Skip to main content

Effect of upscaling of input data on the output of predictive models in relation to catchment water quality


We propose to quantify the effects of 'upscaling', i.e. changes in the scale of temporal and spatial inputs (e.g. soils, landuse, farming practice, climate), on the output of well-characterised process-based models as used to predict water quality in catchments. Model predictions obtained at different scales of resolution of the inputs will be compared with measured data for named chemical species in catchment waters in European cool marine, cold temperate and temperate Mediterranean climates. The test substances will be nitrate-nitrogen and pesticides. The research addresses the following key questions:

1. how can model(s) be run with more limited data (spatial and/or temporal) than was allowed for in the design of the model(s);

2. what effect(s) do the limitations of available data have on model output(s);
3. what can be reasonably predicted - annual mean concentration, peak events maximum concentration etc., and under what conditions;

4. are the problems/consequences the same at all geographic/temporal scales, and with simple versus more complex models;

5. to which factors are the models most sensitive in this context;
6. can these problems be dealt with in a structured manner;

7. can the answers to such questions be formalised so that users know the degree of confidence which can be attached to model output under a range of input conditions.

To keep the research within practical limits, we will model using:
WAVE : for process-based modelling of N species;
MACRO : for process-based modelling including preferential flow, of pesticide species;
SWATCATCH : a less complex model to predict nitrate and pesticide concentrations, respectively, and flow volumes at catchment
ANSWERS 2000 : for distributed modelling.

We also propose to:

- show how the large amounts of relatively simple spatial and, temporal environmental data within the EU can be utilised to model catchment water quality;
- link models to spatially distinct data sets using GIS techniques; - examine Digital Terrain Modelling as a tool to predict spatial soil patterns and associated soil water flow regimes ('routing'); - formalise the output of this research as guidance principles aimed at the user of process-based models in a distributed context.

Thus, the output of this research will be applicable to a wider range of environmental conditions, including climatic change, and models, than those considered specifically within the project.

Funding Scheme

CSC - Cost-sharing contracts


Silsoe Campus Wharley End
MK45 4DT Silsoe,bedford
United Kingdom

Participants (2)

The Swedish University of Agricultural Sciences
Ulls V.
750 07 Uppsala
Rue De La Piscine 1023, Domaine Universitaire
38041 Grenoble