Deciphering the deep: machine learning approaches to understanding oceanic ecosystems

Размещено на http://www.Allbest.Ru/

Institute of marine and environmental sciences university of Szczecin

Faculty of Physical, Mathematical and Natural Sciences University of Szczecin

Maritime University of Szczecin

Deciphering the deep: machine learning approaches to understanding oceanic ecosystems

T. Miller, PhD in biolog. Sci., Ass. Professor

A. Eobodzinska, student of Biotechnology

O. Kaczanowska, student of Oceanography

I. Durlik, Mr Mariner, Research Assistant

P. Kozlovska, student of Biotechnology

K. Lewita, student of Biotechnology

Szczecin, Poland

Summary

This paper presents a detailed exploration of the transformative role of Machine Learning (ML) in oceanographic research, encapsulating the paradigm shift towards more efficient and comprehensive analysis of marine ecosystems. It delves into the multifaceted applications of ML, ranging from predictive modeling of ocean currents to in-depth biodiversity analysis and deciphering the complexities of deep-sea ecosystems through advanced computer vision techniques. The discussion extends to the challenges and opportunities that intertwine with the integration of AI and ML in oceanography, emphasizing the need for robust data collection, interdisciplinary collaboration, and ethical considerations. Through a series of case studies and thematic discussions, this paper underscores the profound potential of ML to revolutionize our understanding and preservation of oceanic ecosystems, setting a new frontier for future research and conservation strategies in the realm of oceanography.

Keywords: Machine Learning, Oceanography, Marine Biodiversity, Predictive Modeling, Deep-Sea Exploration

Анотація

Розшифровка глибин: підходи машинного навчання до розуміння океанічних екосистем

Ця стаття представляє детальне дослідження трансформаційної ролі машинного навчання (ML) в океанографічних дослідженнях, інкапсулюючи зміну парадигми до більш ефективного та комплексного аналізу морських екосистем. Він заглиблюється в багатогранне застосування ML, починаючи від прогнозного моделювання океанських течій до поглибленого аналізу біорізноманіття та розшифровки складності глибоководних екосистем за допомогою передових методів комп'ютерного зору. Обговорення поширюється на виклики та можливості, які переплітаються з інтеграцією штучного інтелекту та машинного навчання в океанографію, наголошуючи на необхідності надійного збору даних, міждисциплінарної співпраці та етичних міркувань. Завдяки серії тематичних досліджень і тематичних обговорень ця стаття підкреслює величезний потенціал МЛ для революції в нашому розумінні та збереженні океанічних екосистем, встановлюючи новий рубіж для майбутніх досліджень і стратегій збереження в царині океанографії.

Ключові слова: машинне навчання, океанографія, морське біорізноманіття, прогнозне моделювання, глибоководне дослідження

Introduction

The New Frontier of Oceanography

The vast and enigmatic realm of the ocean has long been a subject of fascination and study. For centuries, humans have endeavored to comprehend the mysteries that lie beneath the surface of the world's oceans. As our understanding has grown, so too have the tools and methodologies we use to study these crucial ecosystems. Today, we stand on the precipice of a new era in oceanographic research, where the traditional methodologies meet the transformative power of technology: the frontier of Machine Learning (ML) [1]. Machine Learning, a subset of artificial intelligence, has seen widespread adoption across a plethora of scientific disciplines, proving to be an invaluable tool in data analysis and predictive modeling. In the field of oceanography, these capabilities hold significant potential. The oceans are vast and complex systems, with an array of dynamic factors influencing their behavior. Understanding these factors and the relationships between them is a challenge that seems tailor-made for the capabilities of ML [2]. By leveraging ML, we can unearth patterns and connections in oceanic data that might otherwise remain obscured. This ability to glean insights from vast volumes of data, coupled with the predictive power of ML models, stands to revolutionize our approach to oceanographic research. From predicting the movements of ocean currents to understanding the intricacies of marine ecosystems, the possibilities are as deep and wide as the oceans themselves [3].

As we venture into this new frontier of oceanography, it is crucial to understand the role of ML in oceanic studies, the opportunities it presents, and the challenges we might encounter. This new era beckons us to dive deeper than ever before, deciphering the secrets of the deep through the lens of Machine Learning. Through this exploration, we seek not just to understand the oceans, but to shape the future of our relationship with this vital component of our planet. The ocean has always been a frontier; with machine learning, it's a frontier that we're better equipped than ever to explore [4].

The Role of Machine Learning in Oceanic Studies

Machine Learning (ML), with its ability to find patterns in large and complex datasets, has emerged as a powerful tool in oceanic studies. It is transforming the way researchers approach their studies of the sea, revolutionizing how we interpret data, make predictions, and understand the complexity of marine ecosystems [5,6]

Firstly, ML algorithms can help us make sense of the vast and diverse oceanic data we collect. Oceanographic data is inherently multidimensional, encompassing variables such as temperature, salinity, pressure, nutrient concentrations, and biological diversity. Traditional analytical methods often struggle to discern patterns and relationships within such complex datasets. ML, on the other hand, thrives on complexity. Through unsupervised learning, ML can identify patterns and relationships within the data that human analysts might miss, offering us new insights into the workings of the oceanic system [7,8].

Secondly, ML excels in predictive modeling. Predicting future oceanic conditions, such as sea temperatures, currents, or biological productivity, is a critical aspect of oceanographic research. Predictive models built using ML algorithms can learn from past data to predict future conditions with a high degree of accuracy. These predictive capabilities can be invaluable in numerous applications, from climate studies to marine resource management [9,1 0]

Lastly, ML can aid in the understanding of marine ecosystems. Marine ecosystems are complex, dynamic, and interconnected. Machine learning techniques such as clustering and classification can help us understand the structure and functioning of these ecosystems, identify species, and detect changes in ecosystem health over time [11,12].

As we continue to refine these technologies and apply them to the oceanic realm, the role of machine learning in oceanographic studies will only become more significant. It offers a powerful toolset for deciphering the deep, enhancing our understanding of the ocean's many mysteries, and informing our efforts to conserve and manage this vital global resource. With the integration of machine learning into oceanic studies, we are truly entering a new era of discovery and understanding.

Challenges and Opportunities in Applying AI to Oceanography

As we continue to incorporate Artificial Intelligence (AI) and Machine Learning (ML) into the field of oceanography, it is important to acknowledge both the challenges and opportunities that come with this integration [9,13].

On the challenges side, data quality and availability are often the first hurdles. The ocean is vast and dynamic, which makes data collection difficult. Even with advancements in technology, our ability to collect and monitor data over large spatial and temporal scales is limited. Furthermore, the data that we do collect can be noisy, incomplete, or inconsistent, which can complicate the training and accuracy of ML models [14].

Another significant challenge is the need for interdisciplinary collaboration. The application of AI to oceanography involves both ocean scientists and data scientists, fields with different terminologies, methodologies, and perspectives. Bridging this gap requires effective communication and collaboration, ensuring that the machine learning models being developed are both scientifically valid and technologically sound [15].

Despite these challenges, the opportunities for applying AI to oceanography are vast. ML can handle the vast amounts of complex and multidimensional data collected in oceanography, identifying patterns and relationships that may not be apparent through traditional analysis methods. This can lead to new insights about the oceanic environment and its processes [16].

In addition, AI and ML offer powerful predictive capabilities. They can be used to forecast future oceanic conditions, such as temperature, salinity, and biological productivity, aiding in everything from climate modeling to fisheries management.

Perhaps most excitingly, AI has the potential to accelerate the pace of discovery in oceanography. By automating data analysis and prediction, AI allows researchers to focus on interpreting results and formulating new hypotheses. This could lead to more rapid advancements in our understanding of the ocean and its many intricate systems [8].

While the integration of AI into oceanography presents certain challenges, the opportunities for enhanced understanding and discovery are profound. By embracing this intersection of disciplines, we can push the boundaries of our knowledge and usher in a new era of oceanographic research [11].

Case Study: Predicting Ocean Currents Using Machine Learning Models

Ocean currents, the continuous and directed movement of ocean water, play a pivotal role in shaping the climate, marine transportation routes, and marine ecosystems. Predicting these currents has long been a challenge due to the dynamic and complex nature of the oceanic system. However, the advent of machine learning (ML) has opened new horizons in this realm of study. Let's delve into a specific case where ML was successfully used to predict ocean currents [10,13].

In this case study, researchers were faced with the task of predicting the patterns of ocean currents in the Atlantic Ocean. Traditional methods of prediction, which relied heavily on physical oceanographic models, were found to be lacking in both speed and accuracy. To overcome these shortcomings, the researchers turned to ML [9,18].

The researchers used a diverse set of data inputs, including sea surface temperature, sea level anomalies, wind patterns, and historical current data, collected from satellite observations and in-situ measurements. They opted to use a type of ML known as a convolutional neural network (CNN), a deep learning algorithm that excels at processing grid-like data, such as the spatial datasets often found in oceanography [19,20].

The CNN was trained on several years of data, during which it learned to identify patterns and relationships between the various data inputs and the resulting current patterns. Once trained, the model was able to predict future current patterns based on new input data [18,21].

The results were striking. The ML model outperformed traditional predictive models in terms of both speed and accuracy. It was able to generate predictions significantly faster, and its predictions aligned more closely with actual observed current patterns [8,11].

This case study illustrates the potential of machine learning in predicting ocean currents. By leveraging the pattern-recognition and predictive capabilities of ML, we can enhance our understanding of ocean dynamics and improve our predictive models. Such advancements not only aid in scientific understanding but also have practical implications, from improving maritime navigation to informing strategies for mitigating the effects of climate change [12,22].

It's important to note that while this study represents a significant advancement, it's just one example of the potential applications of ML in oceanography. With continued research and development, we can expect to see even more sophisticated and accurate ML models for predicting ocean currents and other aspects of the marine environment [8,23].

machine learning oceanic marine biodiversity ecosystem

Implementing ML Algorithms for Marine Biodiversity Analysis

Marine biodiversity, the variety of life in the oceans, is a key indicator of the health of marine ecosystems. It encompasses everything from the genetic diversity within a species to the variety of ecosystems found across different oceanic regions. Monitoring and analyzing marine biodiversity is a monumental task, but one where Machine Learning (ML) algorithms can provide significant assistance [20,24].

Implementing ML in marine biodiversity analysis involves several steps. Firstly, it requires the collection of relevant data. This data can come from a variety of sources, such as satellite imagery, underwater cameras, acoustic monitoring, and genetic sequencing. Each of these sources provides a different perspective on marine biodiversity, offering information on species distributions, population densities, genetic diversity, and more [23,25].

Once data has been collected, it needs to be preprocessed for use in ML algorithms. This may involve cleaning the data, dealing with missing values, normalizing measurements, and converting the data into a form suitable for the chosen ML algorithm [18,25].

The next step is to choose an appropriate ML algorithm. The choice of algorithm depends on the nature of the data and the specific task at hand. For example, clustering algorithms can be used to identify groups of similar species or habitats, while classification algorithms can be used to identify individual species in images or acoustic recordings. Regression algorithms, on the other hand, might be used to predict changes in species populations or biodiversity indices based on environmental factors [3,26].

Once an algorithm has been chosen, it needs to be trained using a subset of the available data. The trained model can then be validated and tested using separate data subsets to ensure it performs well and avoids problems like overfitting [1,27].

After these steps, the ML model is ready to be used for biodiversity analysis. It could be used to identify trends in biodiversity over time, predict future changes in biodiversity based on environmental factors, identify key species or habitats, and more [12,21].

Implementing ML algorithms for marine biodiversity analysis is not without its challenges. These include the need for large amounts of high-quality data, the complexity of marine ecosystems, and the need for interdisciplinary collaboration. However, the potential benefits, in terms of the insights and efficiencies gained, make it a promising approach for advancing our understanding and conservation of marine biodiversity. As we continue to develop and refine these techniques, we can expect them to play an increasingly important role in marine ecology and conservation efforts [28,29].

Using Machine Learning (ML) to Understand Deep Sea Ecosystems

The deep sea, extending from around 200 meters below the ocean surface to the seafloor, is the largest and least explored habitat on Earth. It houses an incredible diversity of life forms and ecosystems, many of which remain poorly understood due to the difficulties of exploring such a remote and hostile environment. (ML) offers promising new ways to analyze and understand these deep-sea ecosystems [20,30].

One of the most important applications of ML in deep sea research involves analyzing imagery data from remotely operated vehicles (ROVs) and autonomous underwater vehicles (AUVs). These vehicles can descend to the deep sea and capture high-resolution images and videos of the seafloor and the organisms that inhabit it. ML algorithms, particularly those involving computer vision, can be trained to identify and classify organisms in these images, helping scientists to understand the composition of deep-sea communities [31,32].

For example, convolutional neural networks (CNNs), a type of ML model particularly well-suited to image analysis, can be trained to recognize different species of deep-sea organisms, from tiny microscopic plankton to larger animals like fishes and squids. This automated identification and classification can save researchers countless hours of manual image analysis and provide more accurate and consistent results [6,33].

Another application of ML in understanding deep-sea ecosystems involves predicting the distribution of species or habitats based on environmental data. The deep sea is characterized by a range of environmental conditions, including temperature, pressure, and nutrient availability, which can greatly influence the distribution and diversity of organisms. ML models can be trained on environmental data and species occurrence data to predict where certain species or habitats are likely to be found [34,35].

These models can help scientists understand the factors that drive the distribution of deep-sea species and predict how these distributions may change in response to environmental changes such as climate change or human activities. Moreover, these models can inform conservation efforts by identifying areas of high biodiversity or unique habitats that may be particularly worthy of protection [36,37].

In conclusion, the application of ML to understanding deep-sea ecosystems holds great promise. By automating the analysis of imagery data and predicting species distributions, ML can help us to unravel the mysteries of the deep sea and to better conserve and manage this vast and important part of our planet. Despite the challenges inherent in studying the deep sea, advancements in technology like ML are pushing the boundaries of what we can achieve and know about this enigmatic realm [38,39].

Conclusion

Future Perspectives of Machine Learning in Oceanography

As we gaze into the future of oceanography, it is clear that Machine Learning (ML) will play a pivotal role in shaping our understanding of the oceans. By enabling us to glean insights from vast volumes of complex data, ML is revolutionizing the way we study marine ecosystems, predict oceanic phenomena, and conserve marine biodiversity.

The potential applications of ML in oceanography are as vast and varied as the oceans themselves. From predicting ocean currents to identifying marine species, from modeling the spread of pollutants to forecasting the impacts of climate change, the possibilities are virtually limitless. Each of these applications holds the promise of deepening our understanding of the oceans and enhancing our ability to protect and manage this vital global resource.

Looking ahead, we can expect to see the role of ML in oceanography continue to grow. As our computational capabilities advance and our data collection technologies become more sophisticated, the scope and accuracy of ML models will only increase. Simultaneously, the integration of ML with other emerging technologies, such as remote sensing and autonomous underwater vehicles, will open up new avenues for oceanographic research.

However, the future of ML in oceanography is not without challenges. Issues such as data quality and availability, the need for interdisciplinary collaboration, and the ethical considerations of AI use will all require careful attention. Nonetheless, the benefits that ML can bring to oceanographic research are considerable, and the potential for discovery is enormous.

The marriage of oceanography and ML represents a new frontier of scientific exploration, one that promises to deepen our understanding of the world's oceans and their myriad mysteries. By embracing this intersection of disciplines, we can push the boundaries of our knowledge, inform our stewardship of the oceans, and shape the future of our blue planet.

References

1. Sonnewald, M.; Lguensat, R.; Jones, D.C.; Dueben, P.D.; Brajard, J.; Balaji, V. Bridging Observations, Theory and Numerical Simulation of the Ocean Using Machine Learning. Environmental Research Letters 2021, 16, 073008,

2. Jun, H.; Jou, H.-T.; Kim, C.-H.; Lee, S.H.; Kim, H.-J. Random Noise Attenuation of Sparker Seismic Oceanography Data with Machine Learning. Ocean Science 2020, 16, 1367-1383

3. McNabb, B.J.; Tortell, P.D. Oceanographic Controls on Southern Ocean Dimethyl Sulfide Distributions Revealed by Machine Learning Algorithms. Limnol Oceanogr 2023, 68, 616630

4. Salman, A. Editorial: Application of Machine Learning in Oceanography and Marine Sciences. Front Mar Sci 2023, 10

5. Ahmad, H. Machine Learning Applications in Oceanography. Aquatic Research 2019, 161169

6. McEver, R.A.; Zhang, B.; Levenson, C.; Iftekhar, A.S.M.; Manjunath, B.S. Context-Driven Detection of Invertebrate Species in Deep-Sea Video. IntJ Comput Vis 2023, 131, 13671388

7. Castelao, G.P. A Machine Learning Approach to Quality Control Oceanographic Data. Comput Geosci 2021, 155, 104803

8. Park, S.; Byun, J.; Shin, K.-S.; Jo, O. Ocean Current Prediction Based on Machine Learning for Deciding Handover Priority in Underwater Wireless Sensor Networks. In Proceedings of the 2020 International Conference on Artificial Intelligence in Information and Communication (ICAIIC); IEEE, February 2020; pp. 505-509.

9. Dong, C.; Xu, G.; Han, G.; Bethel, B.J.; Xie, W.; Zhou, S. Recent Developments in Artificial

10. Intelligence in Oceanography. Ocean-Land-Atmosphere Research 2022,2022,

11. Sinha, A.; Abernathey, R. Estimating Ocean Surface Currents With Machine Learning. Front Mar Sci 2021, 8,

12. Restreppo, G.A.; Wood, W.T.; Phrampus, B.J. Oceanic Sediment Accumulation Rates Predicted via Machine Learning Algorithm: Towards Sediment Characterization on a Global Scale. Geo-Marine Letters 2020, 40, 755-763

13. Rubbens, P.; Brodie, S.; Cordier, T.; Destro Barcellos, D.; Devos, P.; Fernandes-Salvador, J.A.; Fincham, J.I.; Gomes, A.; Handegard, N.O.; Howell, K.; et al. Machine Learning in Marine Ecology: An Overview of Techniques and Applications. ICES Journal of Marine Science 2023, 80, 1829-1853

14. Chen, G.; Huang, B.; Chen, X.; Ge, L.; Radenkovic, M.; Ma, Y. Deep Blue AI: A New Bridge from Data to Knowledge for the Ocean Science. Deep Sea Research Part I: Oceanographic Research Papers 2022, 190, 103886

15. Weller, R.A.; Baker, D.J.; Glackin, M.M.; Roberts, S.J.; Schmitt, R.W.; Twigg, E.S.; Vimont, D.J. The Challenge of Sustaining Ocean Observations. Front Mar Sci 2019, 6, doi:10.3389/fmars.2019.00105.

16. Faria, M.; Pinto, J.; Py, F.; Fortuna, J.; Dias, H.; Martins, R.; Leira, F.; Johansen, T.A.; Sousa, J.; Rajan, K. Coordinating UAVs and AUVs for Oceanographic Field Experiments: Challenges and Lessons Learned. In Proceedings of the 2014 IEEE International Conference on Robotics and Automation (ICRA); IEEE, May 2014; pp. 6606-6611.

17. Vance, T.C.; Wengren, M.; Burger, E.; Hernandez, D.; Kearns, T.; Medina-Lopez, E.; Merati, N.; O'Brien, K.; O'Neil, J.; Potemra, J.T.; et al. From the Oceans to the Cloud: Opportunities and Challenges for Data, Models, Computation and Workflows. Front Mar Sci 2019, 6

18. Goodwin, M.; Halvorsen, K.T.; Jiao, L.; Knausgard, K.M.; Martin, A.H.; Moyano, M.; Oomen, R.A.; Rasmussen, J.H.; S0rdalen, T.K.; Thorbj0rnsen, S.H. Unlocking the Potential of Deep Learning for Marine Ecology: Overview, Applications, and Outlook. ICES Journal of Marine Science 2022, 79, 319-336

19. Kavousi-Fard, A.; Su, W. A Combined Prognostic Model Based on Machine Learning for Tidal Current Prediction. IEEE Transactions on Geoscience and Remote Sensing 2017, 55, 3108-3114

20. Bayindir, C. Predicting the Ocean Currents Using Deep Learning. 2019.

21. Dauji, S.; Deo, M.C.; Bhargava, K. Prediction of Ocean Currents with Artificial Neural Networks. ISH Journal of Hydraulic Engineering 2015, 21,14-27

22. Immas, A.; Do, N.; Alam, M.-R. Real-Time in Situ Prediction of Ocean Currents. Ocean Engineering 2021, 228, 108922

23. Sadaiappan, B.; Balakrishnan, P.; C.R., V.; Vijayan, N.T.; Subramanian, M.; Gauns, M.U. Applications of Machine Learning in Chemical and Biological Oceanography. ACS Omega 2023, 8, 15831-15853

24. Chen, V.Y.; Lu, D.-J.; Han, Y.-S. Hybrid Intelligence for Marine Biodiversity: Integrating Citizen Science with AI for Enhanced Intertidal Conservation Efforts at Cape Santiago, Taiwan. Sustainability 2024, 16, 454,

25. Brito de Jesus, S.; Vieira, D.; Gheller, P.; Cunha, B.P.; Gallucci, F.; Fonseca, G. Machine Learning Algorithms Accurately Identify Free-Living Marine Nematode Species. PeerJ 2023, 11, e16216

26. Ditria, E.M.; Buelow, C.A.; Gonzalez-Rivero, M.; Connolly, R.M. Artificial Intelligence and Automated Monitoring for Assisting Conservation of Marine Ecosystems: A Perspective. Front Mar Sci 2022, 9

27. Hu, J.-H.; Tsai, W.-P.; Cheng, S.-T.; Chang, F.-J. Explore the Relationship between Fish Community and Environmental Factors by Machine Learning Techniques. Environ Res 2020, 184, 109262

28. Marburg, A.; Bigham, K. Deep Learning for Benthic Fauna Identification. In Proceedings of the Oceans 2016 MTS/IEEE Monterey; September 2016; pp. 1-5.

29. da Silveira, C.B.L.; Strenzel, G.M.R.; Maida, M.; Gaspar, A.L.B.; Ferreira, B.P. Coral Reef Mapping with Remote Sensing and Machine Learning: A Nurture and Nature Analysis in Marine Protected Areas. Remote Sens (Basel) 2021, 13, 2907

30. Tamvakis, A.; Trygonis, V.; Miritzis, J.; Tsirtsis, G.; Spatharis, S. Optimizing Biodiversity Prediction from Abiotic Parameters. Environmental Modelling & Software 2014, 53, 112120

31. Lou, R.; Lv, Z.; Dang, S.; Su, T.; Li, X. Application of Machine Learning in Ocean Data. Multimed Syst 2023, 29, 1815-1824

32. Egbert, H.; Stanchev, L. Hierarchical Classification of Deep-Sea Videos. In Proceedings of the 2022 IEEE 16th International Conference on Semantic Computing (ICSC); IEEE, January 2022; pp. 119-122.

33. Manochkumar, J.; Cherukuri, A.K.; Kumar, R.S.; Almansour, A.I.; Ramamoorthy, S.; Efferth, T. A Critical Review of Machine-Learning for "Multi-Omics" Marine Metabolite Datasets. Comput Biol Med 2023, 165, 107425

34. Han, F.; Yao, J.; Zhu, H.; Wang, C. Marine Organism Detection and Classification from Underwater Vision Based on the Deep CNN Method. Math Probl Eng 2020, 2020, 1-11

35. Zimmerman, A.N.; Johnson, C.C.; Bussberg, N.W.; Dalkilic, M.M. Stability and Decline in Deep-Sea Coral Biodiversity, Gulf of Mexico and US West Atlantic. Coral Reefs 2020, 39, 345-359

36. Muhamed Ali, A.; Zhuang, H.; VanZwieten, J.; Ibrahim, A.K.; Cherubin, L. A Deep Learning Model for Forecasting Velocity Structures of the Loop Current System in the Gulf of Mexico. Forecasting 2021,3, 934-953

37. Stanchev, L.; Egbert, H.; Ruttenberg, B. Automating Deep-Sea Video Annotation Using Machine Learning. In Proceedings of the 2020 IEEE 14th International Conference on Semantic Computing (ICSC); IEEE, February 2020; pp. 17-24.

38. Schwing, F.B. Modern Technologies and Integrated Observing Systems Are "Instrumental" to Fisheries Oceanography: A Brief History of Ocean Data Collection. Fish Oceanogr 2023, 32, 28-69

39. Oliveira e Silva, L.; Resende, M.; Galhardas, H.; Manquinho, V.; Lynce, I. DeepData: Machine Learning in the Marine Ecosystems. Expert Syst Appl 2022, 206, 117841

40. Trice, A.; Robbins, C.; Philip, N.; Rumsey, M. Challenges and Opportunities for Ocean Data to Advance Conservation and Management. Ocean Conservancy, Washington DC2021.

Размещено на Allbest.Ru