the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Sep 2023
LOCATE v1.0: Numerical Modelling of Floating Marine Debris Dispersion in Coastal Regions Using Nested Hydrodynamic Grids and Parcels v2.4.2
Leidy M. Castro-Rosero
Manuel Espino
Jose M. Alsina Torrent
Abstract. The transport mechanisms of floating marine debris in coastal zones remain poorly understood, primarily due to the complex geometry and influence of coastal processes which pose difficulties in incorporating them into Lagrangian numerical models at coastal scales. The numerical model presented in this study addresses these challenges using nested grids of varying hydrodynamic resolutions (2.5 km, 350 m, and 70 m). The model couples Eulerian hydrodynamic data with a Lagrangian particle solver to accurately simulate the motion of plastic particles in coastal regions. A Lagrangian validation of the model was conducted using drifter data to assess the model’s skill and confidence in simulation predictions. The results demonstrated high skill and stability in 72-hour forecast horizons and high skill score values close to the shoreline. Additionally, a beaching sensitivity analysis was performed to determine particle beaching parameterisations suitable for coastal zones. This analysis demonstrated that the real-time distance of particles to the shore during simulations was the most accurate method for detecting the land-water boundary at coastal scales, accurately representing residence times and spatiotemporal beaching patterns. Simulations were conducted using nested grids and a single coarse grid, both using the same plastic input from river outflows in a pilot test at the Barcelona coastline with a complex morphology due to coastal structures, two major river mouths, and a large harbour. The simulations revealed marked differences in beaching amounts and residence times. Both simulations exhibited high levels of beaching, with the nested grid simulation registering 91.5 % beaching and the low-resolution grid simulation showing 95.8 % beaching, surpassing other studies conducted at larger scales. Beaching amounts exhibited high levels of variability within demarcated areas between simulations, with the highest flux observed near the Llobregat River mouth release point. The inner port area showed that particles have 18 times longer residence times when using high-resolution data, demonstrating the model’s ability to solve complex coastal geometrical structures. By using nested grids, the model can resolve coastal processes with high-resolution hydrodynamic data and accurately simulate potential accumulation zones and litter hot-spots at coastal scales.
- Preprint
(9372 KB) - Metadata XML
- BibTeX
- EndNote
Ivan Hernandez et al.
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2023-160', Anonymous Referee #1, 24 Oct 2023
Hernandez et al. introduce a modelling framework (LOCATE) for simulating marine dispersal in coastal regions using nested hydrodynamic grids. The authors apply LOCATE to a nested model of the Barcelona coastline (forced by a regional CMEMS model), and compare trajectory forecasting skill and beaching behaviour between the (higher resolution) nested model and (lower resolution) regional model, as well as the sensitivity of beaching to three beaching parameterisations.
Although LOCATE has been used in previous studies, this manuscript represents a considerable effort to document, describe, and illustrate a potential application for the framework. This is a great example of ‘open science’, and the quality of figures is also very high. Unfortunately, in its current form, the manuscript suffers from a lack of purpose and clarity, and feels more like a detailed supplement to Castro-Rosero et al. (2023) than a free-standing study. If the main purpose of the manuscript is to facilitate the wider adoption of LOCATE, the manuscript should clearly explain the advantages of LOCATE over just using Parcels directly, since the nested grid capability of Parcels is already quite accessible (i.e. if I were to investigate dispersal across nested grids, it is not clear to me why I would use LOCATE rather than ‘pure’ Parcels). If the main purpose is to instead explore the utility of nested grids for particle tracking in coastal waters, more thorough validation is needed than comparison with a single drifter profile. I realise that there is limited observational data available given the small domain size (using drogued drifters may help), but many of the manuscript’s claims about “model accuracy” are weakly supported by evidence, and the manuscript therefore provides limited insights into the (dis)advantages of using nested grids beyond “there are differences”. I also have concerns about some aspects of the methods, and other aspects were very difficult for me to understand.
If the authors can address the general comments above, then I would recommend major revisions to the manuscript (please see the attached document for specific and technical comments).
-
AC2: 'Reply on RC1', Ivan Hernandez, 16 Nov 2023
First of all, we wish to thank the referee for their extensive review and comments which have raised some very valid points.
The purpose of LOCATE is to provide a system to facilitate the study of the dispersion of marine debris in small-scale coastal settings. To achieve this, the use of nested hydrodynamic grids with high-resolution data is applied, and although this could be done using Parcels alone as suggested by the referee, the main purpose of LOCATE’s methodology is to do so in conjunction with an appropriate method for detecting when and where particles arrive on land with as much precision as possible from the hydrodynamic data available and often given the complex coastal geometry. Given that coastal areas can experience very high amounts of beaching, special emphasis has been placed on the detection of the beaching of particles. This is especially relevant for areas such as the Barcelona coastline described in this study characterized by a complex geometry with a major harbour and several marinas (Port Olimpic, Forum), groynes, breakwaters and so on. Barcelona experiences discharge of debris on the coastline from various sources, such as rivers or other discharge outlets commonly found in urbanised coastal areas where debris is transported and interacts with the described complex geometry.
A beaching module based on a distance-to-shore parameter that detects the pre-calculated real shoreline by using distance data within grids during the simulation is included in LOCATE, which could be adapted to other areas where high-resolution coastline data may be available. The beaching module developed within LOCATE allows for a precise determination of which areas can be more affected by particle beaching. The beaching module adapts to each computational grid. In small-scale studies, such as this one, such precision and highly granular information which can be determined in the post-simulation analysis when using the distance-based beaching module, is pertinent to being able to determine which areas, such as specific beaches or port structures, could be more at risk of receiving debris.
In the present work, it is shown that LOCATE’s use of high-resolution hydrodynamic data that resolves coastal processes in conjunction with the distance-to-shore beaching module, allowed for substantially greater precision in terms of the measurements of where particles became beached. The reason for this is that by using these data together, complex geometric structures can be solved to a much greater extent than using the IBI-CMEMS grid alone where such structures are not taken into account or ‘seen’ due to the grid’s coarse resolution. Furthermore, using the current velocity as a land detection parameter to determine when and where a particle becomes beached has been shown in this study to be insufficient and imprecise at localised scales, even if this has been widely used by other studies, albeit at much larger, even global scales.
The LOCATE model offers areas of functionality tailored towards use for coastal areas that would require substantial time investment and programming using Parcels from scratch. As mentioned above, the distance-to-shore beaching module can be used as a tool to determine where exactly particles become beached independently of hydrodynamic data resolution, while recreating real-life scenarios such as the continuous release of particles by a river or other discharge outlet can be easily achieved with relatively minor configurations. LOCATE can also be easily configured to track particle dispersion of one-off releases, such as runoff events, or even track drifters or tracers. Since there are high-resolution circulation data available throughout the Spanish coastline via the Spanish Port Authority (Puertos del Estado), LOCATE could be easily applied for similar studies in any of those areas with relative ease. Although Parcels is the engine behind the LOCATE model, it lacks the necessary considerations and requirements, which are included within LOCATE, to provide precise simulations of marine debris at localised coastal scales.
The concern about the model’s accuracy is duly noted and in return, what we propose is that the model can be used for more precise measurements of particle beaching at localised settings. The abstract and conclusion have been edited to reflect this change. The model’s validation has been extended with data on two more drifters that serve to increase the model’s confidence to be used in coastal settings, where more consistent good skill scores were achieved when using the nested grids than using only the IBI-CMNEMS grid. However, the use of nested grids alone is not the sole focus of LOCATE as outlined above.
It is hoped that the concerns shown in the comments by the referee are addressed in the above statement and in the response to each of their points. Far from being a “detailed supplement” to other published works which have also used LOCATE to simulate the dispersion of particles, this work sets out to present a complete outline of the purpose and functionality available in LOCATE for use in coastal areas which may be affected by high amounts of beaching of debris.
With this in mind, the abstract (below), section 1.4 (objectives) and conclusion have been modified to reflect the points made above to offer greater clarity regarding the purpose of the study and the objectives. To avoid further confusion and add clarity the title has been shortened to:
LOCATE v1.0: Numerical Modelling of Floating Marine Debris Dispersion and Beaching in Coastal Regions Using Parcels v2.4.2-
RC3: 'Reply on AC2', Anonymous Referee #1, 21 Nov 2023
I thank the authors for their detailed response to my review. The purpose of the manuscript is now clearer to me, and I agree with the authors that a well-documented and transferable modelling system including beaching would be a useful contribution to the modelling community. This resolves one of my major concerns about the manuscript. I agree with many of the authors’ responses to other points raised in my review, and look forward to reading the revised manuscript.
However, I am writing this pre-emptive comment as a few points in my review may not have been made clearly enough, and I would like to clarify a couple of (significant) unresolved concerns before the manuscript is revised - please see the attached pdf.
- AC3: 'Reply on RC3', Ivan Hernandez, 24 Nov 2023
-
RC3: 'Reply on AC2', Anonymous Referee #1, 21 Nov 2023
-
AC2: 'Reply on RC1', Ivan Hernandez, 16 Nov 2023
-
RC2: 'Comment on gmd-2023-160', Joseph Harari, 09 Nov 2023
The submission has scientific relevance, quality and reproducibility. But the submitted text is too long and difficult to read, it should be much more concise. The conclusions are too much obvious: a nested high-resolution model provides better results! Should insert some numerical results in the conclusions, proving the modelling improvements and the scientific contribution of the research.
See other comments on the attached file.
-
AC1: 'Reply on RC2', Ivan Hernandez, 10 Nov 2023
Thank you very much for the comments which are addressed accordingly:
- Line 55: Use density instead of specific gravity
- Agree, will change this term
- Line 126: Some scripts and folders are cited but not given in the Appendix
- In section 2.1 The following sentence will be added:
“The scripts used for LOCATE and preprocessing scripts mentioned hereon can be found in the code availability section” - Mention is already made in line 123 to the LOCATE repository, referenced as Hernandez, et al., 2023
- In section 2.1 The following sentence will be added:
- Line 145: Regarding citing the environment variables for CMEMS.
- Reference to these will be removed as this is in the instructions provided in the repository, and agree that it is not necessary.
- Lines 274 and 305: A map showing the release points would be useful.
- Such a map is already provided in Figure 7, where the release points are displayed in 7a, 7c and 7e. Figure 3, however, has also been edited to show these release points in the next version of the manuscript.
- Line 372 and 382: Showing validation results in section 3.1 in a table
- These results are already shown as a plot in Figure 6 instead of a table because it was considered by the authors that it could be better understood that way. Better referencing in the text of section 3.1 to the individual plots will be included in the updated manuscript.
- Line 394: Include a short description of each test in Table 4.
- This will be included.
- Line 483:
- This has already been changed in the next version of the manuscript to “wave-induced Eulerian (mean) currents” as per the recommendation of Referee 1.
- Final comments on the conclusion:
- It is hoped that an updated version of the manuscript will be clearer and easier to read. The LOCATE model required substantial outlining and explaining, as well as the different tests contained within, such as the validation process which has been extended to include more drifter data, the beaching sensitivity test to define the beaching parameter at coastal scale moving forward, and the simulation using realistic debris discharge data to test the model output and beaching patterns.
Regarding the conclusion, this has been reworked given the comments from referee1 as well. The point of this study and in turn, its scientific contribution, which perhaps was not made strongly enough, is that while using high-resolution hydrodynamic data is important at coastal scales because they can resolve (some) coastal processes not available using coarse-resolution data, it can also resolve complex geometric structures which are otherwise indistinguishable using low-resolution data that used in many other studies. This was seen in the port area where particles spent 18 times longer with the nested grids compared to the IBI-CMEMS grid. Resolving complex strucutres is especially relevant at coastal scales to identify which areas were most affected by the beaching of particles. The distance-to-shore-based beaching module was developed to calculate a particle’s distance to the shoreline in real-time using high-resolution shoreline data to adequately consider coastal areas in the model in terms of land-water boundary detection in areas of high complexity. Using the high-resolution hydrodynamic and shoreline data in conjunction can provide much more precise information regarding the affected areas around the coastline. This beaching module can also be used independently of the resolution of the hydrodynamic data and still provide more precise beaching measurements even when not applying nested grids as can be seen in the beaching sensitivity test.
While the ability to implement this development can be scale-dependent, if the shoreline data is available at coastal scales then the beaching module and methodology presented in this study would be beneficial, as the beaching parameters used in previous studies e.g. when a particle moves into a dry cell or stagnation based on current velocity, are insufficient to determine which local areas are more at risk from the deposition of plastic pollution. These points will be extensively presented in the discussion for referee 1, which made similar observations, and will be included in the updated version of the manuscript.
- It is hoped that an updated version of the manuscript will be clearer and easier to read. The LOCATE model required substantial outlining and explaining, as well as the different tests contained within, such as the validation process which has been extended to include more drifter data, the beaching sensitivity test to define the beaching parameter at coastal scale moving forward, and the simulation using realistic debris discharge data to test the model output and beaching patterns.
Citation: https://doi.org/10.5194/gmd-2023-160-AC1 - Line 55: Use density instead of specific gravity
-
AC1: 'Reply on RC2', Ivan Hernandez, 10 Nov 2023
Ivan Hernandez et al.
Ivan Hernandez et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 225 | 50 | 11 | 286 | 1 | 3 |
- HTML: 225
- PDF: 50
- XML: 11
- Total: 286
- BibTeX: 1
- EndNote: 3
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1