Science
Modelling
Models transform governed evidence into testable predictions and decision indicators.
Scientific question
Can models improve groundwater decisions while remaining reproducible, uncertainty-aware and transferable across data-sparse settings?
Methods and evidence
Register model versions, inputs, training/validation partitions, metrics, spatial applicability and reproducible software environments.
Expected scientific outputs
- Groundwater-potential estimates
- Yield ranges
- Decision indicators
- Model registry
Limits and safeguards
Models are constrained by training data and domain shift. No model should issue an unconditional “drill here” instruction.