De Ville, Mitzi; Feuchtmayr, Heidrun. 2017. Constraining uncertainty and process-representation in an algal community lake model using high frequency in-lake observations.Contact CEH NORA team at noraceh@ceh.ac.ukThe NERC and CEH trademarks and logos ('the Trademarks') are registered trademarks of NERC in the UK and other countries, and may not be used without the prior written consent of the Trademark owner.
Abstract 13Excessive algal blooms, some of which can be toxic, are the most obvious symptoms of 14 nutrient enrichment and can be exacerbated by climate change. They cause numerous 15 ecological problems and also economic costs to water companies. The process-16representation of the algal community model PROTECH was tested within the extended 17Generalised Likelihood Uncertainty Estimation framework which includes pre-defined Limits 18 of Acceptability for simulations. Testing was a precursor to modification of the model for real-19 time forecasting of algal communities that will place different demands on the model in terms 20 of a) the simulation accuracy required, b) the computational burden associated with the 21 inclusion of forecast uncertainties and c) data assimilation. We found that the systematic 22 differences between the model's representation of underwater light compared to the real 23 lake systems studied and the uncertainties associated with nutrient fluxes will be the 24 greatest challenges when forecasting algal blooms. 25 blocked filters, poor taste and odour and, in more extreme cases, high levels of algal-derived 32 toxins. Managing these effects costs greater than £50 million per year in the UK (Pretty et 33 al., 2003) and billions of dollars annually in the US (Dodds et al., 2009;Michalak, 2016). 34Implementation of mitigation strategies is becoming more expensive owing to increases in 35 the frequency of blooms (Ho and Michalak, 2015) as a result of nutrient enrichment and 36 climate change (Brookes and Carey, 2011;Paerl and Huisman, 2008; Rigosi et al. 2014) 37 and the effectiveness of interventions is, in some cases, being compromised. It is therefore 38 beneficial to be able to forecast algal blooms to allow the most cost-effective management 39 strategies to be implemented. 40One algal model that has been used in lakes and reservoirs around the world is PROTECH 41 Elliott, 2010Elliott, , 2012Reynolds et al., 2001). PROTECH was used here 42 because it explicitly simulates the dynamics of lake algal community structure and hence 43 algal types of particular interest including cyanobacteria. As real-time forecasting of algal 44 blooms is becoming a priority for the management of lakes and reservoirs used for water 45 supply and recreation, one of the aims of this study is to test the model as a precursor to 46 modification for forecasting purposes. Real-time forecasting places different demands on the 47 analyses were carried out within a hypothesis testing framework where different model 56 representations were considered as competing hypotheses and accepted or rejected bas...