Sampling design optimisation for uncertainty analysis of integrated water quality models

PhD project Alexandre Wadoux on uncertainty analysis

Problem definition and objective

The overall aim of QUICS is to provide high levels of training and carry out research in order to take the implementation of the Water Framework Directive (WFD) to the next level and improve water quality management by assessing the uncertainty of integrated catchment model water quality predictions.

Figure 1: Quantifying Uncertainty in Integrated Catchment Studies (QUICS) is an EU funded FP7 Marie Curie Initial Training Network (ITN) which will run for 4 years from 1st June 2014.


Hydrologic models need input data and data for calibration and validation. Collecting these data is costly and time-consuming. In this respect, we must (i) optimise sampling patterns to gain maximum information for a given budget and (ii) find the optimal sampling design for minimisation of input, calibration and/or validation uncertainty.


Figure 2: Study area of the Upper-Sure in Luxembourg.


The project purpose is to employ advanced geo-statistical methods to optimise sampling designs used in integrated catchment water quality modelling. This research will optimise sampling designs in space and time using statistical tools from spatial uncertainty propagation analysis, statistical sampling theory and Bayesian calibration. Various sampling design optimisation techniques will be developed and tested, by making a trade-off between the cost of data collection and model output accuracy. Testing and validation is done for a Soil Water Assessment Tool applied to a catchment in Luxembourg and for minimisation of spatial and temporal rainfall estimation errors using radar data and rain gauges for a study area in the UK. The main overall targets of this PhD research are to:

•      Develop and test statistical methods for sampling design that minimise the space-time interpolation error of model inputs.

•      Develop statistical methods for sampling design for optimisation of model outputs used for Bayesian calibration of various water quality models.

•      Investigate optimal sampling designs for minimisation of input, calibration and validation uncertainty.

•      Use multi-criteria analysis techniques to merge these into a unified optimal sampling strategy for water quality based models.


-      University of Bristol,  Department of Civil Engineering (3 months in spring 2016)

-      Luxembourg, CRP Henry Tudor (5 months between 2016 and 2018)

Start PhD. project: September, 2015

End PhD. project: September, 2019


please contact me if you need any information about my project (see contact details above).