The study uses machine learning to characterise the decrease in pH, a key factor in assessing the impacts of climate change on marine biodiversity.
TDB keeps you informed. Follow us on Facebook, Twitter and Instagram.
An interdisciplinary team from the Consejo Superior de Investigaciones Científicas (CSIC) in the Balearic Islands has presented the first acidification rate in the coastal area of the Balearic Sea to elucidate the consequences of climate change in coastal areas of the archipelago. The aim of the study has focused on reconstructing incomplete time series of relevant pH through the use of artificial intelligence techniques.
The results, published in the journal Scientific Reports, indicate that these coastal areas show a trend of pH decrease (acidification) of 0.0020±0.00054 pH units per year. This trend is similar to that observed in other global ocean basins and is mainly due to the incorporation of atmospheric carbon dioxide into seawater and increasing temperature.
“The decrease in seawater pH is due to the increase of carbon dioxide in the atmosphere and results in major alterations with a large impact on marine ecosystems. For example, ocean acidification leads to reduced saturation levels of carbonate minerals, which increases difficulties in shell formation for calcifying marine organisms (plankton, molluscs, echinoderms and corals). Measuring how pH is changing in these areas is, therefore, key to characterising the problem,” explains Iris E. Hendriks, principal investigator of the project who works at the Mediterranean Institute for Advanced Studies (IMEDEA-CSIC-UIB).
“Our work is a valuable contribution to understanding the role of coastal areas and the effects of climate change on the ecosystems present in them,” she says.
Artificial intelligence for data reconstruction
The study has been a major operational effort that began in 2018 with the collection of pH data, along with other variables (water temperature, salinity and dissolved oxygen levels), at the monitoring stations of the Balearic Ocean Acidification Time Series (BOATS) network in the Bay of Palma and in the maritime-terrestrial National Park of the Cabrera archipelago, within the framework of the CSIC Water:iOS Interdisciplinary Thematic Platform.
“The maintenance of this type of stations entails difficulties – financial costs, meteorological risks, deployment in areas of high shipping traffic, instrumental failures, etc. – which imply gaps in the data and, therefore, a loss of quality when it comes to preparing global studies,” Hendriks points out.
In order to complete the data and estimate the pH series over a large time interval prior to monitoring, the team applied deep learning techniques, an emerging area of machine learning that has recently made substantial advances in the field of Artificial Intelligence. Specifically, several recurrent neural network models were developed which, in their training, allowed them to relate the pH series to the set of environmental variables obtained, predicting the pH value when it is not available.
The work of obtaining a large amount of data and the subsequent application of these techniques has made it possible to reconstruct the decadal trend of acidification of the Balearic Sea, which is the main result of the work.
The IMEDEA-CSIC-UIB, the Andalusian Institute of Marine Sciences (ICMAN-CSIC), the Institute of Interdisciplinary Physics and Complex Systems (IFISC-CSIC-UIB), the Balearic Islands Coastal Observation and Forecasting System (ICTS SOCIB), and the Institute of Marine Research (IIM-CSIC) have participated in the study. The management team of the Cabrera Archipelago Maritime-Terrestrial National Park and the Regional Ministry of the Environment and Territory have collaborated in the project. It has also been funded by the Ministry of Science and Innovation, the Government of the Balearic Islands and the BBVA Foundation.