BELLAMY, P.H.

Computer Application to Run Catchment Water Quality Models Using Data from Different Sources

BELLAMY, P.H.
Cranfield University, UK
Email: p.bellamy@cranfield.ac.uk

Key words: Catchment Modelling, SWATCAT, SWANCAT, MACRO, ANSWERS, Surface Water, Groundwater, Scaling

An EU project entitled Upscaling, Predictive Models and Catchment Water Quality (CAMSCALE) aims to improve the understanding of the effect of changes of spatial and temporal scale within input data to process-based models. The project is being funded by the Commission of the European Communities under the Framework IV Programme (ENV4-CT97-0439).

As part of the CAMSCALE project, a computer application has been developed to run four process-based models of differing complexity.

The application enables the user to import data for a given catchment from three sources of information (local, regional, and European), store the data in a database and generate input files to run with a range of models typical of those in widespread use with respect to their underlying physics, chemistry, and assumptions:

MACRO is a is a deterministic model simulating dual-porosity non-steady state water and solute transport in field soils;
SWANCATCH is a distributed model for predicting weekly mean nitrate concentrations and flows at the catchment outlet of a river. It is based upon a direct, empirically-derived link between soil type and stream response to rainfall as described in the Hydrology of Soil Types (HOST) system. Nitrate available for leaching in each crop is calculated from simplified versions of the nitrate cycling models SUNDIAL (arable crops) and N-Cycle (grass).
SWATCATCH uses identical hydrological transport to SWANCATCH, but predicts weekly mean and peak pesticide concentrations based on pesticide soil loadings and attenuation within the soil.
ANSWERS 2000 is a distributed, mechanistic model for predicting nitrate concentrations in runoff and groundwater. It has been coupled with a GIS (GRASS).

Each of the models requires data on soils, climate, topology, and cropping regimes. A data dictionary has been written to specify in detail what is required to run each model. Included in this data dictionary are functions required to generate input parameters if they are not available from direct measurements. These functions vary depending on the source of the measured data. For each input parameter there is data (or methods of generating data) for the three different sources of information.

The spatial catchment data is input in the form of an ArcView shape file. The user can display the catchment in the form of a map. Other spatial data can either be input as a raster file or as an ArcView shape file with grid reference and soil (for example) as attributes.

Each model is run over a catchment area defined as a set of independent grids. The size of the grid varies between the models and in the case of ANSWERS the model can be run at different grid sizes. MACRO is a point model which is run for each grid square as a point. The models can be run within the application using the source of data selected by the user. The results from the model runs are concentrations and loads of nitrate and/or selected pesticides in surface water and groundwater as a time series. The application allows stochastic variation in the input parameters to generate a range of results. The results can be displayed in tabular form (as a summary over the whole catchment (SWATCAT and SWANCAT) or for individual grid squares (ANSWERS and MACRO)) and map form, showing values for each grid square for which the model was run (ANSWERS and MACRO).

The application is written to run on a PC in Visual Basic using an Access database to store the data.