In drop-out modelling, a number of constraints on whether or not a spare may be substituted for a dropped-out gun are provided. One of these allowed a 25% volume tolerance on the dropped out gun in the choice of a spare. In other words if a spare passed the other constraints on type and proximity and had a volume within 25% of the volume of the dropped-out gun, it would be allowed. This has been extended to add a 50% tolerance, allowing the user to choose 0% (precise volume match), 25% tolerance of dropped-out gun volume or a 50% tolerance of the dropped-out gun volume.
We have also clarified the relationship between the normalised cross-correlation calculated in the drop-out calculation and the Pearson normalised cross-correlation calculated in the array comparison option. There were minor differences, (the drop-out used a 99.95% energy window whereas the Pearson comparison did not). These have now been rationalised and in essence are the same calculation.
After user request, we have added the array geometry file (.sba file) to the zip downloadable signatures package for full reports, full reports with directivity, environmental reports and signature sweep reports.
Both TPS and GEMINI low-frequency source modelling has been added.
In the case of the TPS source, the match with the Chelminski et. al SEG 2021 paper "Sea trial of a low frequency enhanced pneumatic source" is very close. They show a 26500 cuin TPS at 10m as having a peak amplitude of around 213dB rel. to 1 microPascal/Hz at 1m at 2.8Hz. The Gundalf model gives 213dB rel. to 1 microPascal/Hz at 1m at 2.9Hz.
For the GEMINI source, the match with the ION datasheet is also very close. This gives a 4000 cuin Gemini source at 8m as having a peak amplitude of 209dB rel. to 1 microPascal/Hz at 1m at 3.4Hz whereas the Gundalf model gives 209dB rel. to 1 microPascal/Hz at 1m at 3.9Hz. The 2x4000 cuin GEMINI source is comparably accurate. See the Help -> Calibration sheet for more details.
In this version of Gundalf, there are 25 distinct airgun types to choose from including the two new low-frequency choices. We considered this to be too unwieldy so they have been re-organised in exactly the same way as the filters were re-organised in C8.3k. There is now an "Edit airgun menu" option in the main menu (the one underneath the user's email address). This allows you to select which airgun types you would like to appear in the "Add Gun" menu as you build arrays.
In this version of Gundalf, the full sweep of generated signatures is saved as a SEGY file and can be downloaded by the user along with the other signature downloads.
The half msec filters inApril_geo_lin_0p5ms_noLC.flt and inApril_hyd_lin_0p5ms_noLC.flt have been added.
Gundalf now contains
All are accessible but other than a few defaults, must be enabled by the user with the main menu option "Edit filter menus".
The Toolset reference has been updated to reflect the edit filter and airgun menu options and add a short section on personal configuration with a stateless system like Gundalf Cloud. The Filtering and Signature statistics knowledgebase files have been updated.
Allowing independent specification of the modelling sample interval in Gundalf is an anachronism. It is not in fact constant in any case as the algorithm automatically adjusts it to maintain accuracy constraints and varies between a few microseconds and the value which you could formerly specify, so what you ask for is not necessarily what you get. Furthermore, as we have received more and more calibration data, it became obvious that the modelling sample interval was inextricably linked with the self-consistency of calibration data from multiple sources. We have finally therefore removed the option to make it user-specifiable. The output sample interval remains, as it should, an important part of modelling but Gundalf will assume full control over what needs to be done with respect to the modelling sample interval. In most cases, this will make no difference but will make it clearer for the user.
In view of the importance of filtering in signature modelling, the applicable pre- and post- internal or external filters have been added to the Drop-out, the Optimisation, the array stability, the array comparison and the signature sweep reports.
All the signatures in the sweep are now plotted alongside the existing amplitude spectra.
The legacy version of Gundalf could create basic reports (signature model with or without directivity) shrouding the details of individual guns in the array. This option is one of the few we had not included in the Cloud version. It is now included on the modelling parameters page but is only enabled for the basic reports. It has no effect on any of the other reports.
We have included some information on a study on zooplankton by McCauley et. al. (2017) (reference given in full in the report). This study measured the impact of a single G-GunII on zooplankton. Amongst other things, they give detailed field measurements of the airgun in SPL dB. These measurements agree exactly with a Gundalf model of the same setup. More information is included under Help -> Calibration.
The FAQ have been updated to give tips on how to create an organisation wide set of filter menus. The Calibration notes have been updated to reflect the zooplankton study.
Creation of filter panels for internal filters could fail. This has now been corrected. External filter behaviour is unaffected.
Now that the external filter availability has been extended to all filters, we have introduced a standard filter information sheet for all pre- and post- specified filters. These appear in the report and show the impulse response, amplitude spectrum and a range of metrics for both domains to help classify filters. These filter information sheets appear in standard and directivity reports.
In the full list of filters produced by the edit filter menu, we have added the capability to view the filter information sheet for a particular filter just by clicking on the view option.
We have further upgraded this by adding the ability to include pressure field interaction compensation or not under user choice. This allows these interesting and little known effects to be studied in detail.
We have added a complete suite of GPR filters kindly supplied by Sercel. The total number of filters in Gundalf is now approaching 3,000.
There have been numerous documentation changes accompanying the changes to external filters and also in documenting required signature lengths for very long period airguns in environmental reports. Further changes to information messages issued in marine vibrator modelling have been done to clarify the wording.
We fixed a defect which caused graphics output to fail in some circumstances.
Historically, we added external filters as they were made available to us. This slowly grew and we were able to make them all available within a menu system. In the last few years, this has all fallen apart. We had to start leaving them out because there were simply too many - currently 2882 with several hundred more coming after we have processed them. We added the ability to upload them but this was a stop-gap only.
In this release we finally gave up with this piecemeal approach so that you could get access to them all. We have added a new menu item under the system menu (the drop-down under your email address) called "Edit filter menus". This allows you to tailor the menus yourself choosing which ones you would like to appear when you get to the filter parameters page. Most people have their favourite filters, sort of like a household pet, so you will now be able to create your own menus containing only your favourites. Alternatively, you could get out more.
Currently, your selection will only last for a session and then it will revert to the default. If you wish your selection to last between sessions, you will be able to do this using the existing Save/Restore session menu items. We will continue extending this functionality in upcoming releases. We hope you enjoy this option and that it incrementally makes external filters easier to deal with and manage whilst greatly extending your choices.
When arrays contain very large amounts of pressure field interaction, the energy calculation becomes progressively unreliable. (Typically when efficiency is above 80% or so.) Such arrays are very rare but we have added a documentary warning in the reports when this occurs.
Due to the rarity of their use, these were not migrated from the legacy version. However there are still some older array definitions out there so they have been re-implemented. Please note as before, wave-shape kits may compromise calibration quality a little due to the paucity of high-quality relevant measurements.
Two defects were reported and fixed. The first affected the production of plots with alternative sources. The second prevented array centres from being computed for very shallow arrays (< 0.6m).
When spares are available in an airgun array and Gundalf is instructed to use them if possible, Gundalf will try to find the best combination of spares to mitigate a drop-out. Mitigation can be complete i.e. the array passes, or a minor failure i.e. the array fails on only one constraint or a major failure i.e. the array fails two or more constraints. The documentation for this in the quick reference html app was not very clear. This has now been re-written so that it is obvious which spares had to be used for whatever level of mitigation Gundalf was able to achieve.
It is important to realise though that in both the report and also the quick reference app, Gundalf has already found the best solution it can from the specified spare guns and their constraints. If a particular drop-out combination still fails, the best use of spares has already been applied. Both report and quick reference app have been re-documented to make this clear.
Amplitude spectral scaling in geophysics is sometimes a little opaque with the SEG standards seemingly sanctioning the use of positive frequencies only although the documentation is ambiguous and the worked examples in both the SEG standard and the seminal Fricke paper are insufficiently detailed to be able to tell. This may be a little abstract but the bottom line is that when amplitudes are scaled in this way, the resulting amplitude spectrum and the original signature do not obey one of the most fundamental theorems of Fourier analysis, Parseval's theorem. Gundalf used this historic method until the legacy version v8.1h (30-Mar-2018) after which it was decided that whatever historic practice was, it was not really appropriate to breach such a fundamental theorem and Gundalf was changed from then on to obey Parseval's theorem and an advisory issued. If the legacy method is used, amplitude spectral values 3dB smaller than current Gundalf will be the result. Energy calculations are however unchanged as they calculated in the time domain using the notional source method developed by Laws et. al. (1988).
Gundalf allows either legacy (SEG) scaling or Parseval scaling to be used. The default is Parseval.
We have tweaked this a little to give better performance for band-limited signatures or those filtered with filters with an incorrect index for time zero. If Gundalf can't find a bubble reliably it will say so. Judicious choice of a bubble search start time (which is now annotated on the signature plot), may enable it to find one, however when it is difficult to find a bubble, it may not even be sensibly defined. The bubble in a signature is a legacy of the days when the only tools available to de-bubble were based on predictive deconvolution, which itself needed a primary to bubble in excess of around 3 to work well. Today, things are much more sophisticated to the point that sources in which the concept of a bubble is completely absent, e.g. marine vibroseis, are used.
Enhanced the Optimiser with a new kind of constraint:
Note that if lateral symmetry in the end array is required, guns must be fixed symmetrically at the start. Gundalf will then maintain this symmetry. If lateral symmetry is required and guns are fixed non-symmetrically, Gundalf will be unable to find a solution to improve the array.
An additional column has been added to show the approximate zero to peak contribution of each gun. This may assist in near field phone calibration.
We have also been requested to add the array centres section to the environmental report. This is also available in this release.
Completed work left over from Gundalf legacy so that x,y,z are now written into notional source SEGY files.
The maximum size allowed for the XLA (Extra Large Airgun source) has been increased to 30,000 cuin. The XLA source is still experimental and under development but has been recalibrated and is now the basis for modelling both TPS and Gemini low-frequency source modules.
Clarifications added in a few places, for example that environmental modelling will always use the anelastic estimated surface reflection coefficient. Also error messages for uploaded filters clarified.
The names of externally supplied filters are preserved in Gundalf for familiarity. The downside is that they can be a little inconsistent and opaque. We have added some explanatory text to help. Its not perfect but should help as there are literally hundreds of external filters.
User feedback suggested that the SEG standard method of showing the cumulative energy spectrum simultaneously with the amplitude spectrum as a dual plot could be a distraction to users unfamiliar with this form of display. We have therefore made it optional whether you would like to see the cumulative energy spectrum or not. It is by default not shown.
An array comparison option has been added to allow the comparison of arrays separately consisting of different source types, for example an airgun array with a sparker.
The one msec filters SBGS_M_Geo_Lin_out_200_1p0.flt and SBGS_M_Hyd_Lin_out_200_1p0.flt have been added.
The main signature plot previously used to show the total number of guns for airgun arrays. This is a little confusing if there are spares, so it now shows the total active guns.
In the array geometry tables, the source type is now shown. This is fairly obvious for normal reports but it clarifies the array comparison reports.
These are now computed using a different algorithm. Because they are difficult to interpret (they involve calibration data as well as the model itself), they are turned off by default but we have added an option to turn them on in the array summary table. They are intended only to give a rough idea of the worst case errors in the measurement data and should not be used in array comparisons.
The original None, Harmonic and True GI modes have been expanded. They used to mean Generator:Injector = 100:0, Generator:Injector = 50:50 and Generator:Injector = 30:70. They have been enhanced in the Sercel series to have different values depending on the volume, for example the Harmonic 255 has a Generator:Injector ratio of 59:41. Gundalf has been modified to use these internally and set the injector delay appropriately. The engine will currently reject non-standard volumes on the grounds that the injector times are not defined properly. We will keep this under review. A slight and unimportant change may occur compared with previous Gundalf versions, so we have defined this as a calibration epoch.
To assist in navigating around the final reports and to prepare for a much faster model-filter cycle in preparation, some short-cuts have been added to the beginning of the screen version of the report. These allow direct jumps to the array summary and signature sections at present.
Advice added on how to identify the bubble in the presence of various kinds of filtering.
After several requests, we have added a spectral plot at a specified target frequency of all the amplitude spectra within a signature sweep. This plot is annotated with the minimum and maximum amplitude values at the target frequency along with their difference.
These have been added to the standard plan and side views and the separate diagram removed.
An annotation error meant that in some circumstances the SEL range table defaulted to spherical spreading whatever the spreading regime was set to. This has been fixed.
An inconsistency in session file management could sometimes prevent a work area sub-array file from being downloaded after a session restore. This has also been changed to improve workflow efficiency.
Added a new airgun source from Seismic Source Engineering LLP. These appear in the gun palette as "JA76".
In some circumstances, the two-gun dropout matrix overflowed the right margin for large numbers of guns. We are not entirely sure why but have reduced the size of the table to mitigate.
We fixed a bug whereby contours were not always annotated when requested.
Added an environmental modelling report option for Sparker and Boomer sources. This is not currently applicable to marine vibrators which apart from harmonics do not provoke any frequencies outside the sweep range.
Often clients want sweeps for a much restricted range of theta, phi. We have extended the options here so that this can be done more easily.
The algorithm was further modified and warnings based on it were downgraded to informational. This means that the modelling can still be carried out successfully but the error bounds may be affected. Infeasibly close guns will still raise an error and Gundalf will not attempt to model them as they are effectively caused by user input error.
The half msec filters MASS3_Geo-0p5msec-0Hz_DC.flt and MASS3_Hyd-0p5msec-0Hz_DC.flt have been added.
Added modelling summary. Updated file access for computation of Pearson cross-correlation for PHP 8.0.
This uses a different algorithm based on a more relaxed and wider set of calibration data than Gundalf legacy but this update is closer to the legacy version.
For some time, Gundalf has handled sub-array numbering automatically, but it still appeared as an option in the add gun and add sub-array options even though it was ignored. These have been removed to avoid confusion. Guns separated by more than 2m in the crossline (y-direction) are considered to be on a separate sub-array. We are now looking at user-specified gun grouping using a separate mechanism.
Various small documentation modifications and some minor changes in preparation for upgrading the pdf conversion module. These do not affect functionality here but will be used to enhance pdf reporting as we go through the year.
This release introduces an extended modelling engine enabling several commonly used Sparker and Boomer sources and also a generic marine vibrator to be modelled. After a trial period, this extended modelling engine will optionally be available in both the Designer and Optimiser modules. Modelling of airguns is unaffected by this in either the Designer or the Optimiser.
As with airgun modelling, the following functionality is available:
For more details of this new functionality, see the KnowledgeBase article "Alternative Sources" in the dropdown menu.
This version of Gundalf includes a new option in the drop-out parameters whereby the total array volume can be constrained never to increase. In normal circumstances, as like for like is swapped, this is never an issue but in the last release an option to allow volume matches within 25% of the drop-out was added. This means that a drop-out could potentially increase the total array volume. This new constraint operates on a gun by gun basis so that if left unchecked, only drop-outs of the same or lesser volume under option, can be used.
The notional sources are now available for download with the environmental report.
In Gundalf legacy, the centre of mass was calculated using the maximum and minimum x,y values in contrast to the weighted element approach of the centres of pressure and energy. In this release, the calculation of the centre of mass is normalised to be consistent with the calculation for the centres of pressure and energy. The centre of mass is now the mean position of the x,y,z components for all non-active source elements.
A couple of seal filters were reported as corrupted. They have now been reloaded from another archive and verified against that archive. These filters are named strictly using the supplied archive headers, however, with external filters we can only implement what we are given so if you have a particular favourite filter, we recommend you upload your own copy during processing using the filter upload option.
Fixed naming error on the half msec filter QM-003-200L_HydResp.flt. Now changed to QM-001p5-200L_HydResp.flt.
A patch release addressing two reported problems.
In this case, the bubble was annotated incorrectly because the time-axis is effectively shifted by the undeclared filter anticipation component. Furthermore, the override whereby the bubble start time can be specified did not work in this eventuality.
The logic for this has been re-written to cater for the shifted time axis. The override now functions correctly but in most cases, Gundalf should find the bubble period using the default bubble start time relative to the peak of 0.0s. Note that, when specified, the bubble start time is always relative to the peak.
In general bulk time shifts are not applied to airgun signatures - they should make no difference to the signature. However, this was handled incorrectly in the engine. Gundalf now removes any bulk time shifts before modelling the array, to avoid this eventuality.
Relative time shifts are handled as before and are unchanged by this.
The formatting of all tables in the generated pdf files was homogenized.
A minor release incorporating a number of user suggestions.
If your filters have anticipation components, Gundalf will now self-adjust to display any energy before time zero. This extends the original functionality in Gundalf legacy.
To mitigate the increasing number of filters in Gundalf and to improve client response, we have added upload buttons for both pre- and processing filters. These allow the user to upload their own filters. You will have to do this each time you wish to model as Gundalf is stateless and will not store your filters. Nor will they be accessible by anybody else. Please note that you are responsible for the content of your own filters. Gundalf will check them but cannot check everything. They are currently limited to a maximum of 4,000 samples.
The dialog for uploading arrays now automatically looks for arrays with the suffix .sba.
In view of the length of the filter selection list for external filters, a shortcut is available. Please check the updated Knowledgebase page on "Filtering".
A minor release incorporating a number of user suggestions although the modelling engine has been substantially upgraded internally.
The Gundalf modelling engine has been enhanced to model some sparkers and boomers in addition to all its traditional airgun types. These are not yet accessible but the internal architecture has been changed accordingly, however for normal airgun modelling, behaviour is unchanged in either modelling or in the user interface.
We have included an optional phase check into drop-outs. As a result, along with all the other drop-out checks built into Gundalf, you can now monitor drop-outs for a phase deviation of more than some specified number of degrees at the maximum phase frequency within the specified frequency range. This entailed updating the drop-out part of the engine as well as the quick dropout app.
This was introduced by request of more than one user. It would not be our choice to include this functionality since there is currently no peer-reviewed evidence of which we are aware to support any beneficial claims and drop-out analysis is already complex. If any users are aware of such evidence however, please let us know and we will gladly reference it.
The half msec filters Z700_Geo-0p5msec-0Hz_DC.flt Z700_Hyd-0p5msec-0Hz_DC.flt ZXPLR_Geo-0p5msec-0Hz_DC.flt ZXPLR_Hyd-0p5msec-0Hz_DC.flt have been added. The OBX filter OBX_min_2ms_1p0Hz_LC_combGeoNode.flt has also been added to complement the existing OBX_min_2ms_1p0Hz_LC_combHydNode.flt filter. The number of filters available now in Gundalf is quite large so we are looking closely at a number of suggestions to mitigate this.
The bubble annotation algorithm in the reports has been upgraded to match the internal one in the engine itself along with documentation upgrades.
Additional annotation has been added to amplitude spectral plots for frequency at which the peak occurs and the bandwidth of the 3dB and 6dB points.
The maximum time to display is now correctly applied in array comparisons.
On the amplitude spectrum, the dual y-axes are now colour-coded to match the data they represent.
Added option to specify the maximum frequency to display independently on normal amplitude spectra and on inline/crossline directivity spectra. Previously, the same value was used for both.
Filters Z700_Geo-0p5msec-0Hz_DC.flt Z700_Hyd-0p5msec-0Hz_DC.flt ZXPLR_Geo-0p5msec-0Hz_DC.flt ZXPLR_Hyd-0p5msec-0Hz_DC.flt have been added under half_msec.
A standard granularity of 0.01m has been applied across all coordinate entry, for sub-array positioning or for specification of observation point.
A "continue to physical parameters" button has also been added to the top of the geometry to avoid the need to scroll down to the bottom of a long array to continue to the next page (appears when the working array is not empty).
An additional field displaying the effective maximum bandwidth of the model has been added to the Array Summary to make it clear that some array parameters, notably those involving the peak are strongly dependent on the bandwidth in which the signature is displayed.
The filter OBX_min_2ms_1p0Hz_LC_combHydNode.flt has been added under two_msec.
This has been modified to restore the current user after restoration. Previously, the stored session user was used as author.
Some small documentation changes for clarification. For example, the appearance of a negative percentage change in P/B as meaning an increase in the P/B in a drop-out report is now spelled out more clearly.
A shift error was found for resampled shifted guns with long delays. Now fixed.
Small change to permitted cluster proximity for tight clusters of small guns.
The 32kHz and 16kHz sample rates in normal modelling have been removed because the environmental modelling algorithm uses a more accurate algorithm at high frequencies. The environmental modelling therefore now allows the download of the matching signature and amplitude spectrum.
None of the modelling changes have necessitated re-calibration so the calibration epoch remains unchanged.
Temporally dispersed arrays have been implemented. These were first demonstrated in principle in 1984 in Ziolkowski (1984), "The Delft Airgun Experiment", First Break, 2(6), p. 9-18 where a deliberately detuned airgun array was reconstructed from near-field measurements. This was revisited in the last few years by Ray Abma and Allan Ross of BP with much longer delays and proprietary coding techniques under the name of "popcorn" arrays, "Popcorn shooting: sparse inversion and the distribution of airgun array energy over time", SEG Technical Program Expanded Abstracts 2013. Such arrays reduce the peak amplitude by around 20dB with beneficial effects on marine life. They also have geophysical advantages.
Here we simply allow the randomisation of long delays (up to 10s currently) and compute the output signature in the normal way in the manner of Ziolkowsi (1984), although we use predicted near-field signatures from the Gundalf modelling engine rather than recordings as in the original experiment. Note also that such arrays typically have little or no pressure interaction energy.
This has been upgraded to annotate both the bubble peak and the bubble trough on the signature display for QC purposes. If these do not look correct (usually the result of some kind of filtering delay or bandwidth reduction), try again by specifying the bubble search start time manually.
Summary table of all options included for convenience.
Some annotation changes to improve appearance of tables and one or two typos corrected. We have also updated the description of the Physics to reflect all the recent work on the modelling engine itself.
Patch to fix an annotation error affecting the gun volume in the two-gun dropout table was identified and fixed. No calculations are affected.
The filters SBGS_M1p5_Geo_0p5_200_1p0.flt and SBGS_M1p5_Hyd_0p5_200_1p0.flt have been added under one_msec.
The GI gun is now connected and fully annotated in the reports with mode details, Generator/Injector split and delays.
The array plan view has been modified so as not to distort guns when there is a single inline sub-array or small number of guns.
On customer request, the amplitude spectra have been added to the downloadable archive and appear as the .obp file to supplement the existing sweep signature file .obs. The format is the same with the amplitude spectra separated by the same theta/phi header as the .obs signatures.
On customer request the closest separation of guns now follows Gundalf Legacy but with a computation based on the total amount of air in a cluster to determine the closest practical separation. This now handles the closest recommended separations for tiny Sleeve guns.
The comparison table now includes all the original parameters of the legacy version and the annotation has been cleaned up.
On customer request the signature can now be normalised back to 1m when the observation point is specified, so that the signature is in bar-m. This matches the infinite vertical far-field signature but allows it to be done for any position in the acoustic field. In particular it allows comparison with sweep signatures which are always normalised in this way. Note that a small approximation is implied if the observation point is relatively close to the source.
The two filters OBX_lin-0p5ms-1p0HzLC_combGeoNode.flt and OBX_lin-0p5ms-1p0HzLC_combHydNode.flt have been added under half_msec. Filters Z700_Geo-2msec-0Hz_DC.flt Z700_Hyd-2msec-0Hz_DC.flt ZXPLR_Geo-2msec-0Hz_DC.flt ZXPLR_Hyd-2msec-0Hz_DC.flt have had a large number of almost zero samples trimmed out.
We have added a -18dB line to the existing -3dB and -6dB lines on user request. On the amplitude comparison spectra, the annotation is colour-coded to match the maximum of the two comparison arrays.
Normally the use of spare guns is restricted to the same gun type, same volume and same sub-array. We have added options to relax each of these constraints individually to give more flexibility to the acceptable spare replacements. Even with these relaxations, spares must be within the specified range tolerance to be considered acceptable.
We have also added further documentation to the drop-out matrices to make it clear what information can and cannot be found out. By their nature, you can find out if a drop-out combination can be mitigated by a particular choice of spares, but you can't generally find out which spares are being used. For that you must consult the full tables or preferably, just use the quick drop-out report, which is what it was designed for.
The pop-ups on the guns in the quick drop-out report now have the x and y coordinates to accompany the existing z coordinate. This should emphasize the conventional right-handed nature of the Gundalf coordinate system with x increasing to the right away from the boat, y increasing from starboard to port seen from above and z increasing down.
Finally, spare guns in the 1- and 2-gun drop-out matrices and also the quick drop-out report are now greyed out in the same way as in the full tables.
We added an option to choose the contour annotation levels between 0 (no annotation), 2, 5 or 10dB increments.
We have included two new sample intervals to bring Gundalf Cloud into line with the legacy version. The default remains 0.0005s as the best compromise between accuracy and performance.
This has been substantially improved. In each case, if a drop-out passes without the need to deploy an available spare, this appears in the tables. If multiple spares are available for drop-out combinations, then if the original drop-out fails the drop-out specs, all available spares will be used to attempt to mitigate the dropout.
We have also added an option to number guns in sub-array.gun format under the General Parameters section. Internally, Gundalf numbers all guns in conventional increasing inline and starboard to port numbering without regard to sub-arrays as all guns are treated as interacting. This option is cosmetic only and if requested, applies everywhere except the array plan and side views.
Finally, we have made it clear in the array summary table the length of the window in energy terms over which the RMS was computed. RMS depends directly on the length of the window used and for transients is effectively useless as Fricke et. al realised decades ago, which is where the dB relative to 1 microPascal / Hz at 1m. unit came from. Unfortunately marine biologists use RMS so it has crept back in. Nearly everywhere in Gundalf, the full specified signature window is used, however in drop-outs to reduce useless computation, this window is shortened until it encompasses 99.95% of the energy of the full window. The only place this makes any significant difference is in the RMS computation.
We have added dB annotation of all contour plots. Contour lines appear every 5dB and are annotated every 10dB. to help you find your way around them.
We have rewritten the algorithm which displays the array in the reports. This greatly improves the annotation and scaling and adds information about gun types to supplement the existing information.
The four filters Z700_Geo-2msec-0Hz_DC.flt, Z700_Hyd-2msec-0Hz_DC.flt, ZXPLR_Geo-2msec-0Hz_DC.flt and ZXPLR_Hyd-2msec-0Hz_DC.flt have been added under two_msec.
A small documentation change was made to make it clear that array comparison can not only be used to compare two different arrays, but also by entering an array, modifying its depth, and then entering it again using array comparison, it can be used to compare the same array at different depths. We have also made it possible to enter an alternative name for the working array at the model checking stage. This will appear in the subsequent report.
We have speeded up drop-out computations by a modest amount by using that part of the signature which contributes 99.95% of the total energy. The parameter MXDROP may change by a small amount but other parameters are very robust to this. In essence, we are excluding the tail of the airgun signature after the significant bubble periods. It could be speeded up significantly more by assuming that sub-arrays do not interact as is a common practice. However, they do interact over a considerable distance (it is clearly visible at 30m separation) degrading the accuracy of drop-out computations and we have chosen not to compromise Gundalf's standard assumption that all guns in an array can interact.
We have added the following filters on request, seal20131209_lin-0p5msec-iir2p0Hz-alc2Hz.flt, seal20131209_lin-1msec-iir2p0Hz-alc2Hz.flt, seal20131209_lin-2msec-iir2p0Hz-alc2Hz.flt. We have also widened the scrolling list to make the filter choices clear.
We have added an exclusion zone table to the Southall data section of the environmental report to make it easy to look up minimum ranges of approach for specified commonly occurring fixed SPL and SEL dB levels.
We have annotated this a little more clearly so it is obvious even on a small device where port and starboard are and that the view is a plan view.
The third release of the C8.2 series features an experimental calibration of a large volume airgun. It has been calibrated from published signature parameters of similar sources and should be considered only for exploration of the capabilities of low-frequency extension to the spectrum at this stage for sources relying on the discharge of high-pressure compressed air. In this release volumes up to 20,000 cuin. are catered for.
We have chosen a different method of generating pdf reports specifically for drop-out analyses. For a reasonable sized array, a 3-gun dropout can easily run into 50-100 pages. Quite apart from this being a rather unreadable way of displaying such information, it caused problems for our existing pdf generation and we have implemented a different method which is much faster and does not require so many resources. Early experience is good. For all other reports, we use the original method which gives us much finer control over the pdf generation process.
This was available in Gundalf Windows/Linux but is now available in Gundalf Cloud as well.
Wave Shape Ratio We have now connected the logic for Wave Shape Ratio specification and modelling (as was always present in Gundalf Windows/Linux).
Array validity checking has been improved. It was possible to read in a corrupted sub-array file, and the model would stop while trying to model it. An error was reported but only in the logs. We have now implemented the checking in a way which should stop this happening and make it obvious to the user what is invalid at the geometry entry stage.
Some wording changes in reports for clarification, for example the energy flux in the amplitude spectrum is now explained.
The Drop-out Quick Reference file could be incorrect for 1-gun dropouts when spares were present. This has been corrected.
The second release of the C8.2 series features a significant makeover of drop-out modelling, rationalisation of the attached headers for signature, amplitude spectrum, notional source and signature sweep files, rationalisation of the download names for reports, zip, csv and quick-reference html files, and a patch.
We had not revisited this for some years and it was long overdue some spring cleaning. We have done the following after discussions with several customers and thank them for their valuable feedback and insights.
The .sg1 signature text file has had an extended header with lots of useful information for some time. A customer requested we extend this to some of the other output files which makes really good sense so we have added to the amplitude spectrum text file, the notional source text file and the new theta/phi signature sweep file. Thanks again for this important feedback.
The downloadable outputs pdf reports, signature and sweep zip files, the drop-out quick-reference and the drop-out .csv file all now have meaningful suggested download names which embed the volume and depth of the array along with an identifying prefix and a process number which keeps them unique within a session. They originally just had the process number and it was easier to get confused in a busy session.
This is only the second time we have had to issue a patch release in 15 years but its very simple with the new Cloud architecture, so we decided to go ahead. It was pointed out that the rectangular / polar conversion was not correctly passed to the engine by the interface, so the Cloud version did not agree with the Windows version. This is fixed and they now agree. We apologise for any inconvenience.
The first release of the C8.2 series brings a brand new service to Optimiser, new filters, extended drop-out criteria, and some other minor updates.
Gundalf Cloud is a complete reworking of the Gundalf desktop software into a global architecture of computationally-optimised servers with a modern, platform-independent browser-based interface. No more licence keys or software to manage.