Knowledge development through measurements

Deltares develops knowledge about water, soil and the subsurface. We can predict the behaviour and quality of soil and water using numerical models. New knowledge and better models are based on real-life measurements and physical models.

To predict the consequences of interventions in ground, we use numerical models (software). Ever more powerful computers allow us to make increasingly precise calculations. But models are still only an approximation of real life. So numerical models can always generate results that may not match reality. Comparing numerical models, practical experience and tests with experimental facilities makes it possible to reduce the ‘uncertainty gap’ between predictions and reality in a systematic way. The knowledge innovation cycle (see figure) describes the cyclical succession of observation, hypothesis, prediction and comparison with reality. The difference between predictions and reality is examined by running through the cycle again.

Models and reality

Models are only an approximation of real life. That is why model results are often different from actual observations. We can optimise short-term expectations by taking measurements. We do this, for example, with weather forecasts and water levels in the River Rhine.

Systematic discrepancies

If the mismatches between predictions and the measured effects are small but systematic, there will probably be a problem with the selection of the model parameters. New measurements from the field will then be used to calibrate the model and to determine the parameters more precisely. The behaviour of water or the subsurface can then be predicted more reliably. If the mismatches are substantial, it is also possible that we have not yet properly understood the mechanisms involved. To enhance our insight, we use tests with experimental facilities.

Experimental facilities

Testing with experimental facilities usually starts with "observation tests". This involves the qualitative exploration of the mechanisms involved: do cracks appear in the soil at a given load; do waves break on this slope; does the contamination spread? And how do the processes work? This information is used as the basis for improvements in our conceptual models. More advanced, quantitative, model tests then take place. At this point, the conditions can be set and varied precisely so that a one-on-one relationship is established with the structure in the numerical models. This opens up the possibility of quantitative testing and validation for the conceptual model. The result is that processes in the field can be predicted more reliably. The application of the numerical model in practice brings the innovation cycle full circle.