Using SWAP to quantify space and time related uncertainty in deep drainage model estimates: A case study from northern NSW, Australia

• Deep drainage depending on the occurrence of heavy rainfall relative to land use.
• Variability in rainfall was the dominant factor that influenced uncertainty.
• Variations in soil hydraulic properties had least impact on deep drainage.
• The temporal variability introduced more uncertainty than spatial variability.
• To improve deep drainage predictions, more accurate rainfall data is needed.

Deep drainage can contribute to groundwater table rises and salinity, and is a complex function of rainfall, land management and soil hydraulic properties. Each of these components is uncertain and variable in space and time. This study quantifies the associated uncertainty using a Monte Carlo simulation to calculate deep drainage and estimate deep drainage risk. The 1-D soil water model SWAP was used with multiple realisations of rainfall, land use and soil hydraulic properties over 25 years in northern NSW, Australia. The results confirm that deep drainage is episodic with high monthly variability, depending on the occurrence of heavy rainfall relative to land use. Uncertainty about the spatial and temporal variation in local rainfall was the dominant factor that influenced uncertainty in deep drainage predictions, followed by uncertainty in land use changes and soil hydraulic properties. Uncertainty in soil hydraulic properties had less impact because specific land uses tend to align with soil types. The uncertainty related to the temporal variability in input parameters introduced more uncertainty than the spatial variability. To improve deep drainage predictions, more accurate rainfall data in space and time is needed, as well as data on the temporal and spatial variability of crop rotations.