Effects of changes in the driving forces on water quality and plankton dynamics in three Swiss lakes – long-term simulations with BELAMO
Article first published online: 24 OCT 2012
© 2012 Blackwell Publishing Ltd
Volume 58, Issue 1, pages 10–35, January 2013
How to Cite
DIETZEL, A., MIELEITNER, J., KARDAETZ, S. and REICHERT, P. (2013), Effects of changes in the driving forces on water quality and plankton dynamics in three Swiss lakes – long-term simulations with BELAMO. Freshwater Biology, 58: 10–35. doi: 10.1111/fwb.12031
- Issue published online: 11 DEC 2012
- Article first published online: 24 OCT 2012
- (Manuscript accepted 4 September 2012)
- input loads;
- lake modelling;
- long term;
- model bias;
1. With a modified version of the lake model BELAMO, we were able to describe the essential features of the dynamics of nutrients, dissolved oxygen, phyto- and zooplankton in three lakes of different trophic status over periods of 19–30 years, with essentially the same model parameters for all three lakes. This is remarkable, as the measured nutrient inputs decreased considerably during the simulated time period.
2. Despite having done this before for a period of 4 years with an earlier version of the model, a considerable effort was required that led to a series of model modifications without which the data could not be matched. This demonstrates that long-term calibration of a model that combines processes in the water column with mineralisation in the sediment can be difficult.
3. Due to the necessarily simplified processes within the model, there is a bias in its output. We applied a recently developed technique for model calibration and uncertainty analysis to address bias and multiple calibration criteria. To account for the demanding long-term simulations, a simplified numerical implementation of this technique was used.
4. Our results demonstrate good understanding of the chemical state of the lake during the calibration period but less of the biological variables. The credibility intervals used to visualise this knowledge widen substantially during the prediction period (consisting of the last 10 years of the simulation).
5. The joint calibration of the model with long-term data from lakes of different trophic status is possible but only with considerable prediction uncertainty. Due to the explicit consideration of bias in our calibration technique, we are able to estimate quantitatively the uncertainty of our knowledge about chemical and biological variables in the lake.