre; and for at least one sensing portion of said optical fibre detecting optical radiation which is Rayleigh backscattered from within the sensing portion of said sensing fibre;

wherein the launch rate of said interrogations is such that optical radiation backscattered from one interrogation does not substantially interfere with optical radiation backscattered from another interrogation; and

wherein at least some of the interrogations have a predefined variation in frequency of said at least one interrogating pulse from one another;

the method comprising determining any variation in intensity of said

backscattered radiation resulting from said predefined variation in frequency of the interrogating pulses.

In some embodiments the predefined variation in frequency comprises a cycle of frequency modulation which repeats at a modulation cycle rate. In this case

determining any variation in intensity of the backscattered radiation resulting from the predefined variation in frequency of the interrogating pulses may comprise identifying a first signal in the detected backscatter radiation exhibiting at a frequency equal to the modulation cycle rate. The amplitude of said first signal may be analysed and/or any variation in the amplitude of the first signal over time may be detected. The method may detect when the amplitude of the first signal goes through at least one of a maximum and a minimum. A respective first signal may be generated for each of a plurality of sensing portions of fibre and any variation in amplitude of the first signals from said plurality of sensing portions of fibre may be determined. The method may involve identifying a change in temperature of a sensing portion of the optical fibre by detecting a change in amplitude of the first signal for that sensing portion and/or determine a rate of change of temperature by determining the rate of change of amplitude of the first signal.

The launch rate of said interrogations may be an integer multiple of said modulation cycle rate. The modulation cycle rate may be between 10 Hz and a quarter or a half of the launch rate of said interrogations. Each successive interrogation may have a frequency modulation applied according to said cycle of frequency modulation. Each cycle of frequency modulation may comprise an interrogating pulse at a first frequency and the method may comprise, for at least one sensing portion of said optical fibre, comparing the backscatter from the pulses at the first frequency in successive cycles to provide distributed acoustic sensing. In some instances each cycle of frequency modulation may comprise a plurality of interrogating pulses of different frequencies which are repeated each cycle and the method may comprise, for at least one sensing portion of said optical fibre, comparing the backscatter from said pulses at the same frequency in successive cycles to provide a number of distributed acoustic sensing measurements for the same sensing portion.

The cycle of frequency modulation may comprise a first set of interrogating pulses all at a first frequency followed by at least a second set of interrogating pulses all at a second, different frequency wherein the backscatter from interrogating pulses within at least one set are used for distributed acoustic sensing and backscatter from

interrogating pulses from different sets are compared to determine the optimum set or sets to be used for subsequent analysis.

The cycle of frequency modulation may be applied to a first series of interrogations which is interleaved with a second series of interrogations, wherein the second series of interrogations are not frequency modulated with respect to one another. The method may then comprise for at least one sensing portion of said optical fibre, comparing the backscatter from the interrogating pulses of the second series of interrogations to provide distributed acoustic sensing. Interrogations of the first series and the second series may be alternately launched into said optical fibre. The cycle of frequency modulation applied to the first series of interrogations may involve a frequency modulation about the frequency of the pulses of the second series of interrogations.

In some embodiments the method may comprise, for at least one sensing portion of said optical fibre, comparing the detected intensity from one or more interrogating pulses to provide a measurement signal indicative of any acoustic stimuli acting on that sensing portion. This may involve analysing the variation in intensity of the

backscattered radiation resulting from said predefined variation in frequency of the interrogating pulses to determine at least characteristic of an operating point of the sensor portion. The characteristic of the operating point may comprise a polarity of the variation in intensity compared to at least one of the predefined variation in frequency or the intensity response of another sensing portion to said predefined variation in frequency, in which case the method may involve determining a polarity for each of a plurality of sensing portions and adjusting the measurement signal for that sensing portion based on said polarity. Additionally or alternatively the characteristic of the operating point may comprise a gain value indicating the amplitude of variation and the method may comprise monitoring the gain value for a given sensing portion over time to detect any variation. The gain value for a plurality of sensing portions may be compared to detect any variation. The method may then comprise applying a gain correction to the measurement signals from one or more sensing portions to correct for said variations in gain. In some embodiments the measurement signal may be combined with the polarity and gain values for a plurality of sensing portions to estimate the signal generated on each portion by laser frequency noise.

The method is particularly applicable for relatively rapid detection of temperature changes. Thus in another aspect of the invention there is provided a method of detecting temperature changes in an area of interest comprising:

performing a series of interrogations of an optical fibre deployed, at least partly, in said area of interest, each interrogation comprising: launching at least one interrogating pulse of optical radiation into the optical fibre; and for at least one sensing portion of said optical fibre detecting optical radiation which is Rayleigh backscattered from within the sensing portion of said sensing fibre; wherein the launch rate of said interrogations is such that optical radiation backscattered from one interrogation does not substantially interfere with optical radiation backscattered from another interrogation; and

wherein at least some of the interrogations have a predefined variation in frequency of said at least one interrogating pulse from one another, the predefined variation in frequency comprising a cycle of frequency modulation which repeats at a modulation cycle rate;

the method comprising identifying a first signal in the detected backscatter radiation exhibiting at a frequency equal to the modulation cycle rate; and

detecting any variation in the amplitude of said first signal over time.

This method may be used for detecting a leak in a pipeline by performing the method to detect any sudden changes in temperature, where the area of interest comprises the path of the pipeline.

In a further aspect there is provided a method of distributed fibre optic sensing comprising:

performing a series of interrogations of an optical fibre, each interrogation comprising launching at least one interrogating pulse of optical radiation into an optical fibre and detecting optical radiation which is Rayleigh backscattered from within said fibre to provide a plurality sensing portions;

applying a predetermined modulation at a modulation frequency to the operating point of at least one sensing portion so as to produce in a variation in backscatter intensity from said sensing portion; and

determining the amplitude of said variation in intensity of backscatter radiation at the modulation frequency; and

detecting any variation in said amplitude.

The invention also relates to distributed fibre optic sensor apparatus. Thus in another aspect of the invention there is provided a distributed fibre optic sensor apparatus comprising:

an optical source configured to generate interrogating pulses to be launched, in use, to an optical fibre in a series of interrogations .wherein the launch rate of said interrogations is such that optical radiation backscattered from one interrogation does not substantially interfere with optical radiation backscattered from another

interrogation; a frequency modulator configured to apply a predefined frequency modulation to said interrogating pulses so that at least some interrogations have different frequency interrogating pulses to one another;

a detector for detecting radiation Rayleigh backscattered from within said optical fibre; and

a processor configured to determine, for at least one sensing portion of the optical fibre, any variation in intensity of said backscattered radiation resulting from said predefined frequency modulation of the interrogating pulses. The apparatus of this aspect of the invention provides the same advantages as the methods described above and may be used in any or all of same ways and/or for the same applications as described above.

The invention will now be described by way of example only with respect to the accompanying drawings, of which:

Figure 1 illustrates a conventional fibre optic distributed sensor;

Figure 2 illustrates the propagation of an optical pulse in the sensing fibre;

Figure 3 illustrates the principles of variation in backscatter intensity with bias point; Figure 4 illustrates the effect of different operating points on the output response to a given input stimulus;

Figure 5 illustrates an embodiment of a distributed fibre optic sensor according to the present invention;

Figure 6 illustrates one example of how the frequency of the interrogating pulses may be modulated over time;

Figure 7 illustrates how amplitude variations in the detected measurement signal can be used to track operating point variations;

Figure 8 illustrates interleaving of pulses having a frequency variation with pulses of fixed frequency for DAS; and Figures 9a and 9b shows data from an embodiment of the present invention applied to laser noise identification.

Figure 1 shows a schematic of a general distributed fibre optic sensing arrangement. A length of sensing fibre 101 is removably connected at one end to an interrogator 100. The sensing fibre is coupled to an output/input of the interrogator using conventional fibre optic coupling means. The interrogator unit is arranged to launch pulses of coherent optical radiation into the sensing fibre 101 and to detect any radiation from said pulses which is backscattered within the optical fibre. For a Rayleigh scattering based distributed acoustic sensing (DAS) apparatus the detector will detect radiation which has been Rayleigh backscattered from within the fibre and which is thus at the same frequency as the interrogating radiation. To generate the optical pulses the interrogator unit 100 comprises at least one laser 102. The output of the laser may be received by an optical modulator. Note that as used herein the term "optical" is not restricted to the visible spectrum and optical radiation includes infrared radiation, ultraviolet radiation and other regions of the electromagnetic spectrum. The pulses output from the optical modulator 103 are then transmitted into the sensing fibre 101 , for instance via a circulator 104. The sensing fibre 101 can be many kilometres in length and can be, for instance 40km or more in length. The sensing fibre may be a standard, unmodified single mode optic fibre such as is routinely used in telecommunications applications without the need for deliberately introduced reflection sites such a fibre Bragg grating or the like. Multimode fibre can also be used but the measurements are typically of a lower quality than if a single mode fibre was used. The ability to use an unmodified length of standard optical fibre to provide sensing means that low cost readily available fibre may be used. The optical fibre will typically be protected by containing it with a cable structure. In some embodiments the fibre may comprise a fibre which has been fabricated to be especially sensitive to incident vibrations or the cable structure may have been designed to achieve the same effect. In use the fibre 101 is deployed in an area of interest to be monitored.

Optical radiation which is backscattered from said optical pulses propagating within the sensing fibre is directed to at least one photodetector 105, again for instance via circulator 104. The detector output is sampled by an analogue to digital converter (ADC) 106 and the samples from the ADC are typically passed to processing circuitry 107 for processing (although in theory the base data samples could be output). The processing circuitry 107 may process the detector samples to determine an overall backscatter intensity from each of a number of different channels, each channel corresponding to a different longitudinal sensing portion of optical fibre. In some arrangements the processing circuitry 107 may provide most of the processing to indicate any acoustic signals acting on the channels of the sensor. However in some embodiments the output from interrogator 100 may be passed to an external signal processor (not shown), which may be co-located with the interrogator or may be remote therefrom, and optionally a user interface/graphical display, which in practice may be realised by an appropriately specified PC. The user interface may be co- located with the signal processor or may be remote therefrom.

It will be noted that the interrogator unit may comprise various other components such as amplifiers, attenuators, filters etc. but such components have been omitted in Figure 1 for clarity in explaining the general function of the interrogator.

In amplitude based DAS sensor the laser 102 (and modulator 103 if present) are configured to repetitively produce at least interrogating pulse at a particular launch rate, often called a ping rate. When an interrogating pulse propagates within the optical fibre some light will be scattered from the intrinsic scattering sites within the optical fibre. At least some of this backscattered light will be guided back to the beginning of the optical fibre where it can be detected. At any instant the light arriving at the detector may comprise light scattered from a range of scattering sites distributed through a section of fibre.

Figure 2 illustrates the propagation of a pulse in the optical fibre and shows distance along the fibre against time. Lines 201 and 202 illustrate the leading and trailing edges of the pulse respectively. Thus at time t₀ the leading edge of the pulse enters the optical fibre and at ti the trailing edge of the pulse enters the fibre. The time between t₀ and ti therefore corresponds to the duration of the pulse. The pulse propagates in the fibre at a velocity equal to c/n where c is the speed of light in vacuo and n is the effective refractive index of the optical fibre. In the fibre the pulses will thus have a spatial width, W^ represented by the vertical distance between lines 201 and 202.

As the pulse propagates in the optical fibre some light will be backscattered towards the start of the fibre. This backscattered light will also travel at a velocity equal to c/n. Consider the light reaching the detector. Line 203 represents the trajectory of light which could possibly be received at the start of the optical fibre at a given instant t₂. Any backscattering which occurs at a time and distance into the fibre that lies on line 203 could be received at the start of the fibre at the same instant t₂. Thus it can be seen that light which is scattered at various times from a first section of the fibre as the pulse propagates through that section will be coincident at the start of the fibre (and hence coincident on the detector). It can also be seen that the width of this first section of fibre is equal to half the width of the pulse in the fibre, i.e. W^2. This means that at any instance the backscattered light received at the start of the optical fibre corresponds to backscattering in the fibre from a number of scattering sites distributed through a certain section of fibre. The length of this section of fibre is defined by the pulse width of the interrogating radiation. The minimum spatial size of the discrete sensing portions, which may be referred to as the gauge length, is thus defined by the width of the interrogating pulse.

The backscatter signal received at the detector at any instant is therefore an interference signal resulting from the combination of the scattered light from all of the scatter sites within a section of fibre. The distribution of scattering sites within a given section of fibre is effectively random and thus the number of scattering sites and distribution of such sites within a section of fibre will vary along the length of the fibre. Thus the backscatter intensity received from different sections of fibre will vary in a random way. However, in the absence of any environmental changes affecting the fibre the distribution of scattering sites in a given section of fibre will remain the same and thus the backscatter intensity from a given section of fibre will be consistent for identical interrogating pulses. Any mechanical disturbances of the fibre, such as bending caused by an incident acoustic wave, will change the distribution of scattering sites and the effective refractive index of that part of the fibre and thus lead to a change in the resulting backscatter intensity.

Thus by repeatedly launching interrogating pulses into the fibre and looking at the backscatter intensity the same time after launch for each pulse, any changes in backscatter intensity from a given section of fibre can be determined and hence any acoustic disturbance of the fibre detected. Note as used herein the term "acoustic" is taken to mean any type of pressure wave of vibrational type stimulus and for the avoidance of doubt will include seismic stimuli. The term acoustic shall also be taken to cover ultrasonic and infrasonic stimuli. The pulse characteristics used for one interrogation, i.e. the intensity and frequency of the interrogating radiation for each pulse and the pulse duration and/or number of pulses in an interrogation, should be repeated for a subsequent interrogation in order for the intensity of the backscatter returns to be compared. Clearly if a greater amount of light is injected in one interrogation than the next, for instance by altering the overall pulse duration and/or intensity, this would be expected to result in a variation in backscatter intensity. Altering the number of pulses and/or pulse duration between interrogations would also result in the backscatter from different sections of the fibre being compared. In addition varying the frequency of the interrogating radiation could also lead to a variation in the degree of interference in the backscatter signal.

Therefore in order to ensure that any variation in backscatter intensity is due to disturbances acting on the fibre, rather than a variation in the properties of the interrogating radiation, the backscatter from interrogations having the same pulse characteristics should be compared.

The launch repetition rate, also referred to as the ping rate, is therefore set so that the time between interrogations is at least as long as the round trip time for light to reach the end of the fibre and return (or, for a very long fibre, a distance into the fibre from which no significant backscatter is expected). This ensures that any backscattered light received at the start of the fibre can be uniquely identified with a section of fibre and the backscatter signals from two interrogations do not interfere with each other at the detector. For a fibre which is 40km this would require enough time to allow a round trip in the fibre of 80km. If the refractive index of the fibre is n=1.5 say so that the speed of light in the fibre is roughly 2x10⁸ms^"1 then the time between interrogations should be at least 0.4ms and the ping rate should be less than 2.5kHz. Obviously higher ping rates could be used for shorter fibres. Clearly the ping rate determines the effective sample rate of the sensor and thus the ping rate should ideally be set high enough so that the Nyquist limit is above the maximum frequency of the acoustic signals of interest.

Note that as used in this specification the term interrogation shall be taken to mean an instance of launching interrogating radiation into the fibre and detecting the backscatter signal from the fibre. Typically an interrogation may comprise a single continuous pulse. However in some DAS sensors a single interrogation may comprises two or more distinct pulses that are relatively closely spaced. In such sensors the backscatter signals from the two pulses are intended to interfere at the detector to provide the measurement signal but again successive interrogations are arranged so as to not interfere.

It will be noted that wavelength division multiplexing techniques can be used with pulses of different wavelength being launched at staggered intervals such that multiple pulses may be propagating in the fibre at the same time. In effect a first series of interrogations at a first wavelength may be interleaved with a second series of interrogations at a second wavelength. However only the backscatter intensity from pulses of the same wavelength are compared to determine any change in intensity to detect any acoustic stimulus and only one interrogation of each wavelength is propagating within the sensing fibre at any time. The backscatter signal from a first interrogation does not interfere with the backscatter signal of a second interrogation at the frequencies of interest for the first and second interrogations (although clearly there may be some effects at other frequencies). This can lead to a first acoustic

measurement from the first series of interrogations at the ping rate and an interleaved second acoustic measurement from the second series of interrogations, also at the ping rate. In theory the two measurement signals may be combined to provide a single acoustic measurement signal with an update rate faster than the ping rate.

Such DAS sensors are very useful and have been used in a range of different applications. However one possible issue with such DAS sensors are that the variations in intensity for a given channel in response to a given input stimulus may be non-linear and also the gain of the channels, i.e. the amount intensity change for a given stimulus is variable.

As mentioned above the intensity of backscatter from a given section of fibre will depend on the number and distribution of scattering sites in that section. In a simple model the number of scattering sites can be thought to determine the amount of scattering that could occur and the distribution of such scattering sites determines the interference. An acoustic stimulus leading to a strain on the fibre may result in a change of optical path length within the relevant section of fibre (which could be a physical change in length and/or a change in the effective refractive index in part of the fibre). In this simple model this can be thought of as changing the separation of the scattering sites but without any significant effect on the number. The result is a change in interference characteristics. In effect the acoustic stimulus leading to optical path length changes in the relevant section of fibre can be seen as varying the bias point of a virtual interferometer defined by the various scattering sites within that section of fibre.

Figure 3 illustrates an idealised plot 301 of backscatter intensity against bias point for a given section of fibre. As the bias point is changed, i.e. an optical path length change is imparted to the relevant section of fibre, the backscatter intensity will go through a maximum value to a minimum value (where there is maximum destructive

interference).

In the absence of any external stimulus a given section of fibre can be thought of as having a steady state bias point, i.e. the operating point of a given section of fibre (and equivalently a given channel of the DAS sensor) will lie somewhere on plot 301. Figure 3 illustrates an operating point 302 for a given section of fibre.

Any acoustic stimulus causing a change in path length will thus result in a variation of the bias point about this operating point. A sinusoidal input stimulus resulting in a corresponding optical path length variation is illustrated as input 303. This will cause the intensity of backscatter from that section to vary as indicated thus resulting in the output 304. It will be clear however that different sections of fibre, i.e. different channels, will have different operating points (as well as different maximum possible output intensities). Each channel can thus be thought of as having a different operating curve, i.e. the response to any input stimulus about its current operating point. Figure 4 illustrates the effect of different operating points/operating curves for different channels. Figure 4 illustrates three different operating curves and the resultant response of each to the same input stimulus.

It can be seen that the output response for operating curves 1 and 3 (bottom and top respectively) are both distorted, in different ways, with respect to the input stimulus. Operating curve 2 exhibits a more linear response and thus the output is not so distorted. It can however be seen that in these curves (where the peak to peak variations of the various operating curves have been normalised) that operating curve 2 produces a greater magnitude variation in output intensity for the given input stimulus than operating curves 1 or 3.

It will therefore be clear that a variation in operating point can lead to a non-linear response between the input stimulus and detected intensity variation and that further the gain of a channel depends on its current operating point. It will also be seen that for this operating point the intensity variation exhibited by operating curve 2 has intensity increasing at times when the intensity exhibited by operating curves 1 and 3 is decreasing and vice versa. For the operating point operating curve 2 can therefore be said to lead to a different polarity of response to those of operating curves 1 and 3.

These limitations with amplitude based DAS sensor mean that coherent processing techniques cannot be applied to the data and, whilst incoherent methods are suitable for a wide range of applications, there are some applications where such sensors would not be able to provide accurate enough data.

It should be noted that the explanation given above is a relatively simplistic explanation of the various interactions in a sensing fibre to highlight several factors that provide the variation in operating point the various channels of an intensity-based Rayleigh DAS sensor. In reality there may well be other influences and this explanation is not intended to be limiting. In embodiments of the present invention the operating point of the sensing portions of fibre, i.e. the sensor channels, in a Rayleigh scattering based fibre optic distributed sensor are deliberately modulated. By deliberately modulating the operating point of a sensor channel it is possible to determine information about the current steady state operating point of that channel. This can be useful in helping address some of the limitations of amplitude based DAS sensors described above. Preferably the operating point is modulated in a pre-determined manner. By deliberately modulating the operating point in a known manner any other changes in operating point arising from environmental effects can be detected. In particular low frequency effects, for instance length and/or refractive index variations due to temperature changes or low frequency external strains, can be detected. This allows amplitude based Rayleigh scattering techniques to be used to provide detection of changes of temperature. As will be described in more detail below these temperature change effects can be detected as they occur and thus relatively rapid changes in temperature can be detected - more quickly than using conventional distributed temperature sensing (DTS) techniques.

Figure 5 shows one embodiment of the present invention. Figure 5 shows an interrogator unit 100, which may be the same as the described above in relation to Figure 1 , which outputs an interrogating pulse for sensing fibre 101 as described previously. However in this embodiment, between the interrogator unit 100 and sensing fibre 101 is a frequency modulator 501 which may for instance be a suitable acousto-optic modulator (AOM). The frequency modulator acts to vary the frequency of the interrogating pulses input to sensing fibre 101 between interrogations so as to vary the operating point of the channels of the sensor for different interrogating pulses.

Thus the interrogator unit 100 may output pulses having a constant base frequency, f_B. The frequency modulator 501 applies a time varying frequency modulation to the pulses so that the output pulse which is transmitted into the sensing fibre 101 has a frequency f_B ± Δί where Δί varies between at least some interrogating pulses, i.e. pulses of different interrogations. In one embodiment, which is particularly suitable for continual monitoring of low frequency strains or temperature induced path length changes, a continual frequency variation is applied, i.e. the frequency is varied between each successive pulse.

The phase difference of light reflected from different scattering points is the product of the optical path length between them and the frequency of the interrogating pulse so modulation of the frequency has the same effect as modulating the optical path difference. Therefore changing the frequency of the interrogating radiation between each interrogation will have the effect of changing the operating point of the various channels of the sensor between each interrogation. Changing the frequency of the interrogating radiation has a similar effect to applying a path length variation.

The detected intensity from any given channel will thus have a response that (between interrogations) exhibits a variation due to the applied frequency modulation. Referring back to Figure 4 in this instance the input could be seen as the applied frequency modulation of the interrogating pulses. The resultant output will therefore be a corresponding intensity variation in the detected backscatter signal which varies at the same rate as the frequency modulation is applied.

The extent of the frequency modulation excursion and the cycle rate of the frequency modulation may be chosen for a particular application. For detecting temperature variations which occur over the order of tens of seconds a frequency modulation of the order of a few hundred Hertz may be sufficient. The extent of the frequency modulation applied should be sufficient to result in a detectable change in operating point of the channels but is generally chosen so as to not produce too large a shift in operating point. In some embodiments, especially for continual monitoring of low frequency strains or temperature variations, the extent of the frequency modulation may be chosen to be such that the operating curve can be approximated by a linear fit over the region covered by the modulation. Thus in some embodiments the extent of the frequency modulation applied may chosen so as to produce a phase modulation of no greater than say ττ/4 radians, or ττ/8 radians or even 1/8 radians. A frequency modulation of Δί over a double pass through a fibre length of r metres would result in a phase modulation Φ = Δτ.2Γ.η.2ττ/α If r is 10m say then allowing a phase modulation of say 1/8 would allow a frequency modulation of ±200kHz.

For example consider an interrogator unit arranged to launch pulses at the base frequency, f_B, with a ping rate of 2kHz say. The frequency modulator 501 may be arranged to vary the frequency between say ±200kHz with a cycle rate, f_c, of say 400Hz. This means that over the course of five interrogating pulses the frequency of each interrogating pulse will vary somewhere between f_B + 200kHz and f_B - 200kHz. Figure 6 illustrates how the frequency modulation applied by the frequency modulator 501 could vary with time and indicates the times at which interrogating pulses could be generated and thus the resulting frequency applied. In some applications however it may be preferred to use a lower cycle rate of the frequency modulation as, in some instances, it may be easier to track. The period of the frequency modulation cycle may be chosen according to the application, although the period of the frequency modulation cycle should be much shorter than the period of the low frequency signal of interest. For example a cycle rate, f_c, of the order of 20Hz may be appropriate for some applications.

In other embodiments the extent of the frequency modulation may be greater to allow more of the operating curve to be mapped. This could be of use for example in interpreting measurement signals in response to a high strain stimulus, e.g. allowing unwrapping of the measuring signal.

The modulator(s) used to generate the frequency modulation should be appropriate for the desired frequency modulation. Acousto-optic modulators typically have a working frequency modulation range over which the output is linear and beyond the working frequency range the amplitude of light may start to decrease. Thus the type of frequency modulator used may partly limit the extent of frequency modulation applied. The frequency modulator may apply a single frequency modulation to the entire interrogating pulse. Thus the frequency between successive pulses varies in a step- change manner. If the ping rate is an integer multiple of the frequency modulator cycle rate, say n times the cycle rate, then there will be a repeating cycle of n pulses of different frequency. Thus the frequency modulator may be arranged to cycle through the n various frequency modulations within one cycle. For a step-wise change in frequency between each pulse using a ping-rate which is an integer multiple of the frequency modulator cycle rate limits the number of different frequencies that need to be generated. However it would be entirely possible to have a non-integer ratio between the ping-rate and frequency modulator cycle rate. In other embodiments the frequency modulator may be driven so that the frequency variation applied varies continuously. Clearly the modulation will only be applied when a pulse is passing through the frequency modulator. Driving a frequency modulator in a continuous manner may, in some applications, be easier to implement than a stepwise change in frequency. This does mean however that the frequency modulation applied may vary throughout the pulse duration. The pulse duration is typically quite short however, for example the pulse duration in some applications may be of the order of 100ns or so. With a frequency modulation cycle rate of 400Hz or lower, and a frequency excursion of ±200kHz the variation in the frequency modulation applied over the duration of the pulse is very low.

The amplitude of this variation in detected intensity will, in the same way as described previously depend on the operating point of the relevant channel at the base frequency f_B. In the absence of any strain or temperature induced changes on the optical fibre the amplitude of this 400Hz signal in the detected intensity will thus be constant. In other words the detected intensity from each channel will exhibit a response at 400Hz and, in the absence of any environmental changes affecting the fibre, the amplitude of the variation will be constant.

However if there is a temperature or low frequency strain change affecting the fibre the operating point of the affected channels will also move because of these changes. The amplitude of the 400Hz signal will also thus change. Therefore, by monitoring the envelope or amplitude of the 400Hz signal in the detected backscatter from a given sensing channel, any variations in operating point due to environmental changes can be detected. Figure 7 illustrates this principle and shows how the amplitude of a 400Hz signal in the detected intensity for a given channel could vary over time. The detected intensity for a given channel would be filtered to identify a signal at the cycle frequency of the frequency modulator, e.g. corresponding to the 400Hz modulation in the example discussed above. The amplitude or envelope of this 400Hz measurement signal would then be tracked over time, for example over a few minutes. In this example the detected amplitude is relatively constant for a first period, up to ti , indicating that the steady-state operating point of the channel is relatively constant and thus the relevant section of fibre is at a constant temperature. At ti however the amplitude of the 400Hz signal starts to change, indicating a temperature induced change in the steady state operating point of the channel. The amplitude progressively drops to near zero at t₂ indicating that the operating point is crossing a maximum or a minimum in the operating curve. The amplitude then continues to progress to a maximum at time t₃, which indicates that the operating point is near the maximum gradient part of the operating curve (e.g. operating curve 2 as illustrated in Figure 4 would apply in steady state). It can be seen that monitoring the amplitude of the 400Hz signal can be used to track any variation in the operating point. Further the rate of change of amplitude can be used as an indication of the rate of temperature change. If the temperature variation were to continue the amplitude plot illustrated in Figure 7 would reach another minimum as the next maximum or minimum in the operating curve is crossed. The rate of crossing maxima/minima in the operating curve provides a good indication of the rate of temperature change.

Operating point changes can additionally or alternatively be detected by monitoring a change of polarity of the 400Hz signal in a given channel's output compared to the frequency modulation. Thus if the output intensity for a channel increases and decreases respectively when the frequency of the interrogating pulse increases and decreases respectively the channel could be said to have the same polarity as the applied frequency modulation, or alternatively be in phase with the applied modulation. If however the reverse is true the output channel may be said to be of the opposite polarity to, or in anti-phase with, the applied modulation (although it will be appreciated that the designations of what output response is of the same polarity as, or in phase with, the applied modulation is arbitrary). Whether a given channel is in-phase or in anti-phase will depend on which side of a maximum in the intensity/bias curve the operating point is located on, or put another way the local gradient of the operating curve at the steady-state operating point. If the operating point changes so that the operating point crosses a maximum or a minimum the output response will relatively abruptly change polarity , i.e. change from in-phase to anti-phase or vice versa. Thus detecting a change in phase or polarity of the output variation compared to the frequency modulation applied may be used to detect a shift of operating point of a channel due to temperature variation or a low frequency strain. It will of course be appreciated that if the operating point in the absence of any frequency modulation is near a maximum or minimum in the operating curve the applied frequency modulation may cause the operating point to move across the maximum or minimum at the 400Hz cycle rate. In such conditions the exact phase or polarity of the output may be difficult to determine. However over a time scale of several minutes a progression from one polarity to the opposite polarity, i.e. a relatively sudden change in phase will be apparent and can be used as an indication of a temperature induced change in operating point of the relevant channel.

Such a distributed acoustic sensor may therefore be used to provide monitoring for temperature changes. Monitoring for temperature changes may be useful in lots of different applications. One particular application for instance may be for leak detection in pipelines, especially oil or gas pipelines. Detecting a leak in an oil or gas pipeline is important for both safety and environmental concerns as well as avoiding loss of product. Gas is typically highly pressurised in pipeline and thus if gas escapes via a leak rapid expansion with a consequent cooling effect is experienced. Oil is typically transported at an elevated temperature to improve flow and escape of oil from a pipeline via a leak may cause heating of the local environment. In both cases therefore a relatively sudden onset of a leak may lead to a relatively quick temperature change in the local environment, i.e. of the order of a few minutes. It has been proposed to detect such leaks using conventional DTS sensors based on Brillouin or Raman scattering. As mentioned however such sensors require a relatively long time average to provide accuracy and such sensors may not be suited to rapidly identify small temperature changes. The methods of the present invention provide a gain in signal- to-noise ratio for the thermal noise signal of interest, which is a particular advantage. This allows the methods to usefully be employed even where very little light is being returned from the sensing fibre, for instance for sensing channels at the end of a long fibre and/or where there are various losses between the sensing channel and the detector for instance due to fibre attenuation, connectors, splices or the like. In such low light situations it may not be possible to determine any low frequency temperature effects in the absence of the applied frequency modulation.

In some embodiments the temperature sensing could be a secondary detection effect. For instance the optical fibre could be at least partly deployed within or coupled to a material that exhibits temperature changes in response to certain stimuli or in the presence of certain analytes. For instance sections of the fibre could be coated with a material that exhibits an exothermic or endothermic reaction in the presence of an analyte. Any significant temperature changes in the coated sections could indicate the presence of the analyte. Sections of the fibre could also be uncoated to provide a control indication of environmental temperature changes.

In some embodiments the sensor may be operated to detect acoustic stimuli acting on the sensing fibre in addition to providing sensing for temperature or low frequency strain variations.

Referring to the example discussed above where the sensor varies the frequency between each successive pulse it will be clear that the frequency modulation will result in inherent intensity variations between successive pulses. As mentioned previously for performing DAS it may be preferred to compare the intensity backscattered from interrogating pulses of the same frequency to avoid any artefacts introduced by the frequency modulation.

If the frequency modulator 501 applies a repeating series of frequency modulations to interrogating pulses and the ping rate is an integer multiple of the cycle rate, f_c, of the modulator 501 , then a pulse with a given frequency will repeat once each cycle. It may therefore be possible to compare the response to pulses of the same frequency at a pulse rate equal to that of the frequency modulation. In other words if the ping rate is n times the cycle frequency of the frequency modulator, where n is an integer, then every nth pulse will be a given frequency Thus the intensity response for a channel for every nth pulse could be compared in the same manner as for a conventional DAS intensity based sensor as described above to effectively provide a DAS sensor operating at a ping rate equal to the cycle rate of the frequency modulator. Thus in the example discussed above with a ping rate of 2kHz and a frequency modulation cycle rate of 400Hz, every fifth interrogating pulse will have the same frequency. Thus the response from every fifth pulse could be analysed as discussed above to detect any acoustic stimuli acting on the fibre with a frequency below 200Hz. In this arrangement it would be possible to analyse multiple series of fifth pulses, i.e. the detected response to pulses 1 , 6, 11 etc all at frequency ^ could be compared as one sensor series with the response to pulses 2, 7, 12 etc all at frequency f₂ being compared as separate sensor series and so on.

It will be noted that depending on the frequency modulation applied and the ratio of the ping rate to the frequency modulation cycle, the same frequency could be repeated within a frequency modulation. For example if the ping rate was an even integer multiple of a sinusoidal type frequency cycle modulation and the first pulse in each frequency cycle was synchronised to no applied frequency modulation then the pulse half way through the cycle would also be at the base frequency. These pulses could be used for DAS at an updated rate of twice the frequency cycle modulation rate. In some applications it may be possible to have a cycle of just four pulses, the first and third pulses of each cycle being at a base frequency and the second and fourth pulses having equal and opposite frequency modulations. This would provide a series of pulses at the base frequency at half the ping rate that can be used for DAS as well as the complete series of pulses exhibiting the frequency modulation cycle that could be used to monitor low frequency temperature/strain effects. The repeating series of the second pulses in each cycle could also provide another series at a frequency f_B+Af say at a rate equal to a quarter of the ping-rate and likewise the repeating series of the fourth pulses in each cycle would provide a series at a frequency f_B-Af say.

Alternatively a repeating cycle of four pulses could comprise two pulses of frequency f_B+Af followed by two pulses of frequency f_B-Af. This corresponds to a frequency modulation of ±V2.Af at a cycle rate of a quarter of the ping rate.

If a larger frequency excursion was chosen so that a full cycle of frequency modulation would lead to a variation in operating points over a full cycle of the operating curve, then the slope of the operating curve would be significantly different for each frequency. In such a case at least one frequency should (for a given channel) lead to an operating point relatively near the maximum slope of the operating curve (especially if the value of n is relatively high) and so would produce the largest possible output signal for that channel for a given disturbance to the fibre. If the sensor series outputs from all frequencies were initially analysed then the one with the largest output signal could be selected for further processing. In other words the sensor series derived from the pulses at frequency ^ could be compared with the sensor series derived from the pulses at frequency f₂ and so on to determine which sensor series was exhibiting the greatest variation in intensity, i.e. the greatest gain, for a given channel. The relevant frequency series could then be used as the acoustic measurement signal for that channel - possibly with a periodic recalibration.

Such a large frequency excursion would lead to a relatively distorted signal at the cycle rate of the frequency modulation, which may reduce the usefulness for continual monitoring of low frequency temperature and strain effects. However such a large frequency excursion does ensure that for each channel of an acoustic sensor there is a series of interrogating pulses that should lead to a near maximum gain - albeit at an update rate equal to the frequency cycle rate.

For disturbances that are continuous or repetitive the same approach could be used but instead of cycling through the frequencies on a pulse by pulse basis, one frequency could be repeated for an fixed length of time, say 1 second, before doing the same with the next frequency in the cycle. This would mean data would be available at the full ping rate for each frequency. Additionally or alternatively interrogating pulses with a frequency modulation may be interleaved with one or more pulses at a defined constant frequency. Thus for example every second pulse may have a defined constant frequency and thus can be used for DAS at one half of the ping rate whereas the other pulses exhibit a frequency variation and are used for monitoring any variations in operating point as described above. In effect a series of pulses at a constant frequency and a first ping rate may be interleaved with a series of pulses of varying frequency at the same ping rate. Figure 8 illustrates this type of operation. Figure 8 illustrates that a series of pulses of a fixed frequency f_B may be interleaved with a series of pulses having a repeating frequency sequence f₂, f3, , - Preferably the frequency sequence ^ - f₅ modulates the frequency about f_B as described above. If the overall ping rate was 2.5kHz this would lead to a ping rate of the pulses f_B for DAS of 1.25kHz and a frequency modulation cycle rate of 250Hz. This approach would be useful if the frequency modulation rate was desired to be within the acoustic band of interest, as the measurements at f_B could be used to get the acoustic information without being corrupted by the frequency modulation signal which itself could be measured using frequencies ^ - f₅ provided its amplitude was significantly larger than the acoustic signal at that frequency. Additionally or alternatively wavelength division multiplexing techniques could be applied so that a first series of pulses at a first wavelength is transmitted with each pulse having the same frequency for distributed acoustic sensing. A second series of pulses, at a second wavelength, could be transmitted with a frequency variation between the pulses as discussed above to provide independent monitoring for temperature induced changes.

In some embodiments however the intensity variation detected in response to the frequency modulation around a base frequency, f_B, may be used to infer information about the steady state operating point when interrogating with pulse of frequency f_B. In other words in addition to, or instead of, detecting any variations in operating point arising from temperature variations and the like, a frequency modulation may be applied to at least some interrogating pulses so as to determine the current operating point of the sensor channel when no frequency modulation is applied. This information about the operating point may then be used in processing the data from the channel(s) acquired with no frequency modulation applied.

For instance, the polarity or phase of the channel could be determined, i.e. the sign of the local gradient of the operating curve about the steady-state operating point. This can be determined by looking for whether the variation in detected intensity increases or decreases with an increase in frequency. Alternatively the response from various channels could simply be compared to one another to determine whether any two channels are in phase or in anti-phase. For any channels which are in anti-phase it will therefore be appreciated that if the same acoustic stimulus is received at both channels the output intensity variation resulting from that stimulus would also be in anti-phase. For a group of channels it would therefore be possible to determine, from looking at the response to the applied frequency modulation, a first set of channels in phase with one another and a second set of channels also in phase with one another but in anti-phase with the first set. The actual detected intensity signals from the second set of channels may therefore be inverted so as to bring the second set of channels into phase with the first set. This may then allow array processing techniques to be applied to the returns from all channels.

Note there may of course also be a third set of channels where the phase is difficult to determine. Typically however this will occur where the operating point is near a maximum or a minimum. At a minimum in the operating curve the backscatter signal may be nearly completely faded anyway and thus it may be no detectable signal is present anyway. Even if a detectable signal is present at a minimum in the operating curve the gain of the relevant channel will be low. At a maximum in the operating curve there will be a backscatter intensity but the gain of the channel will be low and thus the signal to noise ratio for that channel may be low anyway.

The method may also be used to detect which channels are at or near a maximum of minimum in their operating curve. As mentioned these channels may exhibit a low SNR. Knowing that a particular channel exhibits a low SNR may be of use in processing looking for a similar signal affecting several channels. Channels with low expected SNR or a low gain may be omitted from processing or flagged with a high probability of error in any kind of confidence based processing. In some embodiments the response to the applied frequency modulation may be used to determine an estimate of gain of a channel or track a change in gain over time. In some embodiments a gain correction factor could then be applied to the measurements detected from one or more channels. For instance the amplitude of the intensity variation in response to the frequency modulation could be determined and monitored over time. If the amplitude variation increases or decreases as a result of wander of the steady-state operating point (for instance due to temperature changes) this will indicate that the gain of the relevant channel is changing. A gain correction factor could be applied to compensate for any such gain wander. Additionally or alternatively the intensity variation from various channels to the applied frequency modulation could be detected and used to derive a normalisation factor across different channels.

In some embodiments non-linearity in the output intensity variation compared with the applied frequency modulation may be determined and/or corrected for. As described above the shape of the operating curve for a given channel will determine the output intensity variation in response to the applied frequency modulation. The shape of the applied frequency modulation is known which allows the detected frequency variation to be used to determine information about the shape of the operating curve for that channel and/or correct for any non-linearity. In effect the detected response may be compared to an ideal linear response to identify the extent of the non-linearity. In this instance it may be beneficial to use a frequency excursion for the frequency modulation which is sufficiently large to determine the shape of the operating curve. Information regarding the shape of the operating curve may then be applied to any acoustic signals detected from a given channel. A variety of non-linear distortion signal recovery techniques could be applied. For example a non-linear scaling factor could be applied based on the detected intensity variation. Adjusting a detected measurement signal to account for the shape of the local operating curve represents another aspect of the present invention. All of these techniques would improve the consistency and reliability of the

measurement signals detected for DAS sensing and may be used to allow coherent processing/array based processing techniques to be applied. The use of a frequency modulation of at least some interrogating pulses may therefore be of use for improving the operation of a DAS sensor, even if temperature change or low frequency strain sensing is not applied.

In such embodiments it will therefore clearly be necessary to transmit a regular series of pulses at a repeated frequency ^ that can be used for DAS whilst also transmitting some pulses which are frequency modulated with respect to ^ to allow the operating point characteristics of the channels to be determined.

One of the main causes of wander of operating point may be temperature related changes and thus monitoring for temperature related changes as described above, with pulses of varying frequency being interleaved with pulses of a fixed frequency, may inherently provide the information suitable for polarity and/or gain correction. However in some instances it may be acceptable determine the operating point characteristics only periodically. Thus on initialisation of the DAS sensor a series of interrogating pulses having a frequency modulation between pulses could be transmitted as part of an initialisation and calibration step. Once the polarity of each channel was established and/or a suitable gain factor determined the interrogator unit could start transmitting pulses of a constant frequency f_B for DAS. Periodically however a frequency modulation could be applied to at least some of the interrogating pulses. For simply determining a polarity and/or gain factor it may not be necessary to apply a constant frequency modulation cycle. It may be sufficient to simply transmit a few pulses of increasing (and/or decreasing frequency) in a frequency ramp and detect whether the channels are in-phase or anti-phase and/or the maximum amount of intensity variation. Transmitting a frequency modulation as a pilot tone, i.e. such as a constant 400Hz variation is useful for continual monitoring for temperature changes and the like as the component in the detected intensity is narrow band with a frequency much higher than the temperature signal being detected but such a constant pilot tone may not be necessary for occasional detection of operating point characteristics.

The discussion above has generally focussed on a regularly varying sinusoidal type frequency modulation being applied. However other forms of frequency modulation may be applied as required. For example there could be a repeating frequency ramp, e.g. the frequency modulator could be driven by a sawtooth type waveform. There may in some application be advantages in applying a random or pseudo-random frequency modulation. Some compressive sensing techniques make take advantage of random sampling. Thus a step-wise type frequency modulator may be driven to apply a random frequency modulation, possible from a selected set of possible frequencies, to interrogating pulses. The series of randomly frequency modulated pulses could be interleaved with interrogating pulses of a known constant frequency to allow for DAS sensing.

Referring back to figure 5 the frequency modulator 501 could readily be arranged to apply a frequency modulation to some pulses and not to others and/or apply any pattern of frequency modulation required in response to a suitable control signal.

Note in Figure 5 the frequency modulator is illustrated as being external to the interrogator unit. This is one possible implementation and embodiments of the present invention may be implemented by retrofitting to existing DAS interrogator units.

However in other embodiments the frequency modulator 501 may form part of the interrogator unit and may be located in the transmit path only. In some instances frequency modulator 501 could be provided instead of modulator 103 illustrates in figure 1 - or modulator 103 could be configured to act as the frequency modulator and pulse generator. In some applications frequency modulation may be applied by modulating the operating conditions of the laser 102.

The discussion above has assumed that the laser 102 (and modulator 103 if present) produce a stable frequency output, for instance that a frequency locked laser is used. The methods described above can also be used with non-frequency locked lasers, where laser noise, i.e. frequency drift, could potentially be a source of low frequency noise, to identify and compensate for any such laser noise.

If the laser 102 and/or modulator 103 did exhibit some frequency drift over time this would result in an unknown modulation of the base frequency f_B. In a conventional DAS sensor any such variation in frequency would thus alter the response of the various channels as discussed above. As the frequency drift of the laser occurs over a relatively long timescales this results in a low frequency noise signal (of the order of a few tenths of Hz or lower). For acoustic monitoring the noise signal itself may not be too much of a problem but it is likely to be so when measuring low frequency effects such as temperature changes. It may therefore be desired to detect and possibly compensate for the effects of such laser noise.

By applying a deliberate known frequency modulation to the interrogating pulses in the manner described above, with a cycle frequency much higher than that of the laser noise, then any laser noise will manifest as a detectable amplitude modulation of the signal at this cycle frequency.

Figure 9a shows some test data with a frequency modulation applied to interrogating pulses of a DAS sensor without a frequency locked laser. A DAS interrogator unit was connected to an 80MHz AOM to which a frequency modulated signal was applied. The frequency modulation had a maximum excursion of ±200kHz at a modulation frequency of 400Hz. The output of the AOM was connected to a length of optical fibre. The results from various sensing channels were recorded and analysed. The measurement signal at 400Hz, i.e. the same frequency as the frequency modulation applied, was analysed and the amplitude of 400Hz signal over time was determined. Figure 9a shows the recorded amplitude of the 400Hz signal from three different sensing channels over a period of about 5 minutes. It can be seen that amplitude signals vary in amplitude and do so on a time scale of about 25 seconds. The variation in amplitude from a maximum to a minimum can also clearly be seen.

The phase relationship of the measurement signals from various channels was also recorded over time. As expected the various channels normally differ in phase by an integer number of π radians - indicating that they are either in phase or anti-phase with each other. At times when the amplitude of the 400Hz signal is low however the phase is poorly defined.

These result shows that the low frequency drift caused by drift of the laser frequency can be identified using the deliberately introduce pilot tone.

The laser noise affects all channels but the phase and amplitude of it will vary according to the location of the point on the operating curve. As a channel drifts to a different point on the operating curves the amplitude and phase of the laser noise signal will vary in the same way as the signal induced by the frequency modulation. The detected response of the various channels to the frequency modulation can then be used to correct for the laser noise. In essence the data could be divided into time bins shorter than the typical period of a fading cycle. For each time step channels exhibiting the largest amplitudes at the applied modulation cycle frequency could be selected (to avoid using channels which are faded at that time bin). The channels selected could be limited to those in a region near to the start of the fibre, e.g. a relatively long length, e.g. 1 km, near the start of the fibre. As the laser noise will be the same throughout the fibre it is not necessary to process every channel - although a suitable number are chosen to provide a good measure of the laser noise. Typically channels near the start of the fibre are used where optical losses will be relatively low. The low frequency signal from each channel could then be scaled according to the amplitude and phase of the applied frequency modulation. The mean of all the scaled channels could then be taken to provide a template for the low frequency signal. The low frequency noise for each channel could then be obtained from the template by scaling it according to the amplitude and phase of the applied modulation on that channel at the particular time.

This technique allows for the low frequency noise affecting all of the fibre

simultaneously to be identified and compensated for including those from further down the fibre that were not used to calculate the template.

It should be noted that low frequency effects such as a temperature changes can be discriminated from laser noise by looking at whether all channels are affected in substantially the same way. Temperature effects, especially from leaks in a pipeline setting, are likely to be relatively local and affect only a few contiguous sensing portions. Laser noise however would affect all channels simultaneously so by forming the template from channels over a large length of fibre avoids it being corrupted by events such as a localised disturbance.

The discussion above has focussed on applying a deliberate operating point modulation by applying a frequency modulation to the interrogating pulses. However it would be possible to achieve a similar affect by applying a predetermined strain modulation to the optical fibre. For example if a strain were applied to part of the sensing fibre at a known frequency, e.g. 400Hz, the operating point of the sensing channel would vary as described above.

For a fibre which is embedded in a medium, such as buried in the ground, such a signal could be applied by transmitting an acoustic stimulus at a known frequency into the medium. The stimulus would be transmitted to several channels of the sensing fibre and would be detectable as a modulation in intensity at the known frequency.

Typically only a fraction of the length of the sensing fibre can be stimulated by a given acoustic source and thus this technique may only be applicable to parts of the fibre at a time (or require multiple sources). Also unknown attenuation and propagation delays may limit the amount of information that could be obtained for gain correction say. Nevertheless it would be possible to detect an operating point variation by detecting a variation in amplitude from any given channel. Modulating the frequency of the interrogating radiation does however allow the operating point of all channels of the sensing fibre to be monitored simultaneously and the applied modulation is reliably known.

Another method of coherently stimulating some or all of a fibre would be to include in the cable a structure that could be electrically stimulated to induce an acoustic signal in the fibre. For example some piezoelectric material could be placed between two conductors and when an alternating voltage was applied across these conductors the piezoelectric material would vibrate causing a signal that could be picked up by the fibre. Yet another method would involve passing an alternating current down a conductor in the cable. Interaction of the current with the earth's magnetic field would cause the whole cable to vibrate slightly, generating an output signal at the frequency of the alternating current. It should be noted that the discussion above has focussed on transmitting a single interrogating pulse per launch (i.e. per ping). It would be possible to transmit other pulse configurations but in a DAS sensor which detects acoustic signals by looking at intensity variations a single pulse is generally preferred as it maximises the amount of light injected into the fibre, and hence sensitivity (as the amount of backscatter depends on the amount of light injected) for a given spatial resolution.

It will be noted that DAS sensor that determine a change in phase in a measurement signal are also known. One known type of phase based DAS sensor transmits a pair of pulses, of different frequency to one another, separated by a gap between the pulses. Light scattered from both pulses interfere to create a measurement signal at a carrier frequency defined by the frequency difference between the pulses. Any path length changes between the two pulses will lead to detectable change in phase in the carrier signal. Such phase based DAS sensors have the advantage of providing a linear response to an incident stimulus and thus many of the problems described herein may not be applicable to such sensors. However the gauge length of a pair of interrogating pulses is defined by the distance between the pulses and for optimum signals this distance should be several times the pulse width. To provide an acceptable spatial resolution the duration of each pulse is typically low and since the amount of backscattered light is proportional to the pulse length the sensitivity of such a DAS sensor will generally be less than that of a single pulse intensity based sensor of similar spatial resolution. Thus there are a number of applications where intensity based DAS sensors may be better suited. Embodiments of the invention may be arranged as part of an interrogator unit for a distributed fibre optic sensor or as an add-on or retrofit to such an interrogator unit. Embodiments of the invention may be implemented as temperature change sensors or distributed acoustic sensor a sensor with combined DAS/temperature variation capability. The methods of data analysis may be applied to data acquired from a suitable sensor to which a suitable modulation was applied. The method may be implemented by software.

The invention has been described with respect to various embodiments. Unless expressly stated otherwise the various features described may be combined together and features from one embodiment may be employed in other embodiments. It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended claims. The word "comprising" does not exclude the presence of elements or steps other than those listed in a claim, "a" or "an" does not exclude a plurality, and a single feature or other unit may fulfil the functions of several units recited in the claims. Any reference numerals or labels in the claims shall not be construed so as to limit their scope.

Claims

1. A method of distributed fibre optic sensing comprising:

performing a series of interrogations of an optical fibre, each interrogation

comprising:

launching at least one interrogating pulse of optical radiation into the optical fibre; and

for at least one sensing portion of said optical fibre detecting optical

radiation which is Rayleigh backscattered from within the sensing portion of said sensing fibre;

wherein the launch rate of said interrogations is such that optical radiation backscattered from one interrogation does not substantially interfere with optical radiation backscattered from another interrogation; and wherein at least some of the interrogations have a predefined variation in frequency of said at least one interrogating pulse from one another; the method comprising determining any variation in intensity of said

2. A method as claimed in claim 1 wherein said predefined variation in frequency comprises a cycle of frequency modulation which repeats at a modulation cycle rate.

3. A method as claimed in claim 2 wherein determining any variation in intensity of said backscattered radiation resulting from said predefined variation in frequency of the interrogating pulses comprises identifying a first signal in the detected backscatter radiation exhibiting at a frequency equal to the modulation cycle rate.

4. A method as claimed in claim 3 comprising analysing the amplitude of said first signal.

5. A method as claimed in claim 3 or claim 4 comprising detecting any variation in the amplitude of said first signal over time.

6. A method as claimed in any of claims 3 to 5 comprising detecting when the amplitude of the first signal goes through at least one of a maximum and a minimum.

A method as claimed in any of claims 3 - 6 comprising determining said first signal for each of a plurality of sensing portions of fibre and comparing any variation in amplitude of the first signals from said plurality of sensing portions of fibre.

8. A method as claimed in any of claims 3 to 7 comprising identifying a change in temperature of a sensing portion of the optical fibre by detecting a change in amplitude of the first signal for that sensing portion.

9. A method as claimed in claim 8 comprising determining a rate of change of temperature by determining the rate of change of amplitude of the first signal.

10. A method as claimed in any of claims 2 to 9 wherein the launch rate of said interrogations is an integer multiple of said modulation cycle rate.

1 1. A method as claimed in any of claims 2 to 10 wherein said modulation cycle rate is between 10 Hz and a quarter of the launch rate of said interrogations.

12. A method as claimed in any of claims 2 - 1 1 wherein each successive

interrogation has a frequency modulation applied according to said cycle of frequency modulation.

13. A method as claimed in claim 12 wherein each cycle of frequency modulation comprises an interrogating pulse at a first frequency and the method comprises, for at least one sensing portion of said optical fibre, comparing the backscatter from said pulses at the first frequency in successive cycles to provide distributed acoustic sensing.

14. A method as claimed in claim 12 or claim 13 wherein each cycle of frequency modulation comprises a plurality of interrogating pulses of different frequencies which are repeated each cycle and the method comprises, for at least one sensing portion of said optical fibre, comparing the backscatter from said pulses at the same frequency in successive cycles to provide a number of distributed acoustic sensing measurements for the same sensing portion.

15. A method as claimed in any of claims 12 to 14 wherein the cycle of frequency modulation comprises a first set of interrogating pulses all at a first frequency followed by at least a second set of interrogating pulses all at a second, different frequency wherein the backscatter from interrogating pulses within at least one set are used for distributed acoustic sensing and backscatter from interrogating pulses from different sets are compared to determine the optimum set or sets to be used for subsequent analysis.

A method as claimed in any of claims 2 - 1 1 wherein said cycle of frequency modulation is applied to a first series of interrogations which is interleaved with a second series of interrogations, wherein the second series of interrogations are not frequency modulated with respect to one another.

17. A method as claimed in claim 16 comprising, for at least one sensing portion of said optical fibre, comparing the backscatter from the interrogating pulses of the second series of interrogations to provide distributed acoustic sensing.

18. A method as claimed in claim 16 or claim 17 wherein interrogations of said first series and said second series are alternately launched into said optical fibre.

A method as claimed in claim in any of claims 14 - 16 wherein cycle of frequency modulation applied to the first series of interrogations applies a frequency modulation about the frequency of the pulses of the second series of

interrogations.

A method as claimed in any preceding claim further comprising, for at least one sensing portion of said optical fibre, comparing the detected intensity from one or more interrogating pulses to provide a measurement signal indicative of any acoustic stimuli acting on that sensing portion.

21 A method as claimed in claim 20 comprising analysing the variation in intensity of said backscattered radiation resulting from said predefined variation in frequency of the interrogating pulses to determine at least characteristic of an operating point of the sensor portion.

22. A method as claimed in claim 21 wherein said characteristic of the operating

point comprises a polarity of the variation in intensity compared to at least one of the predefined variation in frequency or the intensity response of another sensing portion to said predefined variation in frequency.

23. A method as claimed in claim 22 comprising determining a polarity for each of a plurality of sensing portions and adjusting the measurement signal for that sensing portion based on said polarity.

24. A method as claimed in any of claims 21 to 23 wherein said characteristic of the operating of an operating point comprises a gain value indicating the amplitude of variation.

25. A method as claimed in claim 24 comprising monitoring said gain value for a given sensing portion over time to detect any variation.

26. A method as claimed in claim 24 or claim 25 comprising comparing said gain value for a plurality of sensing portions to detect any variation.

27. A method as claimed in claim 25 or claim 26 comprising applying a gain

correction to the measurement signals from one or more sensing portions to correct for said variations in gain.

28. A method as claimed in any of claims 24 to 27 when dependent on claim 23 in which the measurement signal is combined with the polarity and gain values for a plurality of sensing portions to estimate the signal generated on each portion by laser frequency noise.

29. A method of detecting temperature changes in an area of interest comprising: performing a series of interrogations of an optical fibre deployed, at least partly, in said area of interest, each interrogation comprising:

launching at least one interrogating pulse of optical radiation into the optical fibre ; and for at least one sensing portion of said optical fibre detecting optical radiation which is Rayleigh backscattered from within the sensing portion of said sensing fibre;

wherein the launch rate of said interrogations is such that optical radiation backscattered from one interrogation does not substantially interfere with optical radiation backscattered from another interrogation; and wherein at least some of the interrogations have a predefined variation in frequency of said at least one interrogating pulse from one another, the predefined variation in frequency comprising a cycle of frequency modulation which repeats at a modulation cycle rate;

the method comprising identifying a first signal in the detected backscatter

radiation exhibiting at a frequency equal to the modulation cycle rate; and detecting any variation in the amplitude of said first signal over time.

A method of detecting a leak in a pipeline comprising performing the method of claim 29 to detect any sudden changes in temperature, wherein said area of interest comprises the path of the pipeline.

A method of distributed fibre optic sensing comprising:

performing a series of interrogations of an optical fibre, each interrogation

comprising launching at least one interrogating pulse of optical radiation into an optical fibre and detecting optical radiation which is Rayleigh backscattered from within said fibre to provide a plurality sensing portions; applying a predetermined modulation at a modulation frequency to the operating point of at least one sensing portion so as to produce in a variation in backscatter intensity from said sensing portion; and

detecting any variation in said amplitude.

A distributed fibre optic sensor apparatus comprising:

an optical source configured to generate interrogating pulses to be launched, in use, to an optical fibre in a series of interrogations .wherein the launch rate of said interrogations is such that optical radiation backscattered from one interrogation does not substantially interfere with optical radiation backscattered from another interrogation; a frequency modulator configured to apply a predefined frequency modulation to said interrogating pulses so that at least some interrogations have different frequency interrogating pulses to one another; and

a detector for detecting radiation Rayleigh backscattered from within said optical fibre;

a processor configured to determine, for at least one sensing portion of the

optical fibre, any variation in intensity of said backscattered radiation resulting from said predefined frequency modulation of the interrogating pulses.