Water Resource Associates

HYSIM’s parameters define in a realistic way the hydrology and hydraulics of the whole river basin (watershed). Such a model is likely to perform well in climatic conditions more extreme than those in its calibration period. The diagram to the right shows the conceptual basis of the hydrological component of the model.

HYSIM can use data on rainfall, potential evaporation (PET), snow melt and abstractions from, or discharges to, both groundwater and surface water. Only rainfall and PET are essential. The data can be daily or any time step less than a day. The simulation time step can be daily or less than a day.

Not only is HYSIM flexible in its data requirements, it is also flexible in terms of sub-catchments and the reaches for flow routing can be either channels or reservoirs. Flow routing uses the kinematic method. Typical uses of HYSIM have included:

• Using long-term rainfall and PET data to produce long-term flow records
• Flow naturalisation
• Studying the effects of climate change
• Flood studies
• Effects of improved drainage
• Groundwater recharge

The output from the model includes: overland flow, impermeable area runoff, snow storage, soil moisture storage, interflow, groundwater recharge, groundwater storage, total surface runoff, routed flow and actual evapotranspiration.

Output from HYSIM can be go directly into Modflow (as recharge) or Isis (either runoff to channels or routed flow at the upstream boundary).

Complex rivers basins (catchments, watersheds) can be simulated as a series of linked sub-basins. To represent hydrological or climatic variations within a sub-catchment, up to three zones, each with its own parameters and data, can be defined.

HYSIM runs under all recent versions of Windows: 11, 10, 7, Vista/XP, NT and 2000.

In addition to the simulation model HYSIM includes facilities for plotting data and simulated and observed flows and tools for data manipulation.

HYSIM is compiled as an ActiveX component and can be run with the Aquator water resources simulation model developed by Oxford Scientific Software working with Water Resource Associates. For more details on Aquator click here.

You can also download an evaluation version of HYSIM. Before you do that please take look at the evaluation link above which shows you what can you do with the trial version.

To download the model (5Mb), click the link below. This is a single zip file which has to be expanded before use. It has three files: HYSIM.CAB (the program file in a compressed format), SETUP.LST (a list of all files included) and SETUP.EXE (the program which will install HYSIM on your computer). You can also download trial data.

Download the documentation (MS Word, 1Mb) below.

For existing users, you can download the latest version of HYSIM.

This note refers to the free trial version of HYSIM. This version has two limitations which do not apply to the full version:

1. A maximum of 5 years of data can be used (in HYSIM there is no limitation).
2. Only one sub-catchment can be used (in HYSIM a wide variety of multiple catchments and/or multiple hydrological zones is possible).

There are three files you can download:
1. HYSIM.zip. This has three files, HYSIM.CAB, SETUP.EXE and SETUP.LST. There are the files needed to install HYSIM.
2. HYSIMDoc.ZIP. This is the HYSIM User Manual.
3. Trial.zip. This is data for testing HYSIM.

The trial data

The data set has the following files:

• trial.par Parameters.              
• trial.dfl Flow (normally observed but in this case simulated)          
• trial.crf Catchment rainfall.              
• trial.dpe Potential evapotranspiration.              
• answer.par The parameter set used to generate the flow file.          
• trial.hpj Project file.              

The trial data set is in a zipped file call test.zip.

Look at the data and parameters

• Start HYSIM

• The first time you run HYSIM you will get the message "Presently using the application directory for log messages. Change it?". You should click on "Yes" and select a suitable folder. (For quality assurance purposes HYSIM keeps a log of everything it does. If the files are written to the application folder these will have to be deleted manually if you uninstall HYSIM).You can also resize the screen to suit your computer.

• Click on the Icon for Edit. Click on File/Open. Select trial.crf. You can now see the rainfall data. You can move around using the keyboard arrow keys and PageUp and PageDown keys. Click on File/Close to exit the editor.

• Click on the Icon for Parameters. Click on File/Open. Select trial.par. You can now see the parameter file. Click on the buttons for Hydraulics and Hydrology to see the parameter values. Click on File/Close to exit the parameter editor.

• Click on the Icon for Graphics. Click on File/Open. Select trial.dfl. You can now see a graph of the first year's flow. Click on the arrow buttons on screen to see data for other years.

Set up project file and run HYSIM

Before you can run the model HYSIM requires a project file, which identifies where the individual data files are located.

• Click on the Icon for Project.

• Open the file trial.hpj in your chosen folder. This file has references to the location of the data files used for a HYSIM project, including the folder used. As the folder you use will not be the same as the one originally used you will get the message "Parameter file not found. Are all the data files in the same folder?". Click on "Yes". You will now get the message "Do you want HYSIM to correct the project file?". Click on "Yes". Click on File/Save and over-write the original project file.

• Click on File/Close.
• Click on the icon for model.
• Click on File/Open existing project file.
• Click on Run/Start. The model will now run and you will get a summary of the results.
• Click on the "graphics" icon.
• Click on File/Open HYSIM file, select "Simulated and observed flow (*.GSM)" and open trial.gsm. This will give a plot of simulated and observed flow.

Some other things you can do

• Select different forms of output (print options) and choose Moisture storages and Moisture transfers. Run the model. Click graphics and select one the files you have just chosen (extensions GMT and GST). You can now look at the internal movement of moisture (GMT) and storage (GST) within HYSIM.

• Enter the Parameter editor and change the values. Run the model and look at the differences in results.

• Change the parameter values a bit more and use the calibration options.

• To get a better simulation change the Pore Size Distribution Index to one appropriate for a sandy soil and the groundwater recession rate to something closer to that observed. (You will need to look at the manual for details). Then run the multiple parameter calibration option. You should now get a much better fit. The "observed" flow is actually a simulated flow record. The file answer.par is the one used to generate the original flow file so you can check how close you have got.

• You can now use HYSIM to set up and calibrate data for your own catchment. The only limitation is that you cannot test the multiple catchment options and you cannot use more than 5 years data.

HYSIM uses a physically realistic approach to modelling the hydrological cycle. It simulates seven natural storages. These are: snow, interception, upper soil horizon, lower soil horizon, transitional groundwater, groundwater and minor channels. The structure of the model is shown on the figure at the end of this section.

Catchment definition
The catchment being simulated can be divided into any number of sub-catchments. The sub-catchment can be divided into up to three hydrological zones, each of which should be reasonably homogeneous with respect to soil type and meteorology.

Data types
The five types of input data which the model can use are :
i) Precipitation. This is given as catchment areal average.
ii) Potential evapotranspiration rate. Estimates based on an empirical relationship.
iii) Potential melt rate. This can be based on the degree day method or a more complex one.
iv) Sewage flow/direct abstractions. The net figure for these is used.
v) Groundwater abstractions.

None of the types of data is compulsory. The data can be given in a monthly, daily or any shorter time increment, provided there is an integral number of data items per day.

Conceptual storages
Snow storage. Any precipitation falling as snow is held in snow storage from where it is released into interception storage. The rate of release is equal to the potential melt rate.

Interception storage. This represents the storage of moisture on the leaves of trees, grasses etc. Moisture is added to this storage from rainfall or snowmelt. The first call on this storage is for evaporation which, experiments have shown, can take place at more than the potential rate. This can be allowed for in the model. Any moisture in excess of the storage limit is passed on to the next stage.

Impermeable area. A proportion of the moisture in excess of the interception storage limit is diverted to minor channel storage to allow for the impermeable proportion of the catchment.

Upper Soil Horizon. This reservoir represents moisture held in the upper (A) soil horizon, i.e. top-soil. It has a finite capacity equal to the depth of this horizon multiplied by its porosity.

A limit on the rate at which moisture can enter this horizon is applied, based on the potential infiltration rate. This rate is assumed to have a triangular areal distribution, as in the models of Crawford and Linsley, and Porter and McMahon.The potential infiltration rate is based on Philips equation, i.e.

x = F t0.5 + c  t1.0 + w t1.5 + ......

Where x is the distance travelled downwards by the wetting front, t is time since x = 0 and F ,c  +and w  are functions of soil type and condition. It has been shown by Manley that this relationship can be closely approximated to :

x = (2k Pt)0.5 + kt

where P is the capillary suction (mm. of water) and k the saturated permeability of the medium (mm/hr). This allows determination of the potential infiltration rate.
Brooks and Corey have shown that P can be expressed as :

P = Pb/Se1/g

where Pb is the bubbling pressure (mm of water), γ is a parameter (called the pore size distribution index) and Se is the effective saturation defined as :

Se = (m - Sr)/(1.0 - Sr)

where m is the saturation and Sr is the residual saturation, i.e. the minimum saturation that can be attained by dewatering the soil under increasing suction. By simulating the moisture content in the upper horizon the forces causing movement of the water can therefore be simulated. The first loss from the upper horizon is evapotranspiration which, if the capillary suction is less than 15 atmospheres, takes place at the potential rate (after allowing for any loss from interception storage). If capillary suction is greater than 15 atmospheres no evapotranspiration takes place. The next transfer of moisture that is considered is interflow (i.e. lateral flow). The rate at which this occurs is obviously a very complex function of the effective horizontal permeability, gradient of the layer and distance to a channel or land drain. Brooks and Corey have also shown that the effective permeability of porous media is given by :

ke = k (Se)(2+3γ )/γ

where ke is the effective permeability (mm/hr) and the other terms are as defined previously. Because of its complexity no attempt is made to separate the individual parameters for interflow and it is given as :

Interflow = Rfac1(Se) (2+3γ )/γ

Where Rfac1 is defined as the interflow run-off from the upper soil horizon at saturation. The final transfer from the upper horizon is percolation to the lower horizon and is given by :

Percolation = kb(Se) (2+3γ )/γ

where kb is the saturated permeability at the horizon boundary and Se is the effective saturation in the upper horizon. By combining the above equations the rate of increase in storage is given by :

ds = i - (Rfac1+kb) Se (2+3γ )/γ
dt

where i is the rate of inflow and S and t are moisture storage and time respectively. Unfortunately this equation cannot be solved explicitly so it has been assumed that the total change in storage in any time increment is small compared to the initial storage. In this case the equation can be simplified and an approximate solution obtained. As a check for extreme situations the change in storage is constrained to lie within an upper and lower limit. The upper limit is defined by the level of storage at which the rate of outflow is equal to the rate of inflow. The lower limit results from setting i equal to zero in the above equation, in which case an explicit solution is possible.

Lower Soil Horizon. This reservoir represents moisture below the upper horizon but still in the zone of rooting (i.e. the B and C horizons). Any unsatisfied potential evapotranspiration is subtracted from the storage at the potential rate, subject to the same limitation as for the upper horizon (i.e. capillary suction less than 15 atmospheres). Similar equations to those in the upper horizon are employed for interflow runoff and percolation to groundwater.

Transitional Groundwater. This is an infinite linear reservoir and represents the first stage of groundwater storage. Particularly in karstic limestone or chalk catchments many of the fissures holding moisture may communicate with a stream rather than deeper groundwater and the transitional groundwater represents this effect. Its operation is defined by two parameters : the discharge coefficient and the proportion of the moisture leaving storage that enters the channels. Being a linear reservoir the relationship between storage and time can be calculated explicitly.

Groundwater. This is also an infinite linear reservoir, assumed to have a constant discharge coefficient. It is from this reservoir that groundwater abstractions are made. As in the above case the rate of runoff can be calculated explicitly.

Minor Channels. This component represents the routing of flows in minor streams, ditches and, if the catchment is saturated, ephemeral channels. It uses an instantaneous unit hydrograph, triangular in shape, with a time base equal to 2.5 times the time to peak.

HYSIM uses kinematic routing of flows. This enables the hydraulics component to use channel dimensions and Manning's roughness coefficient. The only situations where the equations would not apply are in rivers artificially controlled by sluice gates, influenced by tides or with very flat gradients. In general if a stage/discharge relationship is possible then the kinematic approximation is acceptable.

Wave theory
Whilst flow in open channels can be described most accurately by the Saint Venant equations, these equations are not explicitly solvable. The methods of solution that are available are not suitable for a hydrologic model as they require an iterative solution for small time increments. Fortunately most of the terms in the Saint Venant equations have only minor effects and flow can be described adequately by the simplified form known as the kinematic method (Lighthill & Witham).
The velocity of a kinematic wave, Vw, is given by:

Vw = d Q .......................... 1
.........d A

where Q is the incremental change in flow and A the incremental change in area.

Manning's formula when applied to a triangular channel gives:

Q  a A^4/3

where Q is the discharge and A is the sectional area of flow.

For a broad rectangular channel the equivalent equation is :

Q a A^5/3

Since most channels fall between these two extremes then it has been assumed that

Q = CA^1.5 .............................. 2

For flow in bank, A, as a function of Q, is calculated by re-arranging equation 2.

For flow out-of-bank exponential relationships are developed at the start of the program. They are of the form :

A = a Q^b ................................ 3

where a and b are constant. They are based on the geometry and roughness of the flood plain using Manning's equation. Two such relationships are used, one for when the flood plain is filling up and one for when it is full.

Kinematic waves
The way the kinematic theory is used in the model can best be visualized by reference to the following figure. The input hydrograph is shown as the solid line. Using equation 2 or 3 the sectional area for the input flow is calculated. From equation 1 the velocity of the wave and hence its travel time is calculated. The hydrograph is then displaced by this travel time for each time increment. The result of this stage is represented by the dotted line.

The time increments of the "dotted" hydrograph will not generally be the same as those of the model. This "dotted" hydrograph is therefore redistributed to produce the "dashed" hydrograph.

The situation sometimes arises where a later wave has a higher velocity than an earlier wave, and catches up with it within a section. The program checks whether any such events occur, and if they do it recalculates the time of travel of the resulting "coalesced" wave from the point at which it was formed. This approach is of particular importance when flows go out-of-bank and rapid changes in wave velocity occur.

Whilst the above approach involves a degree of compromise it is very stable and can be used with time steps much longer than the time of travel of a kinematic wave in the section being modelled, an advantage when using daily data but when travel times in a short river have to be simulated.

Data Preparation
The accuracy of a model can never be better than the accuracy of the data. Much of the time and effort in any modelling project is necessarily taken up in the collection, preparation and quality control of the data. In particular all models require data in their own format. HYSIM includes a data conversion module which should enable data of almost any format to be handled without use of computer programming.

Some of the most useful features of the data preparation tools are:

• Data in almost any format can be imported to HYSIM provided it is in a text file. In most cases the data can be translated directly. In a few cases where a complicated format is used some pre-editing may be needed but this is rare. No programming is need for this option.

• The data for modelling can be set up easily from the calculation of the mean through to combining files of different types of data to provide a single catchment data file.

• Some tasks are automated such as infilling missing rainfall data and correction of errors revealed by double-mass plots.

Parameters
The model uses mathematical relationships to define the transfer of moisture. The relationships use variables, which change with time, and parameters which do not. With any model, assigning suitable values to the parameters is crucial to the accuracy of the simulation. There has been, and probably always will be, much discussion between those who adopt a physically realistic approach, and would want all parameters to be measurable, and those who adopt a systems approach, and want to use the minimum number of parameters compatible with accurate simulation. HYSIM is more realistic than many models and consequently has more parameters.

All of the parameters in HYSIM have a realistic interpretation. Although no model is yet at the stage where a catchment can be modelled solely on the basis of physical measurements, it is of great help in calibrating the model to have an awareness of the likely range of parameter values, particularly when getting starting values for the optimisation process.

Ultimately the important question is not "Is the model structure simple?" but "Is the model easy to use?". HYSIM, by providing on line help for parameter estimation and a powerful optimisation process, is easy to use

Graphics
It is easier to get a feel for the quality of hydrological data or the accuracy of simulation if the results can be visualised on a graph. The graphics module of the HYSIM system allows for a variety of output types including the monitor, printer, graphics (WMF) file and output as a text (CSV) file for later input to spread-sheet or text processing programs.

All HYSIM files have a standard structure which makes it relatively easy to plot them. This includes: observed flow, simulated and observed flow, soil moisture storage, soil moisture transfers and rainfall. Other options allow for flow and gaugings to be plotted on the same graph and double-mass plots (very useful for quality control of rainfall data). The image to right shows soil moisture and groundwater storage.

The Simulation Model
The rainfall/runoff model is at the heart of the system. It uses precipitation and climate data to simulate the movement of moisture, both above and below ground, from the moment the precipitation reaches the earth until it flows out of the river basin. Internally the model simulates interception storage, runoff from impermeable areas, overland flow, interflow from the upper and lower and soil horizons, rapid and slow response from groundwater and the hydraulics of flow in river channels. The model can use data on precipitation, potential evapotranspiration, potential snow melt, discharges to and abstractions from rivers, and discharges to and abstractions from groundwater.

The model can be run either with or without optimisation. Without optimisation a range of output files are possible, either for tabulation or for plotting. In this mode the simulation of five years daily flow takes around 1 second on a PentiumII 400 MHz. In the optimisation mode little output is provided which speeds up processing. The optimisation process is based on the Rosenbrook method. There is a choice of three objective functions, depending on whether high flows, low flows or overall accuracy is most important.

The following is a list of publications by Ronald Manley and others dealing with HYSIM and related hydrological topics.

1975: A hydrologic model with physically realistic parameters, UNESCO symposium, Bratislava.

1977: The soil moisture component of mathematical models, J. of Hydrology.

1977: Improving the accuracy of flow measurement, Water Services.

1978: Calibration of Hydrologic Model Using Optimisation Technique, Journal Hydraulics Div., A.S.C.E.

1978: Hydrological model in water resources planning, Proc.I.C.E.

1978: Simulation of flow in ungauged basins, Hyd. Sci. Bulletin

1978: Development of a flow data bank, UNESCO symposium, Leningrad

1978: HYSIM - a general purpose hydrologic model, Pisa symposium (Pergamon Press - 1982)

1980: Mathematical models for flow forecasting, UNESCO symposium, Oxford. (Co-author with J.Pirt and R.Douglas)

1986: Technology Transfer of hydrometeorological applications to developing countries, Am. Met. Soc. Symposium, Miami (Co-author with J.Nemec).

1987: HYDRONIGER project, WMO symposium, Toulouse (Co-author with J-R Thiebaux).

1989: Operational Hydrology Problems in the Humid Tropics , in Bonnel, Hufschmidt and Gladwell, Hydrology and Water Management in the Humid Tropics, Cambridge University Press, 1993 (Co-author with A.Askew).

1990: The possible effects of reduced inflows to Lake Toba on Power and Aluminium production. International Conference on Lake Toba, Jakarta, Indonesia.(co-author with F.A.K. Farquharson and D.G.Knott).

1993: The IUCN Review of the Southern Okavango Integrated Water Development Project, IUCN. Co-author/editor with T Scudder et al).

1994:  The Long-term Development of a Hydrological Model, First International Conference of Hydroinformatics, Delft, The Netherlands.

1994: The Review of the Southern Okavango Integrated Water Development Project, International Conference on Integrated River Basin Development, Wallingford, England. (Co-author with E.W. Wright).

1995: River basin development and management, proceedings of workshop on Water Resources use of the Zambezi Basin, ed T Matiza et al, IUCN

1999: High resolution climate change scenarios: implications for British runoff, C Pilling and JAA Jones, Hydrological Processes.

2003: Climate Scenarios For Mid Century And Some Preliminary Perspectives On Engineering Implications. Charlton, R., Sweeney, J., Fealy, R., Murphy, C., and Moore, S. National hydrology conference.

2005: Implications of climate change for river regimes in Wales – a comparison of scenarios and models J. A. A. Jones, N. C. Mountain, C. G. Pilling & C. P. Holt. Fourth Inter-Celtic Hydrology Colloquium.

2006: Assessing the Impact of Climate Change on Water Supply and Flood Hazard in Ireland Using Statistical Downscaling and Hydrological Modelling Techniques. Rosemary Charlton , Rowan Fealy, Sonja Moore, John Sweeney and Conor Murphy. Climatic Change, Vol 74.

2006: The reliability of an ‘off-the-shelf’ conceptual rainfall runoff model for use in climate impact assessment: uncertainty quantification using Latin hypercube sampling. Rosemary Charlton , Rowan Fealy, and Conor Murphy, Area, Vol 38.

2006: Changing Precipitation Scenarios: Preliminary Implications for Groundwater Flow Systems and Planning, Conor Murphy, Rowan Fealy, Ro Charlton and John Sweeney, National hydrology conference.

2006: Climate Change Impact On Catchment Hydrology & Water Resources. For Selected Catchments In Ireland. Conor Murphy and Ro Charlton, Irish Climate Analysis and Research UnitS (ICARUS), NUI Maynooth. National hydrology conference.

2008: Hydrology simulation of the Vardar River. Manley, Ronald. Dimitrievski, Ljupco, Andovska, Sandra. BALWOIS conference.

What it is
HYSIM is a Hydrological Simulation Model which uses rainfall and potential evapotranspiration data to simulate river flow. Its parameters, for hydrology and hydraulics, define the river basin and channels in a realistic way. Whilst being realistic the model is easy to use. HYSIM  has a long pedigreee and has been described in a number of publications.This has a number of advantages.

• Such a model is more likely to perform well in climatic conditions more extreme than those in its calibration period.

• As the model does not rely entirely on measured input and output data for its accuracy, but can also use data such a soil type and rooting depth, it can be used to study catchment changes.

• It enables good initial estimates to be obtained for model parameters.

• The theory behind the hydrological and hydraulic aspects of the model is described in separate sections

What data does it need?
HYSIM is flexible in its data requirements. As a minimum it needs precipitation and PET data but it can also use data on snow melt, abstractions from and discharge to surface water and pumping from groundwater. The data can be daily or any shorter time step. The time step for different types of data and for calculations do not need to be the same.

Not only is HYSIM flexible in its data requirements, it is also flexible in terms of catchment definition. Within a single sub-catchment it is possible to represent channels by up to 20 reachs and to have three zones, each of which can have its own parameters and data, to represent hydrological or climatic heterogeneity. It is also possible to link a series of sub-catchments to simulate large complex river basins.

What it can do?
River flow extension. Rainfall and other meteorological records are typically of longer duration than flow records. Using a few years of flow and meteorological data to calibrate the one can then simulate flow for a longer period.

Climate change. By simulating a river basin with different climate scenarios it is possible to assess the effects of climate change on river flow and groundwater levels.

Flow naturalisation. HYSIM can simulate a catchment as it is, with man-made influences such as groundwater abstraction or urbanisation. It can be calibrated using a flow record which reflects these influences. The flow extension can then be carried out with the parameters and data modified to remove these effects thus giving a naturalised flow record.

Groundwater/surface interaction. The fact that HYSIM can simulate groundwater pumping, its effects on groundwater storage and consequent effects on runoff, enables it to be used for conjunctive use studies.

How it does it?
HYSIM’s strengths come in the following areas:

• Data entry and preparation, often a very time consuming part of a simulation project, are facilitated within HYSIM. Some of the features include rainfall data infilling, calculation of mean basin rainfall, comparison of flow and current meter gaugings and manipulation of data files.

• Parameter entry is done within the model with no need for editing of text files. On line help is available to aid in parameter selection.

• Graphics facilities include plotting of data, observed and simulated flows and double-mass plots. As well as on-screen plotting the graphs can be sent to a printer or to Windows Metafile format files.

• The model itself has a built-in optimisation algorithm which facilitates calibration.

Other uses
HYSIM has other more advanced facilities which include:

• Calibrating the model with daily data but studying major floods using hourly data. It is possible to output values of moisture storage on a daily basis and use these to restart the model on a particular day. One could then simulate a flood flow for a few days.

• Modelling of the soil layers in a realistic way. HYSIM is therefore able to simulate variations of moisture in the soil which could be used for studies of chemical leaching.

• Outputting intermediate values of calculations. This enables HYSIM to be used with other more specialised models. For example runoff could be used with a hydraulic model for detailed studies of flood alleviation measures or the percolation to groundwater could be used as input to a hydrogeological model.

Documentation
Two documents are provided with HYSIM. The User Guide describes how to set up data and parameters, calibrate the model and produce the results. The guide also contains an exercise covering most aspects of using HYSIM. The Reference Manual gives more details on the theory behind the model and the format of the data files.
Uses of HYSIM

HYSIM has been used in Europe, Africa, South America and Asia. Its uses have included:

• Flow record extension and infilling,
• Flow naturalisation,
• Conjunctive use of surface and groundwater,
• Flood studies, both independently and with hydraulic models,
• Groundwater studies, both independently and with hydrogeological models,
• Studying effects of climate change.

Computer requirements
HYSIM runs under all recent versions of Windows: 11, 10, 7, Vista/XP, NT and 2000.