5:15 PM - 6:30 PM
[ACG41-P05] Linear and non-linear analysis for high temperature and acidification event in the Great Barrier Reef based on the observation in the Coral Sea
Keywords:Great Barrier Reef, high temperature process, Acidification
For predicting the health of coral reefs in the future, it is important to evaluate the impact of global fluctuations caused by the open ocean as well as the effect from the land area. For example, the widespread bleaching of the Great Barrier Reef (GBR) in 2016-2017 was attributed to the high temperature of seawater associated with El Nino event, but past El Nino events and GBR bleaching had not been necessarily linked. Therefore, the detailed analysis is required for future prediction of global fluctuation effects on coral reefs. In addition, the decrease in aragonite saturation state (Ω) due to the ocean acidification will lead to the inhibition of coral growth, but the study about the quantitative evaluation of Ω in coral reefs and the combined process with the high temperature event have not been enough at present.
In this study, we used the data in the Coral Sea, which corresponds to the oceanic endmember of GBR, from 2006 to 2018 observed by the National Institute for Environmental Studies (NIES) as the Volunteer Observation Ship program to identify the high temperature event and the Ω fluctuations in the GBR. In the analysis, the nonlinear analysis by machine learning (Random Forest and Gaussian Process) was applied to construct a regression model of the objective variables and extract the important explanatory variables. We used the Degree Heating Week (DHW; obtained from NOAA Coral Reef Watch), which is a high correlation index with the coral bleaching, and the Dissolved Inorganic Carbon concentration normalized to salinity of 35 (nDIC; calculated from the observation data of NIES), which is the main caution of the fluctuation of Ω as the objective variable of the high temperature event and the acidification, respectively.
The results of the regression model by Random Forest analysis are shown in Fig. 1. For the high temperature event, the atmospheric pressure on the Coral Sea was extracted as an important parameter in addition to the Southern Oscillation Index (SOI). This indicates that there are several meteorological patterns in the GBR during El Nino event, and that certain patterns, such as the increase in the amount of solar radiation, might cause the high temperature event in the GBR. It was also clarified that the fluctuation of Ω was strongly influenced by the sea surface temperature and the concentration of atmospheric CO2. However, the correlation with the former parameter is considered to be apparently significant because both the sea surface temperature and nDIC in the Coral Sea have increased by the recent global warming and acidification. No significant correlation was confirmed between the objective variables DHW and nDIC. This suggests that the high temperature and acidification in the GBR are independent phenomena each other. In the presentation, we will show the results of the linear regression analysis and will show more detailed analysis using the improved regression models.
In this study, we used the data in the Coral Sea, which corresponds to the oceanic endmember of GBR, from 2006 to 2018 observed by the National Institute for Environmental Studies (NIES) as the Volunteer Observation Ship program to identify the high temperature event and the Ω fluctuations in the GBR. In the analysis, the nonlinear analysis by machine learning (Random Forest and Gaussian Process) was applied to construct a regression model of the objective variables and extract the important explanatory variables. We used the Degree Heating Week (DHW; obtained from NOAA Coral Reef Watch), which is a high correlation index with the coral bleaching, and the Dissolved Inorganic Carbon concentration normalized to salinity of 35 (nDIC; calculated from the observation data of NIES), which is the main caution of the fluctuation of Ω as the objective variable of the high temperature event and the acidification, respectively.
The results of the regression model by Random Forest analysis are shown in Fig. 1. For the high temperature event, the atmospheric pressure on the Coral Sea was extracted as an important parameter in addition to the Southern Oscillation Index (SOI). This indicates that there are several meteorological patterns in the GBR during El Nino event, and that certain patterns, such as the increase in the amount of solar radiation, might cause the high temperature event in the GBR. It was also clarified that the fluctuation of Ω was strongly influenced by the sea surface temperature and the concentration of atmospheric CO2. However, the correlation with the former parameter is considered to be apparently significant because both the sea surface temperature and nDIC in the Coral Sea have increased by the recent global warming and acidification. No significant correlation was confirmed between the objective variables DHW and nDIC. This suggests that the high temperature and acidification in the GBR are independent phenomena each other. In the presentation, we will show the results of the linear regression analysis and will show more detailed analysis using the improved regression models.