WO2005101058A1

WO2005101058A1 - Method for geothechnically characterising an underwater floor by means of a broad-band and multiple flatness acoustic wave

Info

Publication number: WO2005101058A1
Application number: PCT/FR2005/000587
Authority: WO
Inventors: Xavier Demoulin; Olivier Sarzeaud
Original assignee: Xavier Demoulin; Olivier Sarzeaud
Priority date: 2004-03-15
Filing date: 2005-03-11
Publication date: 2005-10-27
Also published as: FR2867568B1; FR2867568A1

Abstract

The invention relates to a method for geothechnically characterising an underwater floor (B) involving at least one phase, wherein an acoustic wave is emitted by a source (1) and detected by a receiver streamer (2) and an analysis phase, wherein signals produced by receivers (21-24) are used for determining the physical parameters of a strata forming the floor (B). The inventive method uses the angle and frequency-dependence of the acoustic wave reflection coefficient on the interface formed by each stratum.

Description

METHOD FOR THE GEOTECHNICAL CHARACTERIZATION OF A UNDERWATER BACKGROUND USING A MULTI-RASANCE BROADBAND ACOUSTIC WAVE.

The invention generally relates to sonic excitation prospecting techniques.

More specifically, the invention relates to a method for geotechnical characterization of an underwater bottom, such as an underwater bottom, covered with a sheet of water of determined total height, and comprising a plurality of strata forming between them separation interfaces, each presenting its own physical parameters and extending to various depths below a first layer which forms a separation interface with the water, this process comprising at least one investigation phase, one pre-treatment phase and an analysis phase, the investigation phase itself including an acoustic excitation operation, implemented by emitting at least towards the bottom, from an immersed acoustic source, an incident acoustic wave of known frequency signature, an acquisition operation implemented by producing, by means of a flute of at least four submerged receivers, respective measurement signals resulting from detection of respective acoustic waves reflected by the background, and an operation of reading geometric data, the flute being at least approximately aligned with the source and distant from the background, the receivers being spaced from each other, and the incident wave being emitted in each window of a succession of disjoint transmission time windows, and having a variation time frequency within each transmission window.

A process of this type is for example described in US Pat. No. 6,151,556.

Other methods are also known to those skilled in the art, and in particular those described in patent documents EP - 0 215 703 and EP - 0 553 053.

More generally, the known acoustic systems for the investigation of the funds are either imaging systems or seismic systems.

Conventional imaging systems use acoustic waves of high frequencies, typically greater than 30 kHz, emitted towards the bottom and making it possible to establish an image of the bottom surface independently of the subsoil present, based on the retro mechanism. -diffusion.

The usual seismic systems of the previously defined type use, as a transmitter, a boomer producing an acoustic signal at very low frequency - and therefore having the advantage of penetrating deep into the basement.

The signal from multiple reflections and collected on a flute of hydrophones and / or geophones towed by a ship in synchronism with the boomer is thus representative of the deep layers of the subsoil. On the other hand, the frequency characteristics of the signal transmitted at very low frequency are poorly controlled, so that these systems only provide qualitative indications of the deep strata of the subsoil, at depths typically greater than 10 meters, and with vertical resolutions typically greater than 1 meter.

Recent sediment sounders have the advantage of using controlled signals ("chirp" or parametric sources) which insonate the bottom vertically at high power. However, signal reception, provided by the same transducer, significantly limits performance for instrumental (electronic saturation) and physical reasons (mixture of backscattering and specular reflection phenomena, thickness / speed ambiguity in the layers).

In this context, the present invention aims to propose a process which meets the need, which is today not satisfied, - to allow the quantified determination of the geotechnical properties of the subsoil of an underwater bottom over the first meters, typically over the first five meters at least, and with a resolution at most equal to fifty centimeters.

To this end, the method of the invention, moreover in accordance with the generic definition given by the preamble above, is essentially characterized in that the investigation phase includes a displacement operation, concomitant with the operations of excitement and acquisition, and implementation in simultaneously moving the acoustic source and the flute in the water table, in that the pre-treatment phase is implemented by deducing, from the measurement signals, the signature of the incident acoustic wave, and geometrical data, reflection signals having reached different receivers following the same emission of incident wave, and by grouping, in the form of series of reset reflection signals, reflection signals having reached, at angles of different shades, different receivers coming from the same bottom area following different transmissions of the incident wave, and in that the analysis phase is implemented by deducing the physical parameters of each bottom layer , from the series of recalibrated reflection signals, by exploiting the angular and frequency dependence of the reflection coefficient of the incident wave on each interface, each reflective signal exion re-aligned being interpreted as a sum of delayed and attenuated arrivals of clean radii corresponding to the reflection of the incident wave by the different strata of the bottom area to which this re-aligned reflection signal corresponds.

In practice, each reflected wave reaching the center of the flute in a time window, reception is advantageously derived from the reflection of an incident wave having reached the bottom under a raking angle at most equal to 70 degrees and at least equal to 10 degrees, and preferably between 25 and 65 degrees.

Each receiver is for example separated from. the acoustic source by a distance between approximately 1 time and about 4 times the average height of water under the instruments, each receiver can also be spaced from a neighboring receiver by a distance of between about 0.375 times and about 1 times the average height depending on the number of these receivers.

According to another mode of evaluation of the distances, each receiver is typically separated from a neighboring receiver by a distance corresponding to a difference in angle of shaving at least equal to 2 degrees, and preferably at least equal to 5 degrees, of the reflection signals respectively received by this receiver and by the neighboring receiver. Preferably, the incident wave has, in each emission window, at least one relatively low frequency component, the frequency of which is at least equal to 100 Hz, this incident wave usefully also having, in each emission window, at least one relatively high frequency component, the frequency of which is at most equal to 8 kHz.

. In a possible embodiment of the invention, the incident wave can thus have, • in each transmission window, a continuously variable frequency between a relatively low frequency component and a relatively high frequency component.

'To obtain both sufficiently rich information and satisfactory separation of the reflected waves, the transmission time windows each have, for example, a duration at least equal to 0.1 seconds, and preferably at least equal to 0.5 seconds, and are repeated periodically with a period for example at least equal to 0.5 seconds and preferably at least equal to 1 second.

In • its preferred embodiment, the method of the invention comprises, upstream of the analysis phase, an operation for determining the immersion depth of the acoustic source and / or the immersion depth of a point on the flute and / or water level.

The analysis phase is for example implemented by iteratively traversing a loop including a modeling operation during which are calculated, from fictitious physical parameters virtually attributed to the first stratum at least of the 'plurality of strata, and for the various receivers, respective fictitious reflexed reflection signals derived from measurement signals virtually produced by these receivers in the presence of these _. fictitious parameters, a subtraction operation during which the observed differences are formed between the respective reset reflection signals derived from the measurement signals actually produced by these receivers and the corresponding fictitious reset reflection signals, and an optimization and feedback at the end of which the fictitious physical parameters are ^' revised according to the deviations noted.

The method of the invention thus allows a geotechnical characterization of each stratum from a plurality of strata of an underwater bottom by means of physical parameters chosen from the set of parameters which includes the thickness of this stratum, the density of this stratum, the speed of sound in this stratum, the attenuation of sound in this stratum, the roughness of this stratum, and a gradient speed of sound inside this layer.

In its preferred embodiment, the method of the invention makes it possible to identify the different strata of the sub-aquatic subsoil with vertical resolutions of the order of ten centimeters.

Other characteristics and advantages of the invention will emerge clearly from the description given below, for information and in no way limitative, with reference to the appended drawings, in which:

- Figure 1 is a schematic perspective view of equipment implementing the method of one invention;

- Figure 2 is a schematic diagram showing in approximately chronological order the main phases and operations of the method of the invention;

FIG. 3, made up of 'diagrams 3A to 3C, intuitively and partially illustrates the RECAL step of resetting and grouping of reflection signals implemented during the course of the process of the invention, this step leading in particular to group reflection signals from the same bottom area;

- Figure 4, composed of diagrams 4A and 4B, completes the illustration of the RECAL step; - Figure 5, composed of diagrams 5A and 5B, partially illustrates the multiple acoustic propagation mechanisms in water, and their use in the invention;

- Figure 6 partially illustrates the multiple acoustic propagation mechanisms in the underwater bottom;

FIGS. 7A to 7J are diagrams representing, as a function of time t expressed in number of samples, the relative amplitude A of measurement signals respectively produced by 10 successive receivers increasingly distant from the acoustic source and separated from each other by a large distance, and highlighting the influence of the angles with which the waves are reflected at the bottom; and

FIGS. 8A to 8J are diagrams representing, as a function of time t expressed in number of samples, the relative amplitude A- of measurement signals respectively produced by 5 successive receivers increasingly distant from the acoustic source and for a modulation of low frequencies on the left and for a modulation of high frequencies on the right, highlighting the influence of the frequencies used. RATINGS

The reader will find below some of the symbols used to identify the different '' entities involved in the implementation of the invention: "i" is a counter index used to number each hydrophone, thus distinguishing it from other hydrophones;

"j" is a counter index used to number each stratum of the underwater bottom, thus distinguishing it from the other strata of the bottom;

"J" is an integer denoting the total number of observable strata;

"k" is a counter index used to number each eigen ray of an acoustic wave by distinguishing it from the other eigen rays;

"1" is a counter index used to number each emission of an incident acoustic wave (also called "firing") thus distinguishing it from other emissions of an acoustic wave;

"n" is a counter index used to number each iteration loop of the analysis phase, thus distinguishing it from the other iteration loops;

"e" is an index assigned to an exact final value of a parameter; s (t) denotes the incident acoustic wave;

sι (t, l) designates the measurement signal received on the i ^th hydrophone following the i ^th firing; eι (t, l) designates the envelope of the intercorrelation between the incident acoustic wave s (t) and the measurement signal

rfi (t, l) designates the reflection signal associated with the measurement signal Si (t, l) by application to the envelope eι (t, l) of a reception window;

Rfi (t, B) designates a readjusted reflection signal, originating from the reflection signal rfj_ (t, l) by a registration step RECAL and associated with the background B;

Rfsι (t, B) designates a fictitious readjusted reflection signal obtained by simulation;

Rfsi (t, B) denotes the k ^{lth eigen} ray of the ^readjusted reflection signal Rfi (t, B); and

θi (j) denotes the rasance angle on the j ^-th Strata k ^e own radius of the reflection signal received at the ^ith hydrophone.

As indicated above, the invention relates (FIG. 1) to a method of geotechnical characterization of a sub-aquatic bottom B, such as an underwater bottom, covered with a sheet of water of total height H determined.

This background B comprises a plurality of strata such as Fl, F2, etc. (Figure 6), generically noted Fj and each presenting its own physical parameters. The first stratum FI forms an interface with water, the other strata extending below this first stratum at various depths.

The method of the invention aims, for each of the strata Fj of a set of strata typically comprising, but not limited to, between two and five strata, to determine several distinctive parameters of this stratum, such as the thickness hj of this stratum , its density p _j , the speed of sound Cj in this stratum, and more precisely the speed C _p j for compression waves and the speed C _s j for shear waves, the attenuation of sound o ( _j in this stratum, and more precisely the attenuation _P j for compression waves and the attenuation _S j for shear waves, as well as possibly the roughness in this stratum, or a sound speed gradient.

The process of the invention essentially comprises an INVESTIG investigation phase, a PRE__TRAIT pretreatment phase and an ANA analysis phase.

The INVESTIG investigation phase itself includes a DEPL displacement operation, an EXCIT acoustic excitation operation, an ACQUI acquisition operation, and a RELEV operation for reading non-acoustic data, these operations being implemented so globally simultaneous. The DEPL displacement operation consists in moving in the water table, by means of a vessel N and at ideally constant reduced speed, an acoustic source 1 and a flute 2 above the bottom B, the flute 2 comprising for example three, four, or five acoustic receivers 21 to 24, also called hydrophones.

The RELEV operation for reading non-acoustic data itself includes a CAPT operation for capturing environmental data and a REPER tracking operation.

The CAPT operation for capturing environmental data consists in recording at any time t the depths of the acoustic source 1 and of a point on the flute 2, as well as the water height H.

The REPER locating operation consists in recording, at each instant t, the X and Y coordinates of the point of the ship N.

The EXCIT acoustic excitation operation is implemented by emitting as isotropically as possible, in particular towards the bottom B and from the immersed acoustic source 1, the incident acoustic wave s (t), of known frequency signature.

The acquisition operation ACQUI is, in turn, implemented by producing, by means of the submerged receivers 21 to 24, respective measurement signals sl (t, l), s2 (t, l), s3 ( t, l) and s4 (t, l) (generically noted sι (t)) in response to detection of respective acoustic waves propagated in water, and by recording these signals as a function of time t.

As shown in Figure 1, the source 1 and the flute 2 stay away from the bottom B and are respectively immersed at depths Zs and Zr, possibly equal, the flute 2 being at least approximately aligned with the source 1 and at least approximately parallel to the bottom B.

The immersion depths of the source 1 and of the receivers 21 to 24 are for example recorded by pressure sensors, the water depth H being for example recorded using a bathymetry sounder.

The total height of the water table above the bottom B being noted H, the source 1 and the flute 2 are thus overall, relative to the bottom B, at an average height Hm typically between 10 meters and 20 meters .

The receivers 21 to 24 are spaced apart from each other, each receiver 'being separated from a neighboring receiver by a distance d which can be constant or evolving along the flute 2.

The flute 2 is considered to be delimited between the two extreme receivers 21 and 24, which are therefore equidistant from the center 20 of this flute. We note ^' Di., the horizontal distance separating the transmitter from the i ^th hydrophone and Zri the depth of the i ^th hydrophone.

The ANA analysis phase, which is intended to allow the identification of physical parameters such as h _j , pj, Cj, cj of each stratum Fj of background B, exploits for this purpose the frequency signature of the incident acoustic wave s (t), and the measurement signals Si (t, l) of the various receivers 21 to 24. The method of the invention first concentrates its analysis on the signals - sι (t) propagated between the source 1 and the receivers 21 to 24.

To do this, the incident wave s (t), as isotropic as possible, is emitted by the source 1, of known directivity, in disjoint successive emission time windows, and the measurement signal Sj _. (t, l) of each receiver is itself analyzed in separate disjoint reception time windows, during which the propagated signals are detected.

Under these conditions, each signal received Si (t) consists of attenuated and delayed replicas of the signal transmitted s (t) for the shot noted Tl, the structure of the received signal Si (t) being thus a function of the geometry of the multiple path as the acoustic wave travels through the water (Figure 5).

This multipath mainly comprises a direct path of the acoustic wave from the source 1 to each of the receivers 21 to 24, a path called "background reflected" wave ^acoustics from the source 1 to each of the receivers 21 to 24 via reflection on the background, the acoustic wave having traversed this path giving rise to the reflection signal rfi (t, l), and a path called "reflected surface" of the acoustic wave from source 1 to each receptors 21 to 24 after reflection from the surface of the water.

In the particular case of bottoms with a low water level H or "shallow bottoms", the path of reflected surface, or even the path called "reflected bottom-surface" involving a reflection of the acoustic wave both on the background and on the surface between its emission by the source 1 and its detection by one of the receivers 21 to 24.

The measurement of the respective arrival times of these multiple paths in the water makes it possible on the one hand to determine with precision the geometry of the device at the time of the measurement and on the other hand to establish reception windows at the inside which the analysis of reflected reflections will be concentrated.

More precisely, the method of the invention concentrates its analysis on the measurement signals originating from waves reflected only once by the same background B. As indicated in FIG. 3, whose different diagrams 3A to 3C correspond to different shots , a sequencing of several consecutive shots then makes it possible to collect the signals denoted Rfι (t, B) originating from the reflection of the incident acoustic wave on the same background area B and under as many different shaving angles as there are receivers such as 21 to 24.

As shown in FIG. 4, the different reflection signals rfj. (T, l) collected, during a shot 1, on each of the receivers 21 to 24 in an observation window corresponding to the reception of a single echo on the bottom must therefore be respectively associated with different reflection signals rfi (t, l '), rfi (t, l''), rfi (t, l''') etc., collected, during other shots l ', l'',l'^'1 , etc. on each of the receivers 21 to 24 to form the so-called "background reflected" signals or reflection reset Rfi (t, B), this operation being carried out in a RECAL registration step which will be detailed later.

In reality, each background reflected signal Rfj_ (t, B) is itself composite in the sense that it is formed (FIG. 6) by the propagation, along a multiple path, of the acoustic wave incident in the background B , giving rise to reflections of this incident wave on the different strata Fj of the sub-aquatic subsoil.

Under these conditions, each signal of - reflected background Rfi (t, B) can be described in the form of a set of natural radii denoted Rfsι (t, B), the index k referring to the kth ^natural radius associated with the incident wave Rfι (t, B). The amplitude ratio between the incident and reflected time signal at the interface Fj is directly proportional to the ^' Fourier transform of the reflection coefficient R _j (f, θi (j)), the index k here referring to l 'angle θ _t (j) with which the k ^{lth eigen} ray is reflected on the interface Fj.

However, this reflection coefficient Rj (f -, θ _± ^k (j)), within the framework of a Rayleigh modeling, is linked to the geotechnical parameters of the different strata Fj-1 Fj of the subsoil.

The signals thus produced and used bear the signature of the environment encountered by the detected waves, and therefore contain sufficiently rich and relevant information of the background B to allow to unambiguously extract from it the geotechnical parameters of the different strata Fj of the subsoil.

The echo of an incident wave arriving at the bottom B under a very grazing ance raking angle is practically sensitive only to the acoustic attenuation of the stratum FI.

The echo of an incident wave arriving at the bottom B under an incidence perpendicular to the bottom B is practically sensitive only to changes in acoustic impedance and presents in particular the defect of not being able to allow unambiguously the determination of the thickness-torque speed of each of the strata.

In general, it is therefore preferable to recover signals having undergone reflection at fairly low shaving angles on each of the strata Fj making up the background.

Thus, moderate rasance _^ angles are needed to describe the first stratum FI and larger rasance angles are needed to describe the deeper layers, F2, etc. It therefore appeared very expedient to ensure, for the invention, that each reflected wave reaching one of the receivers 21 to 24 in a reception time window is the result of the reflection of an incident wave having reached the bottom B at a shaving angle θ _k at most equal to 70 degrees and at least equal to 10 degrees.

Although Figures 5 and 6 are both completely schematic and incomplete given their simplified and purely spatial presentation, they intuitively illustrate what the multiple paths of the incident wave are, respectively in the water and in the bottom.

The importance discussed above of the angles of emission with respect to the reflected background can be observed on the curves of FIGS. 7A to 7J, corresponding respectively to the waves collected on receivers arranged at distances corresponding to angles of razor to the fundus. 'd decreasing from 72 to 25 degrees.

Thus, it appears that the curve 7B, which corresponds to strong shaving angles, provides little information relative to the curve 7A since for these angles, the reflection coefficient of each stratum is practically independent of the shaving angle. Similarly, it appears that the curves 71 - and 7J which correspond to small shaving angles no longer provide any information since the waves do not penetrate the background. Conversely, FIGS. 7B to 71, which correspond to a relevant angular scan, exhibit a high variability making it possible to find the characteristics of the background without ambiguity.

Likewise, while the signature of a very low frequency incident wave is difficult to control, and a high frequency wave has a zero penetration capacity in the deep strata, it appeared very expedient, for the invention, using an incident wave having a relatively low frequency component at least equal to 100 Hz, and a relatively high frequency component at most equal to 8 kHz, this incident wave having for example, in each window emission frequency continuously variable between 100 Hz and 8 kHz.

The relation linking the penetration depth of an incident acoustic wave to the frequency of this wave is clearly illustrated by FIGS. 8A to 8J, which represent, for 5 shaving angles θi different from the incident wave between 25 and 70 degrees , the response Rfi (t, B) of the time signal reflected at the bottom, for a transmission signal comprised between 400 and 3200 Hz (FIGS. 8A to 8E) and comprised between 5400 and 8200 Hz (FIG. 8F to 8J). Although having the same temporal resolution as the low frequency signals, the high frequency signals are appreciably impoverished by the attenuation of the waves in the various strata.

Thus, thanks to an appropriate choice of the razor angle and thanks to a variation of the acoustic excitation frequency, the physical parameters hj, pj, Cj, cj of each stratum Fj can be deduced from the incident wave s ( t) and measurement signals sι (t, l) via a physical model involving the acoustic reflection coefficients R _j (θι (j)) of the strata Fj of the background B seen by the natural rays Rfsι ^k (t, B), their attenuation αj, as well as their frequency dependence.

In other words, the physical model used makes it possible to take advantage of the angular dependencies of the reflection coefficient and the frequency dependence of the attenuation in the strata.

In practice, it is often useful to ensure that each reflected wave reaching the center 20 of the flute 2 in a reception time window is the result of the reflection of an incident wave having reached the bottom under a raking angle θi at most equal to 70 degrees and at least equal to 10 degrees, the raking angle θi even being able to more often be ideally chosen in an angular range between 25 and 65 degrees.

Each receiver 21 to 24 is then separated from the acoustic source 1 by a distance D of between approximately 1 and approximately 4 times the average height Hm, the value D = Hm essentially relating to the receiver 21 closest to the source 1, and the value D = 4. Hm rather relates to the receiver 24 furthest from the source 1.

Under these conditions, each receiver, such as 21 to 23, is spaced from a neighboring receiver, such as 22 to 24, by a distance d allowing the grading angle to be as regular as possible.

In practice, the transmission time windows each have a duration at least equal to 0.1 seconds, and preferably rather at least equal to 0.5 seconds. In general, the duration chosen is a compromise between a short signal minimizing the effects of the movement of the ship and a longer signal increasing the signal to noise ratio.

These transmission windows are repeated periodically with a period depending on their individual duration, at least equal to 0.5 seconds, and preferably rather at least equal to 1 second. In general, the duration of silence between two shots is a compromise between a long duration allowing satisfactory separation of signals received and a short duration allowing greater resolution on the ground.

In the preferred case, the measurement signals are analyzed in increments of 2 seconds and processed by shot, a shot corresponding to a set of signals formed by a transmitted signal and corresponding measurement signals received by the respective receivers 21 to 24.

Between the INVESTIG investigation phase as previously described, and the ANA analysis phase which will be described in more detail later, the method of the invention comprises a pre-treatment phase noted PRE_TRAIT.

This PRE_TRAIT preprocessing phase includes a FIL_COR filtering operation, an INVER_GEOM geometric calculation operation, and a RECAL registration step. During the FIL_COR operation, each measurement signal Si (t) is filtered to normalize this signal., Taking into account the frequency signature of the incident acoustic wave s (t).

During the INVER_GEOM operation, the ^'geometry of the instrument during firing is recalculated and the temporal reception windows are identified in order to extract the reflection signals rfi (t, l).

Finally, during the RECAL step, each sample measurement signal is associated with a geographic position of the base referenced by its coordinates X and Y, and readjusted to ^'' to reflect the ship's movement between each shot and shot next, the RECAL step thus producing a set of reflexed reflection signals Rfi (t, B) having seen the same background area B during different shots and under different shaving angles.

The filtering carried out during the FIL_COR operation, called "adapted", is obtained by estimating the envelope of the intercorrelation between the transmitted signal represented by the incident acoustic wave s (t), and each measurement signal Si ( t, 1). The filtering operation first consists in performing the correlation: c _t (t, l) = js (τ) .s _i (t-τ, l) dτ,

then carry out an envelope extraction of the signal Ci (t, l) which is carried out for example using a demodulation phase according to c _i (t, V) ^{.e ~ Λπfct} , where f _c is the frequency central of the signal transmitted and followed by low-pass filtering at the cut-off frequency F _CO u _P > (Fmax-Fmin) / 2 to obtain the signal eι (t, l).

The calculation of the exact instrumental geometry carried out during the operation INVER_GEO is based on the identification of the reflected waves arriving successively on the various receivers 21 to 24.

This identification is obtained for example from a maximum detection inside time windows which are defined from the a priori geometry of the instrumentation. The measured times (τe ^k ) _m corresponding to this detection are compared to the times (τei) _s from a model of clean rays in a stratified medium providing the expression of the theoretical impulse response:

Rieχt) = ∑A _i ^k .δ (t- τef) Equation 1 fc = ι where N represents the number of acoustic paths that can connect a receiver to the transmitter, where t represents the time variable, where τeι represents the delay that presents , with respect to the emission of the incident wave, the reception of a wave having followed the k ^th path, where δ represents the Dirac function, and where i ^k represents the relative amplitude, compared to that of the incident wave emitted, of the wave having followed the k ^th path and reaching the receiver with a delay τeι, the notation τe recalling that these are delays associated with the multiple path of the acoustic wave in water.

The delays associated with the direct path and the reflected background are successively given by: i ^ (Zs - Zr + D ² ₂ J (2.H-Zs-Zr _l ) ² + D _l ² τe. = - - and τe, = - - ¹ C av C av

where C _m cy is the average speed of sound in the column of water crossed by the wave having followed this path and in the particular case of the deep sea where surface reflections are not exploited.

The theoretical instants predicted for the different eigen radii calculated are compared to the instants of arrival waves actually detected by the different receivers 21 to 24.

In the event of a non-negligible difference, a search for the exact geometry can be carried out for example by a simulated annealing algorithm with 2.N variables (Di and Zr _± ) by minimizing a cost function expressed from the differences between the theoretical instants associated with the clean rays and the instants of detection of the waves collected on the N different receivers 21 to 24.

In the particular case of shallow water, other arrivals such as reflected-surface and even reflected-surface-bottom or background-surface are detected and make it possible to dispense with the preliminary measurements of H and Zs, quantities which will be directly sought by the geometric inversion algorithm described above.

From the information then available, which is constituted, for each shot and each receiver "i", by the signal Si (t) coming from this receiver and by the geometry of the system, the time range corresponding to the reception window in which the received signal will be analyzed, that is to say the time range of reception of the only contribution of the reflection signal rfi (t, l).

This selection is for example carried out, in the particular case of shallow water, by a program which, for each envelope of the signal e1 (t, l) coming from the i ^th receiver selects the time window defined by: r (t, = e (e, ³ -Nt _e : re _; ⁴ -Nt _g , /)

where ":" is a symbolic separator placed between the lower time bound and the upper time bound of the selected reception window, where τei ³ represents the delay associated with the reflected _. background, where τei ⁴ represents the delay associated with the first reflected surface-background, where Ν is a parameter depending on the nature of the signal transmitted, and where t _e is the sampling period of each receiver, the indices 3 and 4 being linked to a particular case of shallow water in which the order of arrival of the own rays would be successively the direct path, the reflected-surface, the reflected-background and the reflected-surface-background.

The RECAL registration operation consists in using the position data (X, Y) with the positions of the instruments associated with the corresponding shot so as to calculate the positions Bi ¹ (X, Y) of the bottom seen by the hydrophone i during the shot 1. By repeating the operation for several consecutive shots, it becomes possible to reconstitute the data set Rfi (t, B) made up of the reflected reflections having seen the background B (X, Y) under the different shaving angles θi .

For example, for a configuration defined by a distance of 10m between the source 1 and the first hydrophone 21 of the flute, for a flute with 4 hydrophones separated from each other by 10m, horizontal and located 10m above the bottom, for a ship moving at a regular speed of 5m / s and with a duration of ^' 2 seconds between successive shots, we will successively apply a shift one shot for 2 ^-th hydrophone 22 of two shots to the 3rd hydrophone 23, and three shots for the 4 th hydrophone 24 as mentioned in Table 1 below.

Table 1: Principle of the retiming operation

Once identified and extracted, for each receiver, the detected wave having undergone a single reflection and through the bottom B, the ANA analysis phase is implemented to allow to deduce from this wave the geotechnical parameters of the bottom by means of '' a model based on the expression of the Rayleigh reflection coefficient. As mentioned above, the modeling of the interactions of acoustic waves with the background and its different strata is based on the representation of each stratum on the basis of several parameters including the thickness h of this stratum, the density p (for example in g / cm ³ ), the _. speed of sound C _{p / S} (for example in m / s) and attenuation of sound α _{p / s} (for example in dB / λ); the indices p and s relating respectively to the compression waves and to the shear waves. In the sedimentary layers, roughness or even a speed gradient can in some cases be added to these main physical parameters. Starting from the fact that the reflected background decomposes itself as a sum of eigen rays reflected by the different strata of the background, we can also write its impulse response according to:

RIbχt) = ∑4.δ (t-τbf) 4 = 1

τbi ^k referring to the fact that it is the multiple path of the acoustic wave in the background.

Considering now E (t), the autocorrelation of the envelope of the transmitted signal, we obtain Rfι (t) by:

Rfs t (t, B) = E (t) ®

the index s specifying that it is the reflected simulated background.

Concerning the determination of these eigen radii, it is in particular a question of calculating the k firing angles (raking angle at the level of the source '1) of the k eigen radii sought.

We start from the angle θi ° defined by:

We then perform a dichotomous search for -'firing angle θi (0) between θi ° and θi ° + θ _max using the relations: cos # ¹ (j) _ cosfl ¹ (7 + 1) between strata Fπ and Fη + 1, in

noting θi ^k (j) the angle in the layer Fj of the departing ray with a firing angle θi ^k (0), and

_D _ (2.H-Zr-Zs) _| 2.N (j, k) .h (j) _{| „} So much? (0) i tanθfϋ)

by considering here the k ^{lth eigen} ray of the reflected background associated with the geometry of the sensor i as the eigen ray reflecting in the deepest on the j ^th stratum, and by noting N (j, k) the number of reflections of the radius k in the layer Fj.

A test must be carried out beforehand to ensure the existence of this radius N (l, k).

In the case where Ai is a function of the frequency, we have:

Rf _If (t, B) = ∑4 (t) .E (t-τhï) S

In the particular case of flat interfaces and without multiple reflection in any layer,

where dj is the distance traveled in the stratum Fj, tj = d _j / cj the corresponding time, α _j the sound attenuation in the layer expressed in Neper / m, and R _j (f, θ _x ^k (j)) the reflection coefficient at the j ^th interface.

The Rayleigh reflection coefficient associated with a fluid-fluid interface is defined by:

where we use complex celerities

C, i.a 1 + - 2. ^. 8,686

This model being posed upstream of any investigation, the ANA analysis phase can use it to extract the geotechnical parameters of the different strata of the sub-aquatic subsoil from the reflexed and corrected reflection signals Rf (t, B ) obtained at the end of the PRE_TRAIT phase.

To do this, the ANA analysis phase uses for example a shape matching technique known to those skilled in the art, in which case it consists of traversing, in an iterative manner, a loop including a MODEL modeling operation, an operation SOUST subtraction, and an OPTI RETR optimization and feedback operation. It will be recalled that, by convention, the index "n" will be used below to designate generically the current number of times that the iteration loop has been traversed since the start of the ANA phase, and to identify any magnitude obtained at the ^nth step of the iterative path of this loop.

MODEL modeling consists in calculating, at each step n, the respective fictitious reflexed reflection signals Rfi ⁿ (t, B _n ) that the receivers 21 to 24 would produce if the background B were constituted by a background noted B _n formed of strata Fj defined by fictitious physical parameters h _n j, p _nj C _n j, o. _n j virtually attributed to these strata.

The initial fictitious physical parameters h _Q j, p ₀ j, C ₀ j, _0j are for example chosen on the basis of general geological indications on the site to be explored, or failing this at random in ranges of plausible or simply possible values , the values taken by the parameters h _nj , p _nj , C _nj , c. _nj for _s values _. non-zero of n being defined step by step by the repeated path of the iteration loop.

To these parameters can be added the number J _n defined as the total number of strata assumed, during the execution of the ^nth step of the iteration loop, to have been detected, this number evolving until obtaining, at the end of the iteration, of the exact total number J _e of strata actually detected.

To obtain the exact parameters, the ANA analysis phase firstly includes a SOUST subtraction operation during which the deviations Δ _ni that present, compared to the fictitious re-aligned reflection signals RfSi ⁿ (t, B _n ) previously calculated, the re-aligned filtered filter signals and corrected Rfi (t, B _e ).

Then, the ANA analysis phase includes an optimization and feedback operation 0PTI_RETR at the end of which the fictitious physical parameters are revised according to the deviations observed.

To do this, the method of the invention calculates, at each iteration step n, a cost function given for example by:

K (n) = ∑ (A _ni Ϋ (= 1

with Δ _ni = RfSi ⁿ (t, B _n ) - Rfi (t, B _e ).

This cost function is for example minimized according to a simulated annealing algorithm to produce a refined impulse response Rfsi ^{n + 1} (t, B _{n +} ι) defining a new background B _{n +} ι used in a new iteration step of the phase ANA , the process converging towards the production of exact parameters h _e j,

C _e j, α _e j of the different strata j.

The table below gives an example of a concrete case treated. Geometry: H = 25m, Zs = 7m, Zri = 8m, Dι = 12m, di = 5m, J _e = 5 Background: Cpl ^e = 1750m / s; ρl ^e = 1.8; αl ^e = .0.5; hl ^e = 0.75m; Cp2 ^e = 1970m / s; p2 ^e = 1.93; 2 ^e = 0.4; h2 ^e = 1.75m; Cp3 ^e = 1793m / s; p3 ^e = 1.85; 3 ^e = 0.5; h3 ^e = 0.45m; Cp4 ^e = 2050m / s; p4 ^e = 2 .; α4 ^e = 0.38; h4 ^e = 2.25m; Cp5 ^e = 3000m / s; ρ5 ^e = 2.3; α5 ^e = 0.2; Cs5 ^e = 1300m / s; 5s ^e = 0.3

As will be understood by a person skilled in the art, instead of using a cost function such as K (n) calculated by means of a Euclidean distance, the ANA analysis phase could just as easily use the search for a maximum of correlation between the signals RfSi ⁿ (t, B _n ) and Rfi (t, B _e ).

However, in both of these cases, it is a question of revealing differences between these signals and of making the signal RfSi ⁿ (t, B _n ) evolve in a direction suitable for reducing these differences, the operations previously mentioned of SOUST subtraction and of the formation of such deviations must be understood here in a broad functional sense including in particular such a search for maximum correlation.

Claims

CLAIMS.

1. A method of geotechnical characterization of a sub-aquatic bottom (B), such as an underwater bottom, covered with a water table of total height (H) determined, and comprising a plurality of strata (FI , F2) forming between them separation interfaces, each presenting its own physical parameters (hj, p _j , C _j , c. _J ) and extending to various depths below a first layer (FI) which forms a separation interface with water, this process comprising at least one investigation phase (INVESTIG), a pretreatment phase (PRE TRAIT) and an analysis phase (ANA), the investigation phase (INVESTIG) including itself an acoustic excitation operation (EXCIT), implemented by emitting at least towards the bottom (B), - from an immersed acoustic source (1), an incident acoustic wave (s (t)) of frequency signature known, an acquisition operation (ACQUI) implemented by producing, by means of a flute (2) at least at least four receivers - (21 to 24) immersed, respective measurement signals (Si (t, l)) resulting from a detection of respective acoustic waves reflected by the bottom (B), and a reading operation (RELEV) geometric data (Zs (t), Zr (t), H (t), X (t), Y (t)), the flute (2) being at least approximately aligned with the source (1) and distant from the bottom (B), the receivers (21 to 24) being spaced from each other, and the incident wave (s (t)) being transmitted in each window of a succession of disjoint transmission time windows, and having a variation time frequency within each transmission window, characterized in that the phase investigation (INVESTIG) includes a displacement operation (DEPL), concomitant with the excitation (EXCIT) and acquisition (ACQUI) operations, and implemented by simultaneously moving the acoustic source (1) and the flute ( 2) in the water table, in that the pre-treatment phase (PRE_TRAIT) is implemented by deducing, from the measurement signals Si (t, l), from the signature of the incident acoustic wave (s (t)), and geometric data (Zs (t), Zr (t), H (t), X (t), Y (t)), reflection signals rf _± (t, l) having reaches different receivers (21 to 24) following the same incident wave emission, and by grouping, in the form of series of reset reflection signals (Rfi (t, B)), reflection signals rfi ( t, l) having reached, under different shaving angles (θi), different receivers (21 to 24) coming from the same bottom area following different transmissions of the incident wave, and in that the phase of analysis (ANA) is implemented by deducing the physical parameters (h _j , p _j , Cj, _j ) of each stratum (Fl to F4) of the background, from the series of readjusted reflection signals (Rfi (t , B)), by exploiting the angular and frequency dependence of the reflection coefficient of the incident wave on each interface, each reflexed reflection signal (Rfi (t, B)) being interpreted as a sum of delayed and attenuated arrivals of clean radii corresponding to the reflection of the incident wave by the different strata of the bottom zone to which this readjusted reflection signal (Rfi (t, B)) corresponds.

2. A geotechnical characterization method according to claim 1, characterized in that each reflected wave reaching a receiver comes from the reflection of an incident wave having reached the bottom under a raking angle (θi) at most equal to 70 degrees and preferably at most equal to 65 degrees.

3. A geotechnical characterization method according to any one of the preceding claims, characterized in that each reflected wave reaching a receiver is the result of the reflection of an incident wave having reached the bottom under a raking angle (θi) at least equal at 10 degrees and preferably at least 25 degrees.

4. A geotechnical characterization method according to any one of the preceding claims, characterized in that each receiver (21 to 24) is separated from a neighboring receiver (22 to 24) by a distance (d) corresponding to a difference of raking angle at least equal to 2 degrees, and preferably at least equal to

5 degrees, reflection signals respectively received by this receiver and by the neighboring receiver.

5. A geotechnical characterization method according to any one of the preceding claims, characterized in that the incident wave has, in each emission window, at least one relatively low frequency component, the frequency of which is at least equal to 100 Hz.

6. A geotechnical characterization method according to any one of the preceding claims, characterized in that the incident wave has, in each emission window, at least one component of relatively high frequency, the frequency of which is at most equal to 8 kHz.

7. A geotechnical characterization method according to any one of the preceding claims, characterized in that the incident wave has, in each emission window, a continuously variable frequency between a relatively low frequency component and a relatively high frequency component .

8. A geotechnical characterization method according to any one of the preceding claims, characterized in that the transmission time windows each have a duration at least equal to 0.1 seconds and preferably at least equal to 0.5 seconds.

9. A geoteçhnic characterization method according to any one of the preceding claims, characterized in that the transmission time windows are repeated periodically with a period at least equal to 0.5 seconds and preferably at least equal to 1 second.

10. A geotechnical characterization method according to any one of the preceding claims, characterized in that the physical parameters are at least partly chosen from the set of parameters comprising the thickness (h), the density (p), the speed sound (C _p , C _s ), sound attenuation (α _p , α _s ), roughness, and a sound speed gradient.

11. A geotechnical characterization method according to any one of the preceding claims, characterized in that it includes, upstream of the analysis phase (ANA), an operation (CAPT) for determining the immersion depth (Zs) of the acoustic source and / or the immersion depth ( Zr) from a point on the flute and / or water level H.

12. A geotechnical characterization method according to claim 10, characterized in that the analysis phase (ANA) is implemented by iteratively traversing a loop including a modeling operation (MODEL_FOND) during which are calculated , from fictitious physical parameters (h _n j,

P _n j _Λ C _nj , c. _n j) virtually allocated to the first stratum (Fl) at least from the plurality of strata, and for the various receivers (21 to 24), respective fictitious reflexed reflection signals (Rfsi ⁿ (t, B _n )) originating from measurement signals virtually produced by these receivers (21 to 24) in the presence of these fictitious parameters (h _n j, p _n j, C _n j, α _nj ), a subtraction operation (SOUST) during which are formed '' deviations (Δ _n i) observed between the respective reset reflection signals (Rfi (t, B)) coming from the measurement signals (Si (t, 1)) actually produced by these receivers (21 to 24) and the signals from reflection matched fictitious corresponding (RfSi ⁿ (t, 'B _n )), and an operation of optimization and feedback (OPTI_RETR) at the end of which the fictitious physical parameters are revised according to the deviations observed.