A Package of Script Codes, POSIBIOM for Vegetation Acoustics; POSIdonia BIOMass

1. Introduction

Aquatic macrophyt (algae, grass etc) is one of the  important components playing a crucial role in various ways for the aquatic  ecosystem [1–3]. Considering the volume and  surface area of the marine system, seaweeds and seagrass have significant  contribution primarily with their oxygen production and carbon  absorbance/emission into a way of the ocean-atmosphere interaction. Besides  microscopic algae and photosynthetic bacteria, most of marine macrophytes  reveal status of the marine ecosystem health and ecological  level. Some of them are indicators of pollution whereas some are indicators of  the pristine environment [4,5]. There are a lot of further importance and contributions  of them to the marine system. Unlike the algae, the seagrasses are associated  with the sediments (benthic system) by their below-ground parts (rhizomes, and  roots) [6] and pelagic system by their above-ground parts (sheaths,  and leaves) [5]. 

One of the common seagrasses, Posidonia  species, the phanerogam species which have been sustained well in the  poor-nutrient marine environment and formed colonial life are coastal species,  and are worldwide distributed in the subtropical and temperate waters from  Indo-Pacific Ocean (Australia) to Atlantic Ocean (Mediterranean Sea) systems [7,8]. They are very sensitive to the pollution,  climate change, global warming, biogeochemical cycle, anthropogenic effects and  water movements etc [9–11]. Their biometrical  traits respond rapidly to such global or local events in order for the  scientists to monitor health of the marine system. Like other seagrasses, Posidonia  species are near coastal inhabitants giving many useful opportunities to the  other organisms under their canopies [4,12–15].  One of the common Posidonia species is Posidonia oceanica (Linnaeus)  Delile, 1813 which is an endemic species for the Mediterranean Sea extending to  near coast of the eastern Atlantic Ocean [16].  Most (25-30%) of the Mediterranean coasts are inhabited by P. oceanica [6].

Like other seagrass (angiosperms), Posidonia  oceanica (the Mediterranean meadow) is a prominent component in the zonal  classification in the Mediterranean Sea, and classifies the border between the  Infralittoral and the Circalittoral of the benthic environment [1,3,5,17,18]. Therefore, it is positioned as a  coastal engineer and interior architect in the coastal environment with their  following functions [19,20]. Besides the  ecological indication, Posidonia oceanica also play crucial roles in the  sediment stability and biogeochemical cycle, oxygen, carbon  emission/absorbance, blue carbon, habitat/shelter and food niches for its  interacted organisms, spawning and nursery grounds, competition with invasive  macrophytes, protectors of prey from predator, and extension of their habitants  in range [14,21–30]. Monitoring their dynamics  is a useful tool to sustain the species under the prevailing environment, and  to observe indicative acute or chronic changes derived from the allochthonous  and autochthonous sources in the marine ecosystem. Consequently, this knowledge  is used by the ecosystem modelers, ecologists, environmentalist, biologists,  coastal engineers and managers, and species managers and protectors [31–33].

Compared to fish and zooplankton acoustics,  vegetations acoustics have not however attracted significant attention of  researchers. The destructive sampling methods (SCUBA sampling to pick up the  seagrass) are not generally desired to sample the meadow under the protection [34–36]. Therefore, many studies were fronted to  apply non-destructive methods, remote sensing systems to study sensitive and  vulnerable seaweeds and seagrass [37]. There  are a variety of remote sensing techniques developed as follows: satellites  (especially Sentinel 2), video cameras and acoustics [38–44].  There are some limitations to sample the seagrass significantly for each  technique. Visual techniques (satellite, and video-camera) are limited with  requirements of the suitable atmospheric and sea conditions [45–47]. These techniques were mostly used for the  estimates of coverage and mapping of the different habitats e.g., [48,49]. Compared to the techniques, the acoustical  method is faster, more precise, and easier and does not need clear sea  conditions to sample the data [2,50], and but  have serious problems changing from correct bottom detection, removal of  spurious scatterers to sea-truthing of the targets [51].  Recently, in/ex situ acoustical studies have increased to relate  the biometrics of the seaweeds and seagrasses to the Elementary Distance  Sampling Unit (EDSU) [52], for instance, Depew  et al. [53] focused on Cladophora sp., Monpert  et al. [54] on P. oceanica, and Zostera  marina, Llorens-Escrich et al. [55] on P.  oceanica, Shao et al. [56] on Saccharina  japonica, and Minami et al. [57] on Sargassum  horneri. There were some studies to calibrate the acoustical data versus  the biometrics of two seagrasses, P. oceanica and Cymodocea nodosa  [58–61] and to identify the seagrass, P.  oceanica and C. nodosa [62].

Like that in the other remote sensing techniques,  acoustic data alone are inherently ambiguous with regard needed to the  identities of the target species. The mapping of macrophytes can be  accomplished with the software developed for the submerged vegetation  (EcoSAV:Echo Submerged Aquatic Vegetation, Visual Habitat and Visual Aquatic)  by BioSonics Company. Further methods of vegetation acoustics were developed  based on algorithms to classify bottom types [63–66].  They are OTCView, Echoplus, BioSonics VBT (Visual Bottom Typer), and EchoView [67–70]. While the acoustic reflectivity of aquatic  vegetation having an acoustical impedance of 1.026 to seawater [71] is known and used for the detection of  submerged aquatic vegetation (SAV) [72], the  basis of this phenomenon has not yet been figured out to solve the problems.  The other detection factor is the acoustical frequency used during the  vegetation acoustical studies. Although the most effective frequency for these  techniques is 420 kHz, it has been concluded that 200 kHz is also effective for  vegetation acoustics [73].

The present study is aimed to solve the problems,  detect and discriminate the Mediterranean meadow, and estimate the leaf biomass  for the future works and studies needing the biomass purposed by the researches  for sustaining the meadow. The presence of life on the sea bed causes  significant problems during acoustic post-processing, including spurious  targets, background noise, artificially generated-noise, reverberations, dead  zones, interference, ‘lost’ bottoms (misestimated bottom, too deep, out of  range of sound penetration by the frequency limitation depending on the source  level). In particular, the problem of ‘lost’ bottoms in stock assessments of  the species creates challenges when attempting to recover bottom echoes. In  some cases, commercial software and computer scripts require interventions to  manually recover ‘lost’ bottoms, which could be time-consuming. This study is  the first attempt to regard the matters of the acoustical data post-processing  aimed to estimate biometrics of P. oceanica, and presents an auto-self  (no often intervention) and easy used-package of the scripts codes to solve the  problems during the estimation with advantages and disadvantages of the codes  in comparison to the other software.

2. Material and Methods

Overall, acoustical raw data (*.dt4) used for the  present study were collected with a Biosonics DT-X  digital scientific echosounder which had a split beam transducer in a shape  of circle beaming a width (θ) of 6.8°,  operating 206 kHz using a software package of “Visual Acquisition” (v.  6.3.1.10980, BioSonics inc.). The pulse width (pulse length, pulse duration, τ) was set to 0.1 ms (highly recommended for  the seagrass studies), a ship speed to approximately no more than 5 knots, and  no deeper bottom than 50-60 m. The transducer was  mounted at a draft depth of 2.5 m down-looking on the starboard of the R/V Akdeniz Su [74,75]). During the  surveys of different sampling locations and months, different bottom types  covered by the seagrass were encountered namely, rock, sand, mud and matte [5,18].

Three main script codes which run autonomously after all settings completed by the users were written in MatLab (vers. 2021a, Matworks inc.) to correct the bottom depth, to remove spurious weak and strong scatterers else than Posidonica oceanica and to sort out the species regarding to its vertical rhizome and sheath. To recover removal of the debugs in the package, around 4000 acoustical files (300-9000 pings, 4500 pings per file on average) derived from two different projects [74,75] were subjected to the codes. Three main codes subsequently run as follows: “AbdezdeR1”, “AbemsiR1” and “SheathFinder1” to estimate ultimate data of the biomass (see Mutlu and Balaban [58] for the description of abbreviations of the algorithm names). There are several auxiliary scripts and acoustical data associated with the three main codes. All three codes run under a script as a main menu and manager to reach them. It is called in an acronomy of ‘POSIBIOM’ as abbreviation of POSIdonia BIOMass. The present study was inspired from an algorithm, ‘SheathFinder’ [58]. However, the algorithm was not a package, easy-used, and was debugged to estimate real bottom echo without infinitive permutation until all ‘lost’ bottoms completed disregarding seasonal biomass estimates of the seagrass. Therefore, all algorithms were highly revised and modified completely for the present study. In the present, the acoustical data collected only with only the BioSonics inc. echosounders can be used for the POSIBIOM since I am currently a user of BioSonics echosounder. To run the scripts, the data (*.dt4) was converted to a spread sheet of Comma-Separated Value (CSV) using a post-processing commercial program, “Visual Analyzers” (v. 4.1.2.42, Biosonics inc.). During the post-processing, the settings were configured to get output data in a vertical resolution of count-to-count (equivalent of ~one-eighth of the pulse width in cm depending on seasonal sound speed) and in a horizontal resolution of ping-to-ping. The processes were repeated to get the layered data including the seagrass from the uneven (not flat) bottom up to a stratum number of 249 (this is the last number allowed by the Visual Analyzer). This could be done once for the flat bottom to get count-to-count data of the vertical resolution. This is important to determine precise and accuracy of the vertical resolution of leaf length, hence biomass. This process is but time-consuming. It would be helpful and faster for processing the data if the scripts (POSIBIOM) read the collected data file of the BioSonics directly. In order for POSIBIOM to be professional software, the package would be coded to a programming language of the ‘C’ from the MatLab by its relevant tool.

The POSIBIOM is functioned as a manager to load input files (acoustical and coast line files, Marine Region [76]), to start each of three analyses, and to map the results of the biomass and leaf length of the seagrass (Appendix 1).

2.1. Algorithm of ‘lost’ bottom (AbdezdeR1)

The algorithm was scripted in a file, namely, AbdezdeR1 (see Mutlu and Balaban [58] for description of the abbreviation). Fixation of lost bottom is based on a root of adjacent bottom angles and depths. Configuration for recovery of real bottom is highly flexible during the processing. The fixed lost bottom is corrected with the following three successive methods, namely; the “chirping”, “running average” and “filling gap”. All methods were originated for the present study. The algorithm then estimates the dead zone using three methods. One method was created during the present study, and the other two methods have been described by EchoView [77] and Mello and Rose [78]. The algorithm has some options to alert users automatically in some cases to correct the bottom before the analyses (Appendix 2), and for last check of the bottom correction after the analyses (see Table 1, Figure 7). The algorithm is also useful for other purposes (e.g., fisheries acoustics, and partly plankton acoustics) else rather than the vegetation acoustics.

2.1.1. Real bottom recovery

Fixing misestimated bottom regards the sudden change of the bottom angle in successive pings. The angle to search the fake (second echo of the bottom, sheer of returned echo with the next pulse in the water column) bottom was followed within a certain bottom window of the successive real bottom depth. There are some flexible options to regard the depth deviations depending on steepness of the real bottom. Besides, surface, volume and bottom reverberations were considered for the process to figure out the real bottom by the algorithm. Furthermore, an additional prompt of an autonomous option appears (Appendix 2) when necessary in a case of existence of misestimated bottom at the first pings. Additionally, another subcall is functioned to track the real bottom between the last ping of the previous input file and the first ping of the next file. During the process, the script determines first the overall lost bottom in the next-to-next form of a block or single ping range. Therefore, the algorithm searches lost bottom in a narrow range for the wide block ranged-lost bottom, and then increases slowly the range to search the next lost bottom in a manner like the “spider net webbing”.

Of the three correction methods aforementioned;

• The chirping of the Sv of the bottom echo has a priority to the next two methods to correct the bottom. This method runs well in cases with clear bottom echo in the misestimated bottom section of the data.
• The running average is automatically called by the processor if there is no clear (weak) bottom echo. This method is functional by correcting the bottom lost between the previous and next corrected bottom depths and angles.
• The filling gap is then intervened to correct the bottom. In some case of occurrence of no bottom echo (a gap) due to the occurrence of surface or volume reverberation or intense fish school, the bottom echo disappears. Before filling the gap, the method checks the depths and angles.

All three methods run automatically regarding to their specific cases. In some case, for instance of bottom depth close to the near-field of the transducer, the correction could not be performed well. This action is not so rapid depending on many factors unexpected in the acoustical data by ranging pings of the lost bottoms in unicycle or multicycle. Therefore, some manual and automatic auxiliary options were added to the methods to control the range and cycle. The more the Visual Analyzer fixes the correct bottoms, the less solution of the methods takes time.

2.1.2. Dead zone estimates

After completing the correction for the entire data file, three different equations were used to estimate the total dead zone (TDZ) consisting of dead zones of the flat and angled bottom, regardless of the side lobe effects.

The first equation (Equations 1-2) was established with the present study, and was based on vertical range (Vd, from transducer to the bottom) and horizontal distance (Hd) estimated from the geographical successive distances or pings (Rp, Rc and Rn, choosing the shallowest distance, Rp or Rn referring to Rc) using the echosounder parameters (Figure 1). However, the GPS reports same geographical coordinates in case of echosounder pinging more than 1 per seconds. In this case, the β could be 90 degrees (Figure 1) so that the estimated angle could be set to zero degree. This problem was considered in another dead zone solution by equation 4 [77]. Therefore, the dead zone estimated by equations 1-2 alters to horizontal distance based on the ping instead of the geographical distance. This equation is obligatory in the algorithm to estimate the dead zone.

Total Dead Zone, TDZ= Dead zone of flat bottom (DZ) in m + Dead zone of angled bottom (DZ1) in m

$$\mathrm{DZ}=\frac{\tau /1000\ast \mathrm{c}}{2}$$

(1)

$$\mathrm{DZ}1=\tan \left(\beta \right)\ast \mathrm{H}\mathrm{d}$$

(2)

where τ is pulse width (msec), c is sound speed (m/s), $\mathsf{\beta }=\mathrm{acot}\left(\frac{\mathrm{H}\mathrm{d}}{\mathrm{V}\mathrm{d}}\right)$ in degree, Hd is the horizontal distance based on either geographical distance (m) or ping, Vd is the vertical distance (m) based on the range, R (not real bottom).

The second method and equation (equation 3) for estimating the dead zone was developed by Mello and Rose [78]. This equation is optional set to estimate the dead zone.

$$\mathrm{TDZ}=\left(2404\ast \frac{\mathrm{d}\ast {\mathrm{t}\mathrm{a}\mathrm{n}}^4\mathsf{\theta }}{{\mathsf{\theta }}^2}\right)+\mathrm{c}\ast \tau /4$$

(3)

where d is the bottom depth (m), θ=α/2 in degree (α is acoustical beam angle), τ is pulse width (msec), c is sound speed (m/s),

The third equation (equation 4) was used and offered by EchoView [77]. This equation is optional to estimate the dead zone.

$$\mathrm{TDZ}=\left(\frac{\mathrm{d}}{{\mathrm{s}\mathrm{i}\mathrm{n}(90-\mathrm{q})}^{}}-\mathrm{d}\right)+\mathrm{c}\ast \tau /2$$

(4)

where d is the shortest range (indeed TSD) to the bottom in the acoustical beam, hereby the shortest range (Rn) between successive pings (m), q= the slope of bottom in degree, τ is pulse width (s), c is sound speed (m/s),

All estimates could be similar particularly when using the equations 1, 2 and 3.

2.2. Algorithm of Noise and Reverberation (AbemsiR1)

Overall, determinations and removals of the noise, interference and reverberation were performed following the procedure for signal-to-noise ratio (SNR) described by De Robertis and Higginbottom [79].

The SNR estimates were based on three different algorithms in the present study modifying the procedure slightly in a script, AbemsiR1 (see Mutlu and Balaban [58] for description of the abbreviation). Two different measurements were pre-required for the present study. These measurements would be fundamental reference data in a priority to the SNR solution. The background natural noise including electrical and physical ambient noises changes in time and space. Therefore, at least one measurement is recommended for each of two noise types during an acoustical survey. During the survey with a research vessel, some other acoustical devices are used for the measurements of other purposes such as water current, bottom depth sounding, fishing net sounding etc. Such instruments can interferer the echoes of the study echosounder, and are classified as artificial noise-generators. Furthermore, the interferences occurred owing to biological sound sources and sheering of echo returned from the bottom with echo (pulse) of the next pinging. Both noise measurements can be performed using listening mode of the echosounder and used as a hydrophone. During the present study, one noise measurement was conducted switching off the depth reader, fish finder (echosounder) or similar acoustical devices (e.g., ADCP) of the research vessel (ship) (Figure 2a) and the next measurement switching on the devices (Figure 2b).

Three methods for removal of background noise (b), interference (i) and variety (surface, volume and bottom) of reverberations (r) were cased conditionally with some additive functioning ranges (see Table 2, Figure 9) between measured (MSNR) [79] and expected (ESNR) data (e.g., in Figure 2) as follows:

if MSNRb > ESNRb, removal of the data,

if not, set ESNRb into the acoustical data for the corresponding depth, instead of -999 as suggested by De Robertis and Higginbottom [79],

if MSNRir > ESNRir, removal of the data,

if not, nothing done,

if MSNRir/MSNRb > a criterion value automatically estimated between the ratios for each acoustical data file, removal of the data,

if not, nothing changed (this option is useful tool for detection of air-bubble generated by SCUBA divers).

The expected SNR (ESNR) versus the water column depth was estimated using a logarithmic regression between acoustical data (Sv) without Time-Varied-Gain (TVG, one-way transmission loss, TL=20logR + 2Rα, where R is the range in m, and α is absorption coefficient in dB/m) and depth for both background noise (Figure 3a) and interference (Figure 3b).

2.3. Algorithm of “SheathFinder” & Leaf length and biomass (Sheathfinder1, and 2)

This script consisted of sections of simultaneous removal of weak and strong scatterers, and simultaneous calibration and fixation of the leaf and sheaths or vertical rhizome of Posidonia oceanica (Figure 4). All processes were visually performed looking at the calibration file. Visual calibration file is pre-required before switching to the next real process to estimate the biometrics. The calibration file possesses a section having the meadow (Figure 4d,e). In some temporal cases, the vertical rhizome and sheath could be too short for the detection or within the deadzone [5]. In these months, the filtration continues until leaf parts near the sheaths disappear on the enhanced echogram.

After fixation of the sheaths, the algorithm seeks the count-to-count echo upward to determine the leaves until there are a few gaps between the successive counts. The rule for the determination is that volume backscattering strength has to be in a decreasing trend from the bottom to top of the leaves.

After completing the seeking above the dead zone, this height reads the leaf length in a precision of one-eighth of the pulse width depending on number of strata entered in “Visual Analyzer”. Thereafter strong scatterers suchlike fish are removed from each leaf, and the EDSUs (Sv: volume backscattering strength in dB/m³ and Sa: area backscattering strength in dB/m²) are calculated following the methods as suggested by Simmonds and MacLennan [52]. Besides, the algorithm discards echo without sheath, and categorizes that as a false leaf (Figure 4e).

After finishing processing the acoustical data of a file, leaf biomass is assessed using different EDSUs in relations to the biometrics of the meadow in time [60]. Of the six different temporal relationships [60], the algorithm choices right month matched with month of the acoustical data over yearday. It shall choice the closest month if the yearday of the data is outside the six months.

Methodologically, each script needs an input file or data produced as an output file in MatLab format (*.mat) by the previous script throughout AbdezdeR1 to Sheathfinder2. Initially, AbdezdeR1 only needs output files in CSV format, processed by “Visual Analyzer” since the package of POSIBIOM cannot directly read file.dt4 in the current status.

The package is installed according to window size based on the pixels of the monitor or display owned by the users.

3. Results

Besides the data (acoustical data in *.CSV to *.mat) and auxiliary (baseline or coast map in format of *.bln to *.mat, reference noises in *.CSV, sheath calibration in *.mat, and coast blanking in *.bln to *.mat) files opening/loading, a main menu for controlling all processes is configured in a window called POSIBIOM (Figure 5).

3.1. Flowchart of POSIBIOM

The diagram shows the flowchart in order to finalize all analyses (Figure 5). There is also an option called “Tools” to plot results for leaf length and biomass distribution in format contour or line on the map.

To load the data for acoustical or other type data, opening in “File” is associated with that of Microsoft Window, embedded in the MatLab. For each analysis, opening files is specifically called from the menu (Figure 5). Files for analysis of “Lost Bottom” are opened in format of “*. CSV”, output of Visual Analyzer. The next analyses use format of “*.mat” as output of the previous successive analysis.

The output files are saved in a folder “AbdezdeOut”, created by the package, under the folder where the user opens the data for “Lost Bottom” analysis. The next analyses save the outputs into a folder “AbemsiOut”, and “PosiBiom” for “Noise & Reverberation”, and “SheathFinder & Leaf and Biomass” algorithms, respectively in the way aforementioned. In other word, the previous output files become input files for the next analysis. Output file of the last analysis (Figure 5) is in format of “*.xls*” for the users to use the data for their different purposes.

3.2. Lost Bottom and Dead Zone

This algorithm is purposed to correct the lost bottom, and then to estimate dead zone depth using three different equations (Equations, 1-4). After opening single or multiple input data files the users want to process (Figure 5), the algorithm needs also to open a coastline file from ‘Base Line for Map” before starting the analysis clicking “Lost Bottom” in “Analyses” in POSIBIOM menu (Figure 6).

The algorithm starts to read each of the files sorted in an alphabetic order of A-to-Z after starting the analysis (Figure 7), and then a window appears to control the analysis (Appendix 3) and get information of the acoustical data and echosounder configuration during the sampling (Figure 7, Appendix 4). After all setting (transducer depth, calibration offset, types of dead zone estimates, save etc., Table 1) is performed in the window, the analysis proceeds tapping button START (Figure 7). Description of each option in Figure 7 is presented in Table 1 with their functions, case to use, recommendation and possible negatives.

Figure 7. Graphical menu for control, configuration and info menu of “Lost Bottom” (see Table 1 for actions of each coded option and function with number).

Figure 7. Graphical menu for control, configuration and info menu of “Lost Bottom” (see Table 1 for actions of each coded option and function with number).

During the process with the “Lost Bottom”, enhanced echogram first appears in pings versus depth for unprocessed acoustical data with marks of the estimated lost bottoms in green lineation (Figure 8a). These lineation occurred based on the algorithm of bottom angle and bottom depth in the adjacent real bottoms (Figure 8b, the figure appearance is optional because of some cases, no 23 in Table 1). As the analysis progresses, bottom depth correction starts from the beginning to the end of the pings (Figure 8c, the figure appearance is optional because of some cases, no 23 in Table 1). If there is any spurious scatterer through the water column in front of the bottom, the processing cleans them in order to reach the bottom. Nevertheless, original acoustical data are saved into output file. After finishing the analysis for the lost bottom, the algorithm estimates the dead zone depths using the equations user selects, and then final enhanced echogram comes out in the screen. If user checks option “save P”, the final echogram is saved as format of *.jpeg (*, input filename). If user checks option “save F”, the final results with the variables necessary to the next analysis is saved as *.mat (*, input filename) to folder “AbdezdeOut”. Finally, the corrected bottom depth is plotted in line on a map after correction rate info appears on a 3D-plot to be checked by user to see uncorrected and corrected bottom depths (Figure 8e, e’, respectively).

From this example in Figure 8, the result showed a significant of correction rate on comparison of plot in Figure 8e with that in Figure 8e’. Overall, goodness of the correction was estimated in a percentage of more than 95 after all files (approximately 4000 files belonging to different seasons and years) were analyzed by the “Lost Bottom”. The plots in Figure 8d, e are reset every 100 files processed, and the analysis then needs to press START to continue the next files.

3.3. Noise, Reverberation and Interference

This algorithm is purposed to remove background natural noise, reverberation and interference (artificial noises) using three different conditional functions based on the SNR (Figure 9). After opening single or multiple input data files the users want to process (Figure 5), the algorithm also needs to open a coast file from ‘Base Line for Map” in the main menu, and needs two different noise files as described in Figure 2, Figure 3 and Figure 5 before the analysis. Starting the analysis is performed clicking “Noise” in the “Analyses” in the POSIBIOM menu (Figure 6). This algorithm first runs a subcall in association with “Noiseback1” to estimate the expected SNR (ESNR) for background noises and interferences (Figure 3).

The algorithm “Noise & Reverberation” is achieved depending on the configuration the users perform to configure the options flexibly; particularly for the reverberation intensified at the surface and water column (volume reverberation) (Figure 10a). The surface reverberation is sourced by the wavy sea, strong winds rocking/rolling the ship, and speed of its propel at stop.

The interference had different patterns in relation to the bottom depth, and pulse width, ping rate and frequency of the ship and researcher echosounder (Appendix 5).

The menu needs to enter an approximate maximum canopy height of the seagrass and the transducer draft depth installed during the survey, and if no entry, the default values appear (Figure 9, Table 2).

Figure 9. Graphical menu for control, configuration and info menu of “Noise & Reverberation” (see Table 2 for actions of each coded option and function with number).

Figure 9. Graphical menu for control, configuration and info menu of “Noise & Reverberation” (see Table 2 for actions of each coded option and function with number).

After finishing to run “Noiseback1”, the program calls a subprogram“Sheatfinder2” to start the process for the removal of the spurious echoes such as the noises, and reverberations. The subprogram demonstrates first output data, processed by the “Lost Bottom”, in an enhanced echogram (Figure 10a), and bottom depth and average dead zone line estimated by the methods the user checks in (option numbers of 26-28 in Figure 9). If any method is not checked by the user during the processing by “Lost Bottom”, the method option is disabled automatically in the analysis of “Noise & Reverberation” (Figure 9). The algorithm starts to fix the noises and reverberations for the removal after the trail of three conditions (Figure 10b, c) between the measured and expected SNR. Thereafter, the output file with the variables (Figure 10d) necessary to the next analyis, “SheathFinder, Leaf & Biomass” will be saved as format of “*.mat” in to a folder, “AbemsiOut” under folder, “AbdezdeOut” if the user wants to save the outputs (Figure 9, function no 31).

Figure 10. Graphical outputs of the “Noise & Reverberation” analysis. a) Enhanced echogram before the start of the analysis, b) an algorithm to figure out detection of the overall noises, spurious echoes (optional), c) the SNR for each background natural and artificial noise (optional), d) removed noises and reverberations, in this example the configuration needs optimum adjustments furthermore to remove the entire reverberations.

Figure 10. Graphical outputs of the “Noise & Reverberation” analysis. a) Enhanced echogram before the start of the analysis, b) an algorithm to figure out detection of the overall noises, spurious echoes (optional), c) the SNR for each background natural and artificial noise (optional), d) removed noises and reverberations, in this example the configuration needs optimum adjustments furthermore to remove the entire reverberations.

The noise, especially interferences had different patterns extending down to the bottom in time and space (Appendix 5). This situation creates a lot of negative and spurious sources suchlike fake acoustical characteristic similar that of the sheaths and vertical rhizome. Besides, there is different reverberation rather than that generated by the ship propel; for instance, that generated by SCUBA divers as studying a purpose within beam of the transducer (Appendix 6). During the analysis both unrecovered and unfixed noises and reverberations are removed by the next analysis, “SheathFinder, leaf & Biomass”, and echoes through the canopy height are kept to the next analysis.

3.4. Leaf and Biomass Estimation

Recognition of different traits of P. oceanica in the algorithm was performed basing the descriptions on the experimental results obtained from Mutlu and Olguner [60]. After cleaning out of the spurious echoes from the acoustical data, the algorithm gives an opportunity for visual calibration to fix the seagrass (Figure 4). Therefore, this algorithm gives a favor for the users which are not much familiar with the acoustical theory. For the calibration, three sliders which guide the users with each of color scale specific to the each slider are available to remove weak, strong scatterers, and to fix the leaf and sheath or vertical rhizome, respectively (Table 3, Figs. 4, 11). For that, user needs only to load a file containing the seagrass from the main menu, POSIBIOM (Figure 5). Before starting the analysis, both acoustical and calibration data files are required to open (Figure 5). At the same time, the coast file is needed to plot results on map.

Compared to the other algorithms (Figure 7 and Figure 9), the present algorithm is the easiest and simplest graphical demonstration during the usage (Figure 11). There are some options to save or plot the estimated results (leaf biomass and length) based on the different EDSUs. Overall, the biomass estimate based on Sv is plotted on a map on the colored trackline (Appendix 7c, d). For the contour or line plots, a further option called ‘Tools’ was embedded in the main menu with some more graphical properties.

During the analysis, the biomass is estimated using a combination of Sv and Sa with eight equations formulated by Mutlu and Olguner [60]. When tapping “Result Table”, the results flow down ping-to-ping, but this option slows down the analysis. The algorithm saves the results comprising of three different estimate of the biomass, including other informative variables (file name, geographical coordinate, yearday informing time in format of day,hrs+min+s of a day, bottom depth, dead zone depth, canopy height (leaf height), the biomasses (based on Sv, Sa in “Cut” experiment, Sa in “Leaf” experiment [60], and month range of which the equation is used to estimate biometrics (Figure 12).

During the analysis, the algorithm starts to scan structure suchlike the sheath or vertical rhizome of the seagrass in ping-to-ping depending on the vertical resolution analyzed with “Visual Analyzer” (Figure 12a). The green scanning vertical line proceeds slower when determining the sheaths than that during absence of sheaths in the ping. There are two theories applied to remove the fake seagrass; one is presence/absence of the sheaths, and another is structure suchlike the sheaths whether to have real leaf or not as described in Material and Methods.

After completing to scan the sheaths and leaves, strong scatterers like fish individuals among the leaves are removed according range of the expected threshold of the Posidonia [60]. The next step is the biomass estimates of the leaves which are regressed with seasonal equations according the current sampling month [60], followed by saving the output data in a format of “*.xls” without variable names into an automatically created folder “PosiBiom” The variable names are in the order of each column as follows; geographical coordinates (latitude, longitude), yearday, bottom depth, canopy tip depth, leaf height (canopy height since orientation of leaves standing on the bottom; right, semi-flat and flat position, [62]), and three different successive biomasses (g/m²) estimated by Sv and Sa (“cut experiment”), and leaf calibration (“leaf experiment”) [60]. There is an auxiliary script, namely ”PosiDrwTool” in “Tools” of the main menu, POSIBIOM to draw distribution of the estimates in a format of trackline or contour (Appendix 7 and Appendix 8).

4. Discussion

The present study has discussed the advantages and disadvantages of the algorithm, POSIBIOM with the other acoustical commercial software to estimate bottom type and vegetation since the POSIBIOM is a production of the first study for Posidonia oceanica and its biometrics. Some scripts of the algorithm were related to solution of the general acoustical problems (e.g., lost bottom correction and noise and reverberation removal) during the post-processing of the acoustical data purposed for the other studies, such as fisheries acoustics. For such problems, there are commercial and professional software. In their some cases, a manual intervention is needed to correct the bottoms even if they have many functions for this matter. The mapping of macrophytes can be accomplished with the software developed for the submerged vegetation (Echo Submerged Aquatic Vegetation; EcoSAV, Visual Habitat and Visual Aquatic) by BioSonics Company. Further methods of vegetation acoustics were developed based on algorithms to classify bottom types [63,64,65,66]. They are OTCView, Echoplus, BioSonics VBT (Visual Bottom Typer), and EchoView [67,68,69,70]. However, previous studies on the vegetation acoustics have remained at the assessment of percent coverage and canopy height of the seagrass (e.g., van Rein et al. [50]; Lee and Lin [80]) depending on the limitation of echosounder parameters [81]. Unlike the POSIBIOM, there is overall a maximum bottom depth restriction with a depth value entered in the software to remove suchlike the seagrass or macrophytes at the greater depth. There were two attempts to calibrate the acoustical data to estimate the biometrics of the seagrass by developing the nonprofessional algorithm; one was based on data of VBT process [59], and the next one was based on “SheathFinder” to fix the sheaths/vertical rhizome [58]. The present study modified and revised the nonprofessional algorithms completely and significantly to be used professionally for the seagrass studies.

Regarding to the package POSIBIOM, one issue is the avoidance from the usage of the destructive methods to study the meadows under the protection [34,35,36]. Besides, the previous remote sensing techniques produced only coverage and mapping of the different habitats (e.g., Fakiris et al. [48]; Dimas et al. [49]). Compared to the techniques, the acoustical method is faster, more precise, and easier to ground-truth the data [2,50], and develop the algorithms to remove spurious scatterers [51]. Another issue is detection of the seagrass by the acoustical frequency. The meadow is detected well by a frequency of 206 (200) kHz to assess biomass variables even if the higher frequency (420 kHz, Biosonics inc.) is recommended to study the seagrass and vegetation acoustics. Quintino et al. [82] recommended 50 kHz to study soft bottom classification, but 200 kHz is good effective for the classification of the bottom types including seagrass.

Vegetation acoustics is not however easier at practice to process its data than the fish acoustics. The fisheries acoustics had many problems overcome during the processing [52]. In many cases, the macrophytes and seagrass had unstable relationships specifically with oscillation between the EDSUs and biometrics in time and space [60,61]. These all relationships biased the estimates besides the other scatterers were available suchlike as epiphytes on the leaves and fishes among the seagrass [35,59,61]. The meadow showed inconsistent linear relationships in some seasons [60] when its biological and physiological activities were intensified [83]. Another seagrass, C. nodosa produced a more stable relationships in time than P. oceanica performed [60,61]. The POSIBIOM regarded these matters to estimate biomass by using three different equations of the relationships established in each season [60]. However, any flower (only one flower found) and fruit of the meadow was not encountered during the all surveys of the present study. The flower and fruit which occurred often in the West Mediterranean Sea [84,85] would change the acoustical traits of the meadow in other parts of the Mediterranean coasts. The main factor producing the variation in the estimates is the strength and magnification of the physiological activities of the meadow, calcification, oxygen gas releasing, beam pattern (orientation), age/type (juvenile, middle and adult leaves) and hardness of the leaves, and bottom depth of the occurrence of the meadow [60,61,62,86,87,88,89,90]. The relationships used in the present study were established by means of an in/ex situ study at a bottom depth of 15 for each season. The 15 m was a critical range with its inconsistent biometrics in high variation between shallow and deep Posidonia [5,18]. The fish removal was standardized according to the results of the acoustical ranges of the meadows, but possible significant epiphytes could be disregarded for the removal. For this, SCUBA sampling was needed for assess acoustical contribution to the EDSUs of the meadow leaves by solving the forward model solution [60]. This was not a practical way for the clear biomass estimates with frequent SCUBA samplings as a destructive method. Mutlu and Olguner [61] estimated the contribution of a calcareous alga, Pneophyllum fragile to the EDSUs of leaves of C. nodosa, and the contributions remained below the threshold of the leaves depending on abundance (ind/m²) of the alga. Another calcareous alga, Hydrolithon boreale occurred recently on leaves of P. oceanica at the present study, its abundances were found to be low as compared of that of C. nodosa [35,61].

The POSIBIOM could be the best offer currently for the biometric estimates regarding the matters aforementioned, and some improvements are needed during the next studies to decrease possible disadvantages. There are some biases in estimating the biomass and bottom coverage compared to the estimates by SCUBA (Appendix 8 and Appendix 9). This could be due to misestimates of bottom recovery for the coverage and the equations established at 15 m for the biomass [60]. A bottom depth of 15 m was found to be critical depth for occurrence of the high variations of the biometrics [5,18]. The negatives/troubleshooting were discussed for each algorithm in Table 1, Table 2 and Table 3.

As a conclusion from the present to future studies featured to improve the algorithms;

The algorithm would open acoustical data of the different echosounders manufactured by other companies rather than that by BioSonics,

A way would be developed for reading the acoustical row data directly, needless of the post-processors,

The debugs would be considered to get rid of them if the feedback about that was returned by the users after the package would be a profession version converted to a programming language of C,

The multi-frequencies would be applied if the data was collected, and then the algorithm would be figured out and configured accordingly,

The experiments would be repeated to establish the relationships between the EDSUs and biometrics at the different bottom depths, else than 15 m,

Leaf types would be also taken into account of the equations of the relationships,

Mapping the results would be associated with different types of the coastal map,

Eventually, the software of the package would be hardware connected to the echosounder, which needs too much works associated with the electronics.