Summary
This study demonstrates the application of Random Forest machine learning to estimate root zone soil moisture at daily resolution in a small agricultural catchment, comparing performance against conventional process-based hydrological models. The findings show that data-driven approaches offer practical advantages in regions with incomplete soil hydraulic parameter data, though limitations exist in capturing extreme moisture conditions due to sparse training data. The work suggests machine learning presents a viable alternative to process-based models where the primary objective is soil moisture state estimation rather than mechanistic understanding of subsurface processes.
UK applicability
The methodology could be applicable to UK agricultural water management and catchment studies, particularly where soil hydraulic parameters are poorly characterised. However, the study was conducted over a single small catchment with specific climate and soil conditions; validation across diverse UK soil types, climates and scales would be necessary before operational deployment.
Key measures
Root zone soil moisture estimation accuracy (interpolation and extrapolation); comparison of machine learning versus process-based model predictions; performance on extreme moisture events
Outcomes reported
The study evaluated Random Forest algorithm performance for daily-scale root zone soil moisture estimation and compared predictions against a process-based hydrological model with data assimilation. Random Forest achieved slightly higher accuracy for interpolation and similar accuracy for extrapolation, though performance was reduced for extreme wet and dry conditions.
Topic tags
Dig deeper with Pulse AI.
Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.