Capstone on the machine learning (ML) use for the prediction of the anthropogenic carbon in the ocean
Tobia Tudino (TTudino)
This project provides an initial exploration of the machine learning potentialities to predict the oceanic anthropogenic carbon. This aims to:
- find an alternative approach to the currently used methods,
- reduce the uncertainty in the anthropogenic carbon estimates (currently at approximately 20%).
[1] Friis, K., Körtzinger, A., Pätsch, J., and Wallace, D.W.R.: On the temporal increase of anthropogenic CO2 in the subpolar North Atlantic, Deep Sea Res. I, 52, 681-698, 2005.
[2] Gruber, N., Sarmiento, J.L., and Stocker, T.F.: An improved method for detecting anthropogenic CO2 in the oceans, Glob. Biogeochem. Cycles, 10, 809-837, 1996.
[3] Khatiwala, S., Primeau, F., and Hall, T.: Reconstruction of the history of anthropogenic CO2 concentrations in the ocean, Nature, 462, 346-349, 2009.
[4] S.K. Lauvset et al. A new global interior ocean mapped climatology: the 1◦ x 1◦ GLODAP version 2. Earth Syst. Sci. Data, 8, 2016. doi: 10.5194/essd-8-325-2016.
[5] Redfield, A.C.: On the proportions of organic derivations in seawater and their relation to the composition of Plankton. In J. Johnstone memorial, Liverpool University press, 176-192, 1934.
[6] Sabine, C.L., Feely, R.A., Gruber, N., Key, R.M., Lee, K., Bullister, J.L., Wanninkhof, R., Wong, C.S., Wallace, D.W.R., Tillbrock, B., Millero, F.J., Peng, T.-H., Kozyr, A., Ono, T., and Rios, A.F.: The Oceanic Sink for Anthropogenic CO2, Science, 305, 367-371, 2004.
[7] Waugh, D.W., Haine, T.W.N., and Hall, T.M.: Transport times and anthropogenic carbon in the subpolar North Atlantic Ocean, Deep Sea Res. I, 51, 1475-1491, 2004.
[8] https://www.kdnuggets.com/2018/04/right-metric-evaluating-machine-learning-models-1.html
[9] http://www.mvstat.net/tduong/research/seminars/seminar-2001-05/
The GLODAPv2 climatology dataset is too large to be provided on Github. Please refer to https://www.glodap.info/ and the article [4] above for getting and use the data. Feel free to contact me if necessary.
The initial analysis is summarised at https://medium.com/@tudinot/machine-learning-estimates-of-oceanic-anthropogenic-carbon-cant-82b3dc9951ec.
Please read it through and feel free to ask/suggest any change.
The code is provided as jupyter notebook (ML_Cant_estimate.ipynb) and python code (ML_Cant_estimate.py). Adjust the existing paths to data where necessary.
The code is set to run on a small subset of data randomly extracted from the provided dataset. Remove this limitations if your machine is powerful enough to deal with a bigger dataset. The results improve.