Font Size: a A A

Global Ocean Carbon Source/Sink And Acidification Research Based On A Machine Learning Algorithm

Posted on:2024-03-21Degree:DoctorType:Dissertation
Country:ChinaCandidate:G R ZhongFull Text:PDF
GTID:1520307160958499Subject:Marine Chemistry
Abstract/Summary:PDF Full Text Request
With the increasing carbon dioxide emission since the Industrial Revolution,about a quarter of the carbon dioxide emitted by human activities has been absorbed by the oceans,resulting in a continuous rise in the global surface ocean partial pressure of carbon dioxide(pCO2)and a decrease in the sea water pH(ocean acidification).The strengthening global ocean carbon sink is an important driver of surface ocean acidification,but it is not clear whether the global ocean carbon sink always continues to strengthen with the increase of atmospheric CO2 concentration,and whether ocean acidification is accelerating.This is mainly caused by the lack of long-term,continuous and high coverage measurements of surface ocean pCO2 and sea water pH has severely limited our research on the evolution of global ocean carbon sink and ocean acidification.Although some studies have tried to use different machine learning algorithms to construct the gridded data of the global surface ocean pCO2,there are still many problems,leading to great uncertainty in the global ocean carbon sink estimated based on constructed pCO2 gridded data.In addition,there is currently no report of applying machine learning algorithm to the construction global seawater pH grid data from pH measurements.In order to obtain long-term,continuous and high coverage data supporting the study on global ocean acidification process during the last 30 years and its response to the changes in global ocean carbon sink,this study developed a new machine learning algorithm,called stepwise feed forward neural network algorithm,to construct the gridded data of the global surface ocean pCO2 and seawater pH at 0-2000 m depth.Based on the surface ocean pCO2 measurements from the SOCAT data set and seawater pH measurements from the GLODAP data set,this new machine learning algorithm was applied to select out the most closely related predictors from 73 possible parameters related to the change of pCO2 and pH,for fitting the nonlinear relationship between surface ocean pCO2,seawater pH,and their predictors,then the gridded data products was constructed.This study constructed the first set of global seawater pH gridded data product with a total of 41 layers at 0-2000 m depth and the global surface ocean pCO2gridded data product with the best accuracy at present.These two data products are both from 1992 to 2020,with a monthly time resolution and a 1°×1°spatial resolution,covering global open oceans.Based on the data products,this study analyzed the variability of global ocean carbon sink and seawater pH during the past 30 years,and the main conclusions are as follows:(1)During the past 30 years,the global ocean carbon sink has experienced a large fluctuation,and started continuously strengthening since 2008,reaching2.14±0.33 PgC yr-1 in 2020.The change in global ocean carbon sink during the past 30 years can be divided into four phases according to the trend.After significantly receding in the 1990s,the global ocean carbon sink has been strengthening rapidly since 2001,weakened again during 2003-2008,and then started restrengthening since 2008,reaching 2.14±0.33PgC yr-1 in 2020.The repeated weakening and strengthening were mainly caused by the Pacific Ocean carbon sink variability driven by the ENSO events and were not well captured by previous research.During El Nino events,deep water transported to the surface by upwelling decreases dramatically,leading to weakening equatorial Pacific carbon sources.Furthermore,during La Nina events,strengthening carbon sources in the equatorial Pacific can release more carbon dioxide into the atmosphere.The intensity and change trend of global ocean carbon sink after 2008 estimated using the constructed surface seawater pCO2 gridded data are well consistent with the results of the GCP model.Compared with other studies based on surface seawater pCO2,this study found that since the amount of pCO2 observed in the Southern Ocean in the SOCAT dataset in winter was almost less than a quarter of that in summer,previous studies using machine learning algorithms to estimate the oceanic carbon sink based on this dataset probably overestimated the intensity of the Southern Ocean carbon sink.After correcting for the effect of this uneven data distribution on grid data construction,this study found that the intensity of the Southern Ocean carbon sink was lower than without correction and more similar to the results of other past studies using Southern Ocean buoy data.The corrected Southern Ocean sink shows a similar short-term change,but the long-term change shows little increase over the past 30 years,which is different with previous understanding.(2)The global surface seawater pH has decreased by 0.0015 per year on average during the past 30 years,where the Arctic Ocean experienced the fastest surface ocean acidification,the Pacific Ocean and Atlantic Ocean the slowest.Although the global ocean carbon sink was overall strengthening,there is no obvious accelerating trend of acidification in the global surface ocean.Although the acidification rate of global ocean surface seawater has significantly fluctuated in the last 30 years,there has been no significant acceleration trend of global ocean surface seawater acidification in the last 30 years.Based on the global ocean carbon source sink estimated by surface pCO2,it is found that the carbon source sink intensity is the most critical factor affecting the acidification rate of surface seawater in the south temperate Pacific,the subpolar North Pacific,and the entire Indian Ocean,and the average carbon sink intensity and the average acidification rate of surface seawater show a strong positive correlation during the past 30 years.In the equatorial Pacific,however,there is a very high correlation between the short-term inter-annual variation of surface pH and the El Nino index.During two intense El Nio events,1997-1998 and 2015-2016,the pH of surface seawater in the equatorial Pacific increased by about 0.01 quickly.However,the influence of upwelling intensity changes on the acidification rate of surface seawater was more short-term but had little influence on the long-term trend of the acidification rate.In the northern temperate Pacific,although there is a significant east-west difference in the intensity of the carbon sink,and the east-west difference has been increasing continuously during the past 30 years,the east-west difference in the acidification rate of the surface seawater in the northern temperate Pacific is not significant.In these regions where the correlation between the intensity of the carbon sink and the acidification rate of the surface seawater is low,the biological pump also affects the surface seawater’s acidification rate.More carbon dioxide absorbed by surface waters is converted into particulate organic carbon for transport to the deep ocean.(3)It is first found that acidification also exists in deep seawater.In different oceans,the average acidification rate of seawater at the bottom of the mixed layer is faster than surface seawater.From the mixed layer depth to 2000m,the overall acidification rate gradually slows with the increasing depth.During the past 30years,acidification in the Indian Ocean at 140-300m and Pacific Ocean at 650-750m has been accelerating.However,it has not accelerated in the other ocean from the surface to the 2000m depth.During the past 30 years,ocean acidification has been extended to the intermediate ocean at 2000m depth.The average pH of the global ocean in the range of 0-2000m has been decreasing continuously.The average acidification rate of seawater at a depth of160m is faster than that of the surface ocean,with an average annual decrease of 0.0016units per year.However,in the range of 160-2000m,the average acidification rate gradually slows with increasing depth,and the average acidification rate is only 0.0003yr-1 at 2000m.Meanwhile,compared with the trend of pH data corrected to 25℃,the direct impact of seawater warming on acidification during the past 30 years is about0.0001-0.0003 yr-1,and the deeper the depth,the less the direct impact of seawater warming on acidification.Except for the Arctic Ocean,where the surface acidification rate is the fastest,in different regions,we found that the acidification rate of the subsurface is faster than that of the surface.Moreover,the depth of the fastest acidification rate is very close to the maximal mixing layer depth in most regions,which reveals that the most rapid acidification rate of the global ocean has happened at the bottom of the mixing layer in the recent 30 years.The increase in the remineralization of organic matter may mainly be attributed.By analyzing the long-term trend of annual pH growth rate in different depths of oceans,it is found that acidification in the Indian Ocean at 140-300m and the Pacific Ocean at 650-750m has been accelerating during the past 30 years.However,no similar phenomenon has been found in other oceans.In addition,acidification in the Atlantic Ocean below 1000m was slowing down during the past 30 years.Although seawater acidification is accelerating in some regions,the acidification rate remains unchanged or slows down in most regions.Overall,the change of global average pH in the range of 0-2000m does not show a significant accelerating acidification.
Keywords/Search Tags:Ocean acidification, Carbon sink, Seawater pH, Machine learning, Surface ocean pCO2
PDF Full Text Request
Related items