Modelling tool capability overview

A set of commonly used modelling tools in South Africa was reviewed for the WRC “Critical catchment model inter-comparison and model use guidance development” project. This set included the major tools developed in South Africa, ACRU and the Pitman model-based tools (WRSM-Pitman and SPATSIM-Pitman), as well as two tools that were developed in the northern hemisphere, but have been used globally, SWAT and MIKE-SHE. Locally developed modelling tools can have certain advantages from being designed with the South African context in mind, both in terms of data availability and local climate characteristics, ecosystems, soils and geologic types, as well as land and water management practices. SWAT and MIKE-SHE have resourced development teams behind them that continually update the tools and adapt them to make use of developing globally available data sources, such as remotely sensed data, and generally improved access to greater computing power. This suite of tools covers a diversity of model structure and algorithm type options.

The two tables below summarise basic information about the tools in this set: the first gives intended uses and broad structural characteristics and the second gives an overview of modelling capabilities across the tools that are likely to be in demand for many typical use-cases.

Background & basic characteristics of reviewed modelling tools

Characteristic	WRSM-Pitman	SPATSIM-Pitman	ACRU	SWAT	MIKE-SHE
Developed in South Africa	yes	yes	yes	no	no
Current curator / developer	Bailey & Pitman Water Resources Ltd	Rhodes University, Institute of Water Resources (IWR)	University of KwaZulu Natal, Centre for Water Resources Research (UKZN-CWRR)	Texas A&M University & US Department of Agriculture (USDA)	Danish Hydrologic Institute (DHI)
Free access	yes	yes	yes	yes	no
Version reviewed	WRSM-Pitman version 2.9	SPATSIM GWv3 Global Options Threaded model	ACRU 4	SWAT & ArcSWAT 2012	MIKE-SHE & MIKE Hydro River, version 2017
Reference documents	Theory manual: (Bailey, 2015); User manual: (Bailey and Pitman, 2016)	Theory papers: (Hughes, 2004, 2013; Kapangaziwiri, 2007); User manual: (Hughes, 2019)	Theory manual: (Schulze, 1995); User manuals: (Clark et al., 2012; Schulze and Davis, 2018)	Theory manual: (Neitsch et al., 2011); User manuals: (Arnold et al., 2012)	Theory manuals:(DHI, 2017a, 2017b); User’s manuals:(DHI, 2017d, 2017c)
Intended spatial scale (catchment or model area)	Local to regional: no suggested min-max model size	Local to regional: 10-10,000’s of km2, more typical: 100-1,000’s km2	Field to regional: no suggested min-max model size	Field to regional: no suggested min-max model size	Field to regional: no suggested min-max model size
Spatial discretisation	Modules ('runoff' modules/subcatchments, special sub-areas, channels, reservoirs) linked by routes	Subcatchments + limited internal sub-area types	HRUs within subcatchments	HRUs within subcatchments	Fully distributed (gridded) OR combinations of grids and zones for different process calculations within subcatchments (if all process zones align, would act like HRUs)
	Intended subcat size < 1,000 km2		Intended subcat size 5-50 km2; HRU size < 30km2
Timestep	Monthly*	Monthly*	Daily	Daily, sub-daily	Daily, sub-daily (dynamic timestep length, can vary across processes)
Intended modelling applications (as documented):
Water balance estimation	yes	yes	yes	yes	yes
Design hydrology (flood peaks)			yes	yes	yes
Supply planning (general)	yes	yes	yes	yes	yes
Reservoir yield	yes	yes	yes	yes	yes
Irrigation planning	yes		yes	yes	yes
Groundwater recharge	yes	yes	yes	yes	yes
Groundwater-surface water (GW-SW) interactions & pumping impacts	yes	yes			yes
Land cover change impacts	yes	yes	yes	yes	yes
Climate change impacts	yes	yes	yes	yes	yes
Application limitations (as documented)	Not for peak flow, flood assessment, or design hydrology	Not for peak flow, flood assessment, design hydrology	Not represent deep GW processes - not for GW pumping impact	Not represent deep GW processes	(None listed for the modelling system as whole, only for certain process representation options.)
Specific development focuses particular to tool	Flexible network for tracking managed system transfers, GW-SW interaction, IAP & plantation forestry water use	Parsimony, Uncertainty assessment, GW-SW interactions	Land cover type representation, Crop & irrigation detail, IAP & plantation forestry water use	Land cover type representation, Crop & irrigation detail, Coupling to GIS tools	Spatial discretisation & fine scale processes, GW-SW interaction, Coupled hydraulic channel model with overbank flood process representation

The table below gives an overview of a range of capabilities of the modelling tools under review. This summary is supported by more detailed comparison of the structural and process algorithm differences between the modelling tools. Comparing the capabilities of a modelling tool to the needs of a modelling exercise is a critical step in tool selection. When a modelling tool lacks a capability that would be desirable, but not critical for the exercise, there may be ways to adapt the model set-up to account for this, such as adjusting parameter values for implicit representation of a feature or process the tool does not include explicitly.

Modelling tool capabilities overview

Capability	WRSM-Pitman	SPATSIM-Pitman	ACRU	SWAT	MIKE-SHE
Climate (rain & ET demand)
Spatially variable across model domain	yes	yes	yes	yes	yes
Spatially variable within subcatchment	(limited)	no	yes	no	yes
Inter-annual variability in ET demand	no	yes	yes	yes	yes
Land cover & change
Processes explicitly linked to land cover	(limited)	(limited)	yes	yes	yes
Multiple land cover types included	(limited)	(limited)	yes	yes	yes
Cover has explicit location in subcatchment	(limited)	no	(limited)	(limited)	yes
Cover can vary over model run timespan	yes	no	(limited)	no	(limited)
Irrigation + dynamic demand & supply	yes	yes	yes	yes	yes
Potential direct ET from GW (deep root)	yes	yes	(limited)	(limited)	yes
Peak flows & flooding
Max daily or subdaily peak flow est.	no	no	yes	yes	yes
Explicit impacts of channel capacity on flow	(limited)	(limited)	(limited)	(limited)	yes
Calculation of flooded area extent	(limited)	no	(limited)	(limited)	yes
Flood water subject to infiltration, ET, etc	(limited)	no	yes	no	yes
Reservoirs, dams & channel flow modification
Reservoirs explicitly modelled	yes	yes	yes	yes	yes
Facility to represent many small dams	no	yes	yes	(limited)	no
Abstractions & external inputs	yes	yes	yes	yes	yes
Internal transfers between model units	yes	no	yes	yes	(limited)
GW representation & GW-SW interactions
Dynamic, 2-way, GW-SW exchange	yes	yes	no	yes	yes
GW table elevation predicted	(limited)	(limited)	no	(limited)	yes
GW pumping included	yes	yes	no	yes	yes
Wetlands & riparian zones
Wetland processes included	yes	yes	yes	yes	yes
On-channel wetlands	yes	yes	yes	yes	yes
Off-channel wetlands (fed by channel spill)	yes	yes	yes	no	yes
GW fed (receive GW from surrounding)	(limited)	(limited)	(limited)	(limited)	yes
Other catchment & vegetation processes
Sediment movement	no	no	yes	yes	yes
Water quality	no	no	yes	yes	yes
Crop yield	no	no	yes	yes	no
Uncertainty & parameter calibration
In-built tools for uncertainty, parameter sensitivity, & auto-calibration (batch runs)	no	yes	no	no	yes

components of the catchment and concept of a subcatchment affects how the river is defined and the location of the outlet
semi-distributed subcatchments, each with an outlet streamflow is directed to a particular channel unit (or catchment outlet) semi-distributed land units	land areas (modules) draining into a river network, streamflow estimates available at each module outlet lumped land units	flow paths defined by the topography and material of individual land units/cells streamflow estimated at each cell and between cells and vertical units distributed land units
SPATSIM-Pitman ACRU SWAT	WRSM-Pitman	MIKE-SHE
scale of representation affects how the process is calculated and the meaning of parameter values
subcatchment	HRU	grid cell
ACRU SPATSIM-Pitman WRSM-Pitman	SWAT ACRU MIKE-SHE	MIKE-SHE
land cover type discretisation and representation affects whether the land cover is explicitly represented and the ability to pinpoint/track specific impacts of a land cover type on the water balance
no restrictions explicitly represents multiple land uses		considerable restrictions does not fully represent multiple land uses
ACRU MIKE-SHE SWAT WRSM-Pitman		SPATSIM-Pitman
connections between landscape units affects the flow of water between land units before reaching the outlet
outflows routed to the catchment (i.e. subcatchment units to catchment outlet)		interaction of flow between subcatchments (i.e. subcatchment to subcatchment interactions then routing to catchment outlet)
ACRU (unless a riparian zone is added) SWAT		SPATSIM-Pitman WRSM-Pitman MIKE SHE
aquifer representation and connectivity determines the spatial scale of groundwater flows
subcatchment scale	HRU scale	3D modelling mesh (user-defined grid cells)
no regional aquifer water exchanges between subcatchments	no flow between subcatchments	all units can interact
distributed MIKE-SHE SWAT SPATSIM-Pitman WRSM-Pitman	ACRU (with the exception of version 3)	distributed MIKE-SHE
No channel transmission		Channel transmission (contributes and redistributes water to the unsaturated zone, not the aquifer)
ACRU MIKE-SHE SPATSIM-Pitman WRSM-Pitman		SWAT
reservoir representation and connectivity affects the setup demand & accessibility of water storage for irrigation and inflows into the storage
Can include farm dams within a subcatchment farm dams can be used as a water supply for irrigation (practical and useful feature in some landscapes)		Implied reservoir within the catchment an internal water storage unit (ponds) are used to imply reservoirs, ponds cannot be used as a water supply for irrigation, ponds can act as flood control structures with outflow rules	Does not include farm dams within a subcatchment each small dam needs to be input separately (intensive task)
ACRU SPATSIM-Pitman WRSM-Pitman		SWAT	MIKE SHE