Journal of Conference Abstracts

Regional-Scale Nutrient Modeling ­ Adaptation/Transformation of
Physically-Based Model Output

Paul. F. Quinn (Paul_Quinn@adas.co.uk) & Steven G. Anthony (Steven_Anthony@adas.co.uk)

ADAS R&D, Wergs Road, Wolverhampton, WV6 8TQ, U.K.

This paper will show a modeling system for the prediction of NO₃^- concentrations in surface waters of any size catchments. The final model system will be used by the Ministry of Agriculture and other associated environmental agencies to determine future land use strategies, as well as for the reduction of NO₃^- levels in UK catchments to below the legal limit stated by European Law. We must thus ask:

1. What are the critical Nitrate issues at the regional scale?

2. What models are available to predict Nitrate concentrations?

3. What data sets are available at the regional scale, what is the data source, scale and quality?

We will demonstrate clearly that the model structure must change as we upscale. Certain processes will change as we upscale and some processes and parameters will dominate over others. The keen understanding of this behavioural pecking order becomes the key to the overall model design. We argue that the process based models can not be used beyond the research scale as they have too many parameters and contain parameters that cannot be justified at the catchment or regional scale. The modeling approach supported, relies on low parameter models, catchment scale statistics, and some astute scale related assumptions. The models are described as Minimum Information Requirement (MIR) models which require a minimum amount of available data to run. The models are a distillation of the processes seen at the research scale. Two criteria are need for a MIR: (1) The model must mimic closely the output of the fully physical model and must be validated on field scale data (e.g., the ability to mimic closely the total NO₃^- loss per unit of drainage). (2) The parameters remaining must be physically meaningful and be related to the field scale measurements. Thus, the MIR models are 'validated' by field scale models and measurements.

At the regional scale, uncertainties in the estimates of local rainfall, or the amount of NO₃^- being applied to land are high, in fact they negate the use of detailed process models. However, understanding of rainfall dynamics and farmer husbandry practice can lead to a modeling strategy that is appropriate to the scale of the problem. In essence, we can use scale to our advantage. For example two farmers may apply very differing amounts of fertiliser and manure to their land, and may receive very differing volumes of rainfall. However, as we upscale, a statistical representation of the mean, rainfall distribution, and total NO₃^- can be made. This leads to a sound set of model parameter values and a clear estimate of their uncertainty. Thus the final model operates by having :

1. A set of representative land use classes.

2. A scale at which the rainfall inputs are meaningful.

3. Each land use class contains a validated field scale MIR of that scenario.

4. A simple hydrological model that uses scaled routing parameters for moving water through the landscape.