A comprehensive study of the parameters affecting the stable isotopes in the precipitation of the Bangkok metropolitan area using model-based statistical approaches

ABSTRACT

This study determined the main local and regional parameters affecting the stable isotopes (¹⁸O, ²H) in the Bangkok precipitation and developed the Bangkok meteoric water line (BMWL) (δ²H = (7.68 ± 0.07) δ¹⁸O + (7.25 ± 0.48)). First, Pearson correlation coefficients were used to determine the correlation between local and regional parameters. Six different regression methods were used based on Pearson correlation coefficients. The stepwise regression had the most accurate performance among them according to the R² values. Second, three different methods were used to develop the BMWL, and their performances were also studied. Third, the stepwise regression method was used to study the effects of local and regional parameters on the stable isotope content of precipitation. The results showed that the local parameters had a greater effect on the stable isotope content than the regional ones. The stepwise models developed based on the northeast and southwest monsoons showed that moisture sources also affected the stable isotope content of precipitation. Finally, the developed stepwise models were validated by calculating the root mean square error (RMSE) and R². This study demonstrated that the local parameters mainly controlled the stable isotopes in the Bangkok precipitation, while the regional parameters had a slight effect on them.

KEYWORDS:

• Bangkok
• stepwise model
• precipitation
• stable isotopes
• local and regional parameters
• regression models
• Pearson correlation coefficients
• validation

Acknowledgments

The authors gratefully acknowledge the Global Network of Isotopes in Precipitation (GNIP) for providing the rainfall isotope data. We also thank the National Oceanic and Atmospheric Administration (NOAA) (US Department of Commerce) for providing us with the dataset of teleconnection indices. The authors would also like to express their appreciation to Dr. Supaporn Buajan for her support during this study.

Data tools

All statistical data analyses were done using the R software (version: 4.1.3). The required R packages were ‘devtools’, ‘tidyverse’, ‘corrplot’, ‘caret’, ‘leaps’, ‘MASS’, ‘olsrr’, ‘GGally’, ‘glmnet’, ‘Metrics’, ‘dplyr’, ‘pls’, ‘lattice’, ‘quantreg’, and ‘ggplot2’. The codes needed for data processing are available in Supplementary Files 1–3.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The first author received a grant for his postdoctoral fellowship [grant number MU-PD-2021-13] from the Faculty of Environment and Resource Studies, Mahidol University.