pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning
Abstract The decreasing seawater pH trend associated with increasing atmospheric carbon dioxide levels is an issue of concern due to possible negative consequences for marine organisms, especially calcifiers. Globally, coastal areas represent important transitional land-ocean zones with complex inte...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2022-07-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-022-17253-5 |
| _version_ | 1811222907653718016 |
|---|---|
| author | Susana Flecha Àlex Giménez-Romero Joaquín Tintoré Fiz F. Pérez Eva Alou-Font Manuel A. Matías Iris E. Hendriks |
| author_facet | Susana Flecha Àlex Giménez-Romero Joaquín Tintoré Fiz F. Pérez Eva Alou-Font Manuel A. Matías Iris E. Hendriks |
| author_sort | Susana Flecha |
| collection | DOAJ |
| description | Abstract The decreasing seawater pH trend associated with increasing atmospheric carbon dioxide levels is an issue of concern due to possible negative consequences for marine organisms, especially calcifiers. Globally, coastal areas represent important transitional land-ocean zones with complex interactions between biological, physical and chemical processes. Here, we evaluated the pH variability at two sites in the coastal area of the Balearic Sea (Western Mediterranean). High resolution pH data along with temperature, salinity, and also dissolved oxygen were obtained with autonomous sensors from 2018 to 2021 in order to determine the temporal pH variability and the principal drivers involved. By using environmental datasets of temperature, salinity and dissolved oxygen, Recurrent Neural Networks were trained to predict pH and fill data gaps. Longer environmental time series (2012–2021) were used to obtain the pH trend using reconstructed data. The best predictions show a rate of $$-\,0.0020\pm 0.00054$$ - 0.0020 ± 0.00054 pH units year $$^{-1}$$ - 1 , which is in good agreement with other observations of pH rates in coastal areas. The methodology presented here opens the possibility to obtain pH trends when only limited pH observations are available, if other variables are accessible. Potentially, this could be a way to reliably fill the unavoidable gaps present in time series data provided by sensors. |
| first_indexed | 2024-04-12T08:22:49Z |
| format | Article |
| id | doaj.art-1b55ae89e956487d848e84da3df1d5f4 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-04-12T08:22:49Z |
| publishDate | 2022-07-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-1b55ae89e956487d848e84da3df1d5f42022-12-22T03:40:29ZengNature PortfolioScientific Reports2045-23222022-07-0112111110.1038/s41598-022-17253-5pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learningSusana Flecha0Àlex Giménez-Romero1Joaquín Tintoré2Fiz F. Pérez3Eva Alou-Font4Manuel A. Matías5Iris E. Hendriks6Instituto de Ciencias Marinas de Andalucía (ICMAN-CSIC)Instituto de Física Interdisciplinar y Sistemas Complejos, (IFISC-UIB-CSIC), Campus UIBInstituto Mediterráneo de Estudios Avanzados (IMEDEA-UIB-CSIC)Instituto de Investigaciones Marinas (IIM-CSIC)Balearic Islands Coastal Observing and Forecasting System (SOCIB)Instituto de Física Interdisciplinar y Sistemas Complejos, (IFISC-UIB-CSIC), Campus UIBInstituto Mediterráneo de Estudios Avanzados (IMEDEA-UIB-CSIC)Abstract The decreasing seawater pH trend associated with increasing atmospheric carbon dioxide levels is an issue of concern due to possible negative consequences for marine organisms, especially calcifiers. Globally, coastal areas represent important transitional land-ocean zones with complex interactions between biological, physical and chemical processes. Here, we evaluated the pH variability at two sites in the coastal area of the Balearic Sea (Western Mediterranean). High resolution pH data along with temperature, salinity, and also dissolved oxygen were obtained with autonomous sensors from 2018 to 2021 in order to determine the temporal pH variability and the principal drivers involved. By using environmental datasets of temperature, salinity and dissolved oxygen, Recurrent Neural Networks were trained to predict pH and fill data gaps. Longer environmental time series (2012–2021) were used to obtain the pH trend using reconstructed data. The best predictions show a rate of $$-\,0.0020\pm 0.00054$$ - 0.0020 ± 0.00054 pH units year $$^{-1}$$ - 1 , which is in good agreement with other observations of pH rates in coastal areas. The methodology presented here opens the possibility to obtain pH trends when only limited pH observations are available, if other variables are accessible. Potentially, this could be a way to reliably fill the unavoidable gaps present in time series data provided by sensors.https://doi.org/10.1038/s41598-022-17253-5 |
| spellingShingle | Susana Flecha Àlex Giménez-Romero Joaquín Tintoré Fiz F. Pérez Eva Alou-Font Manuel A. Matías Iris E. Hendriks pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning Scientific Reports |
| title | pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning |
| title_full | pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning |
| title_fullStr | pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning |
| title_full_unstemmed | pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning |
| title_short | pH trends and seasonal cycle in the coastal Balearic Sea reconstructed through machine learning |
| title_sort | ph trends and seasonal cycle in the coastal balearic sea reconstructed through machine learning |
| url | https://doi.org/10.1038/s41598-022-17253-5 |
| work_keys_str_mv | AT susanaflecha phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning AT alexgimenezromero phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning AT joaquintintore phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning AT fizfperez phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning AT evaaloufont phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning AT manuelamatias phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning AT irisehendriks phtrendsandseasonalcycleinthecoastalbalearicseareconstructedthroughmachinelearning |