Requirements

The requirements to build and run a MIKE SHE model depend on the purpose of the model and the trade-offs that must be made between conceptualization and the practicality of simulation time.

1. Input requirements

The flexibility of MIKE SHE means that there is no predefined list of required input data. The required data depends on the hydrologic process included and the process model selected, which, in turn, depend on what problem you are trying to solve with MIKE SHE. However, the following basic model parameters are required for nearly every MIKE SHE model:

• Model extent - typically as a polygon,
• Topography - as point or gridded data, and
• Precipitation - as station data (rain gauge data).

Additional basic data is required depending on the hydrologic processes included, and their options:

• Reference evapotranspiration - as station data or calculated from meteorological data,
• Air Temperature - for calculating snowmelt (station data),
• Solar Radiation - for calculating snowmelt (station data)
• Sub-catchment delineation - for runoff distribution
• River morphology (geometry + cross-sections) - for river flow and water level calculations
• Land use distribution - for vegetation and paved runoff calculations.
• Soil distribution - for distributing infiltration and calculating runoff.
• Subsurface geology - for calculating groundwater flow.

If you also want to calculate water quality, then additional basic information includes:

• Species to be simulated, and
• Source locations

The data items listed above are the basic input data that define your problem. They are not usually part of the calibration. If we now look at each of the hydrologic processes, and the process models available for each, then we can separate out the principal calibration parameters.

Table 1.1: Principal parameters for MIKE SHE

	Principal calibration parameters	Other parameters
Overland flow (finite difference)	Surface roughness	Detention storage
Overland flow (sub- catchment-based)	Surface roughness	Detention storage Slope parameters
River flow	Riverbed roughness Riverbed leakage coefficient
Unsaturated flow (finite difference)	Saturated hydraulic conductivity	Soil water contents at saturation, field capacity, and wilting point Soil function parameters
Unsaturated flow (2-layer method)	Saturated hydraulic conductivity	Soil water contents at saturation, field capacity, and wilting point Capillary thickness
Actual Evapotranspiration	Leaf Area Index Root depth	Canopy Interception FAO Crop coefficient. Kristensen and Jensen ET parameters
Groundwater flow (finite difference)	Hydraulic conductivity Specific yield Specific storage	Drain level. Drain time constants
Groundwater flow (linear reservoir)	Reservoir time constants Reservoir volumes (specific yield, depths)	Interbasin transfers (dead zone storage)
Water quality	Porosity Soil bulk density Dispersivities Sorption and degradation rate constants	Source strength

The parameter list in Table 1.1 is not complete. There are many other parameters that can be modified if you are trying to simulate something specific, such as snowmelt. If you do not simulate a process, then a place holder parameter is usually required that will need to be calibrated. For example, if you do not simulate the unsaturated zone and evapotranspiration, then precipitation must be converted to groundwater recharge using the Net Rainfall Fraction and Infiltration Fraction parameters to account for losses to evapotranspiration and runoff.

2. Model limits

Although, there are no physical limits to the size of your model, there are practical limits and hardware limits.

The practical limits are generally related to run time. We all want the model to be a little bit bigger or more detailed. However, that little extra detail or slightly smaller grid size can quickly lead to long run times.

The physical limits are generally related to memory size. If you model requires more memory than is physically installed on the computer, then the computer will start to swap data to the hard disk. This will vastly slow down your simulation.

If your model reaches the practical or physical limits of your computer, then critically evaluate your model to see if you really need such a large, complex model. For example, maybe you can reduce the number of UZ nodes or increase the grid size.

If the model is simply too slow, then you may be able do an initial rough calibration with a less complex model. For example, during the initial calibration, you could double the grid spacing or shorten the calibration period. After- wards, you can switch back to the original configuration for the final calibration. You might even be surprised that the rougher model is actually good enough.

3.MIKE SHE demo model limits

If no dongle is installed, or if a valid license is not available, then MIKE SHE will run in demo mode. In this case, the model size is restricted. If you need a full-size MIKE SHE to perform your evaluation, then you are welcome to contact your local DHI office to request a 30-day evaluation license.

The current demo restrictions are as follows:

• number of cells in x- and y-direction: 70
• number of computational cells per layer (incl. boundary cells): 2000
• number of computational saturated zone layers: 2
• number of river links: 250
• number of computational UZ columns (multi-layer UZ): 155
• number of nodes per UZ column (multi-layer UZ): 100
• simulation time: 4444 hours or 185 days
• number of UZ time steps: 800
• number of SZ time steps: 200
• no steady-state SZ
• no overbank spilling
• no MIKE ECO Lab linkage
• no irrigation

Further, there are some restrictions in the rest of the MIKE Zero tools in demo mode. The most critical of these is that the Grid and Time Series Editors do not allow you to save files in demo mode.

4. Hardware requirements

The hardware requirements for MIKE SHE depend on the model that you are trying to simulate. As a rule of thumb, any good quality, new computer should be sufficient for an average MIKE SHE model. Thus, a typical machine for an average MIKE SHE model will have at least a 2GHz CPU, 8-16GB of RAM, and 100-500 GB of free disk space.

However, these are minimum requirements. In particular, data storage is often a problem. A large model with a long simulation period and a short saved time step interval can easily generate very large output data sets. If you save multiple simulations (e.g. calibration runs or scenarios), then you can quickly have hundreds of Gigabytes of output data.

Note

MIKE SHE must run in a Windows environment and will not run on Linux workstations.

64-bit CPU

All of the DHI numerical engines are only compiled for a 64-bit processor, including MIKE SHE.

Multi-core/processor computers

The numerically intensive operations in the MIKE SHE engine have been optimized for multi-core computers. However, not all of the hydrologic processes scale equally well. Thus, the simulation speed improvements on multi-core computers depends on the model.

MIKE SHE always runs with the maximum allowed number of threads, which will not be slower, but may be less efficient if you are running multiple simulations at the same time. There is an extra parameter option ("max number of threads") that you can use to control the parallelization of each simulation.

Also, hyper threading is less efficient than physical cores. So, there might also be some benefit to turning off hyper threading for some simulations (This is a BIOS setting).

The AUTOCAL program for parameter optimization and sensitivity analysis has been updated to automatically spread out the simulation load to the available cores.

The standard MIKE Zero license supports up to eight cores/processors. If you want to take advantage of more than eight cores, then you will need to contact your local DHI sales office to obtain additional runtime licenses.

However, since 2017 Release all the cores available can be used. There are no longer any restrictions on the number of cores - as long as you have a Corporate License and valid SMA).

RAM

MIKE SHE does not dynamically allocate RAM. That is, the amount of RAM required by the model is allocated at the beginning of the simulation based on the specified number of nodes. If you don’t have enough RAM, then MIKE SHE will swap to the hard disk, which can drastically slow down your simulation.

The amount of RAM may also be important when running multiple simulations at the same time, since each simulation will require a full memory space.

CPU Speed

In general, the higher the CPU clock speed, the faster the calculations. However, simulation speed also depends on the chip design, which depends on the manufacturer (e.g. Intel vs AMD), the platform (e.g. laptop vs. desktop), etc. Given the huge range of chip designs and the rapid pace of development, it is difficult to give specific guidance on choice of CPU - other than “faster is usually better, all other things being equal”.