Review Reports

Round 1

Reviewer 1 Report

This papers continues the research described in a series of works by D.R. Dowling and coauthors on the analysis of the so-called autoproducts. The autoproduct is the product of two solutions of the Helmholtz equation at two different frequencies. There are reasons to expect that the autoproducts  may mimic genuine acoustic fields at the difference and sum frequencies. The analysis of autoproducts allows one to obtain information about the components of the field at frequencies outside a recorded field’s bandwidth and apply this information for acoustic remote sensing.

 

This paper compares the measurement data of the spatial correlation of the acoustic field and its autoproducts on the aperture of a towed horizontal array about 8 km long. The antenna recorded signals emitted by the airgun array in the 10-200 Hz band. In the processing, the field correlation functions at the antenna aperture were obtained. This is done for both the registered field (genuine field) and autoproducts. Since the frequency bands of autoproducts and genuine field partially overlapped, it was possible to compare the corresponding correlation functions. It turned out that, as a rule, the autoproducts have a better spatial correlation than the genuine field. To explain this effect, the authors present some considerations based on the analysis of destructive interference of upward propagating and downward propagating surface reflected sound. However, why auto products are better correlated than genuine field needs further research.  In Table 1, the found field spatial correlation scales are compared with the correlation scales measured by other authors in other experiments. It can be seen from it that the estimates of the scales of spatial correlation found in this paper agree in order of magnitude with the data of other authors, obtained in most cases at much greater distances.

 

The most interesting, in my opinion, results were obtained in Sec. 3.3 on matched field processing (MFP). The authors use MFP to solve the source localization problem. In this case, it is an estimate of the distance to the source. It is shown that the use of autoproducts for this purpose is more effective than the use of genuine field. Uncertainty functions for autoproducts presented in fig. 9 and 10 have maxima at distances corresponding to the true position of the source, while there are no such maxima for genuine field. The authors explain this effect by the fact that the autoproduct is significantly more strongly correlated at the antenna aperture than the genuine field. This result shows that the autoproduct is unexpectedly close to the replica weight vector calculated for the genuine field using the idealized environment model.

 

I think that the article is interesting and can be published in the present form.

Only one thing should be corrected: in the top panel of Fig. 1 there is no colorbar that establishes a relationship between the color and bottom depth.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

In this paper, the authors report results for the spatial coherence of the genuine field, the frequency-difference autoproduct, and the frequency-sum autoproduct as determined from data collected during the Cascadia Open-Access Seismic Transects (COAST 2012) experiment. The work looks solid and interesting. Some suggestions are presented as follows.

1. The authors are suggested to reorganise the main contribution of this paper in Section to highlight the meaning of the sea trial. 

2. Matched field processing is the common way for sonar signal processing so the authors are suggested to briefly summarise Section 2.3.

3. As figure 2 plotted the layout of sea trial, the author are suggested to discuss a little more about submarine topography effect.   

Author Response

Please see the attachment.

Author Response File: Author Response.pdf