Summary
This study evaluates Random Forest, a machine learning ensemble algorithm, as an alternative to process-based models for estimating root zone soil moisture at daily timescales in agricultural catchments. Random Forest demonstrated comparable or slightly superior performance to process-based modelling with data assimilation, whilst offering advantages in data-poor regions where soil hydraulic parameters are incomplete. However, the approach showed reduced accuracy during extreme wet and dry conditions, attributed to infrequent sampling and incomplete subsurface process representation in the model covariates.
UK applicability
The findings may be applicable to UK agricultural water management contexts, particularly where detailed soil hydraulic characterisation is limited. However, the study's small catchment scale and specific climatic/soil conditions require site-specific validation before adoption in British farming or hydrological practice.
Key measures
Root zone soil moisture (RZSM) predictions; interpolation and extrapolation accuracy; model performance during extreme moisture conditions; comparison with process-based modelling approaches
Outcomes reported
The study compared Random Forest machine learning predictions of root zone soil moisture against process-based model simulations with data assimilation over a small agricultural catchment (2016–2018). Random Forest achieved slightly higher accuracy for interpolation and similar accuracy for extrapolation, though both approaches performed less accurately during 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.