https://doi.org/10.65770/MLDT4430
ABSTRACT
Campaign hydrography in glacier-fed bays is sparse, making prediction at unsampled stations and dates difficult. We evaluated spatial and temporal transfer and physical constraints using 1,982 quality-controlled YSI EXO profiles from 24 Admiralty Bay stations, Antarctica (2018–2022). A graph-attention model with GRU-D-style temporal decay predicted temperature and salinity anomalies; a residual multilayer perceptron added static-stability and temperature–salinity (T–S) hull penalties. We compared five validation schemes with climatology, a spatial generalised additive model (GAM), gradient boosting, inverse-distance weighting (IDW), and persistence. In leave-one-site-out tests with same-day neighbours, IDW achieved a temperature root-mean-square error (RMSE) of 0.110 °C, versus 0.249 °C for climatology. In buffered leave-one-day-out tests (±7-day buffer), the lagged graph model achieved skill scores of 0.45 for temperature and 0.32 for salinity; graph parameters were fixed at leave-one-site-out medians. The apparent advantage of gradient boosting under station withholding disappeared when same-day data were excluded (temperature skill = −0.14), revealing campaign-identifier leakage. On a 12-station subset, increasing the stability-penalty weight from 0 to 3 reduced the share of profiles with density inversions from 37% to 18%, close to the base model’s 20%, while increasing temperature RMSE by 0.004 °C. A summer T–S hull excluded 56% of same-day IDW predictions for June–November. Conformal coverage was near nominal across stations but deteriorated under seasonal shift for several models; same-day IDW remained near nominal. The stability penalty improved plausibility relative to the unconstrained residual model but not accuracy; the hull penalty had little effect on leave-one-site-out accuracy. Recent neighbouring profiles contained most of the transferable information.
References
- [1] Angelopoulos, A. N., & Bates, S. (2023). Conformal prediction: A gentle introduction. Foundations and Trends in Machine Learning, 16(4), 494–591. https://doi.org/10.1561/2200000101
- [2] Barber, R. F., Candès, E. J., Ramdas, A., & Tibshirani, R. J. (2021). Predictive inference with the jackknife+. The Annals of Statistics, 49(1), 486–507. https://doi.org/10.1214/20-AOS1965
- [3] Battaglia, P. W., Hamrick, J. B., Bapst, V., Sanchez-Gonzalez, A., Zambaldi, V., Malinowski, M., Tacchetti, A., Raposo, D., Santoro, A., Faulkner, R., Gulcehre, C., Song, F., Ballard, A., Gilmer, J., Dahl, G. E., Vaswani, A., Allen, K., Nash, C., Langston, V., . . . Pascanu, R. (2018). Relational inductive biases, deep learning, and graph networks [Preprint]. arXiv. https://doi.org/10.48550/arXiv.1806.01261
- [4] Beucler, T., Pritchard, M., Rasp, S., Ott, J., Baldi, P., & Gentine, P. (2021). Enforcing analytic constraints in neural networks emulating physical systems. Physical Review Letters, 126, 098302. https://doi.org/10.1103/PhysRevLett.126.098302
- [5] Bolibar, J., Rabatel, A., Gouttevin, I., Galiez, C., Condom, T., & Sauquet, E. (2020). Deep learning applied to glacier evolution modelling. The Cryosphere, 14, 565–584. https://doi.org/10.5194/tc-14-565-2020
- [6] Che, Z., Purushotham, S., Cho, K., Sontag, D., & Liu, Y. (2018). Recurrent neural networks for multivariate time series with missing values. Scientific Reports, 8, Article 6085. https://doi.org/10.1038/s41598-018-24271-9
- [7] Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29(5), 1189–1232. https://doi.org/10.1214/aos/1013203451
- [8] Gade, H. G. (1979). Melting of ice in sea water: A primitive model with application to the Antarctic ice shelf and icebergs. Journal of Physical Oceanography, 9(1), 189–198. https://doi.org/10.1175/1520-0485(1979)009<0189:MOIISW>2.0.CO;2
- [9] Gneiting, T., & Raftery, A. E. (2007). Strictly proper scoring rules, prediction, and estimation. Journal of the American Statistical Association, 102(477), 359–378. https://doi.org/10.1198/016214506000001437
- Holmberg, D., Oskarsson, J., Wikingsson, E., Lindsten, F., & Roos, T. (2026). Njord: A probabilistic graph neural network for ensemble ocean forecasting [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2605.15470
- Howard, C. C., & Apeli, A. (2026). Hybrid CNN–LSTM–attention forecasting of toxic contaminants in produced water: Evidence from Niger Delta flow stations. World Scientific News, 219, 21–39. https://doi.org/10.65770/ZIWD5587
- Howard, C. C., & Howard, I. C. (2026). Hybrid neural network approaches to produced water quality forecasting and management in Nigerian oil fields. International Journal of Mathematics, Statistics, and Computer Science, 4, 527–543. https://doi.org/10.59543/7fjxt923
- Intergovernmental Oceanographic Commission, Scientific Committee on Oceanic Research, & International Association for the Physical Sciences of the Oceans. (2010). The international thermodynamic equation of seawater–2010: Calculation and use of thermodynamic properties (IOC Manuals and Guides No. 56). UNESCO. https://doi.org/10.25607/OBP-1338
- Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S., & Yang, L. (2021). Physics-informed machine learning. Nature Reviews Physics, 3, 422–440. https://doi.org/10.1038/s42254-021-00314-5
- Lam, R., Sanchez-Gonzalez, A., Willson, M., Wirnsberger, P., Fortunato, M., Alet, F., Ravuri, S., Ewalds, T., Eaton-Rosen, Z., Hu, W., Merose, A., Hoyer, S., Holland, G., Vinyals, O., Stott, J., Pritzel, A., Mohamed, S., & Battaglia, P. (2023). Learning skillful medium-range global weather forecasting. Science, 382(6677), 1416–1421. https://doi.org/10.1126/science.adi2336
- Maciel-Seidman, M. L., Ryan, J. C., Esenther, S. E., Rennermalm, Å. K., Smith, L. C., & Fettweis, X. (2026). A transfer learning approach for simulation of Greenland Ice Sheet meltwater runoff [Preprint]. ESS Open Archive. https://doi.org/10.22541/essoar.15008734/v1
- Mas e Braga, M., Cowton, T., Slater, D., Inall, M., Johnstone, E., & Fraser, N. (2025). Controls on fjord temperature throughout Greenland in a reduced-physics model. Geophysical Research Letters, 52(17), e2025GL116902. https://doi.org/10.1029/2025GL116902
- Meyer, H., & Pebesma, E. (2021). Predicting into unknown space? Estimating the area of applicability of spatial prediction models. Methods in Ecology and Evolution, 12(9), 1620–1633. https://doi.org/10.1111/2041-210X.13650
- Millero, F. J., & Poisson, A. (1981). International one-atmosphere equation of state of seawater. Deep-Sea Research Part A: Oceanographic Research Papers, 28(6), 625–629. https://doi.org/10.1016/0198-0149(81)90122-9
Download all article in PDF
![]()



