Inferring Soil Drydown Behaviour with Adaptive Bayesian Online Changepoint Analysis

Download

• arXiv

Abstract

Continuous soil-moisture measurements provide a direct lens on subsurface hydrological processes, notably the post-rainfall “drydown” phase. Because these records consist of distinct, segment-specific behaviours whose forms and scales vary over time, realistic inference demands a model that captures piecewise dynamics while accommodating parameters that are unknown a priori. Building on Bayesian Online Changepoint Detection (BOCPD), we introduce two complementary extensions: a particle-filter variant that substitutes exact marginalisation with sequential Monte Carlo to enable real-time inference when critical parameters cannot be integrated out analytically, and an online-gradient variant that embeds stochastic gradient updates within BOCPD to learn application-relevant parameters on the fly without prohibitive computational cost. After validating both algorithms on synthetic data that replicate the temporal structure of field observations-detailing hyperparameter choices, priors, and cost-saving strategies-we apply them to soil-moisture series from experimental sites in Austria and the United States, quantifying site-specific drydown rates and demonstrating the advantages of our adaptive framework over static models.

Citation

Gong, M., Nemeth, C., Killick, R., Strauss, P. and Quinton, J. (2025). Inferring Soil Drydown Behaviour with Adaptive Bayesian Online Changepoint Analysis. arXiv preprint.

@article{gong2025inferring,
  title={Inferring Soil Drydown Behaviour with Adaptive Bayesian Online Changepoint Analysis},
  author={Gong, Mengyi and Nemeth, Christopher and Killick, Rebecca and Strauss, Peter and Quinton, John},
  journal={arXiv preprint arXiv:2509.13293},
  year={2025}
}