Uncertainty Quantification Framework for Nutrient Load Estimation
Bruno Bezerra Bluhm from the Technical University of Munich (TUM) stayed at the ICWRGC from 13 May – 12 November 2019 to write his master’s thesis on the topic “Uncertainty Quantification Framework for Nutrient Load Estimation”. He successfully defended his thesis at TUM on 5 December 2019.
Nutrient loads, such as phosphorus and nitrogen, are a key indicator of water quality monitoring, management and assessment, particularly regarding eutrophication. Although flow data is commonly available in continuous forms, monitoring programs often sample concentration following low-frequency policies, which hinders flux calculations. This consequently limits purposes of load estimations, since, in order for the metric to be used in evaluations, its uncertainty must be within an expected threshold. This Master’s thesis conceives an uncertainty quantification framework to be used by an existing load estimation algorithm from GEMStat. The framework calculates expected errors by matching a target station, i.e. scarce data, and a reference station, i.e. continuous data. Brazil, a major GEMStat partner, has been selected as a case study to define the target station baseline. By performing Monte Carlo Simulations and credibility-based assessments, the framework evaluates which estimation algorithm is most suitable and what are the expected levels of uncertainty, given in three forms: credible interval, relative absolute error, and classification (under or overestimation). Expanding the current database to include a greater range of reference stations is the main challenge to improve the framework’s usability in the future.