Water Resource Associates

USING HYSIM

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:

w 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.

w 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.

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

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.

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.