Statistical analysis water quality in Flanders

The Flemish Land Agency aims for a customized implementation of the manure legislation regulations for farmers located in areas where water quality does not meet the standards of the Nitrates Directive. The underlying causes of poor water quality were studied with global statistical analysis for different geospatial aggregation levels of the variables. Causal relationships between predictor and response variables were investigated using a range of algorithms: principal component; Multivariate regression; Non-parametric causal Random forest and Multi Cross Convergent Spatial Mapping. With the acquired knowledge, recommendations were introduced for the implementation of an adjusted manure policy.

The Flemish Land Agency aims for a customized implementation of the manure legislation regulations for farmers located in areas where water quality does not meet the standards of the Nitrates Directive. The underlying causes of poor water quality were studied with global statistical analysis for different geospatial aggregation levels of the variables. Causal relationships between predictor and response variables were investigated using a range of algorithms: principal component; Multivariate regression; Non-parametric causal Random forest and Multi Cross Convergent Spatial Mapping. With the acquired knowledge, recommendations were introduced for the implementation of an adjusted manure policy. 

The objective of this study is to acquire insight into the causes of the good or bad water quality in certain areas by carrying out a statistically based analysis of available data. The study consists of 4 parts: 

  • Part 1: Exploration of the available data. The purpose of this section is to define the relevant variables, to build a uniform dataset for the continuation of the study, to describe the necessary transformations and normalizations and to acquire first insights in the data by exploring the data.
  • Part 2: Global statistical analysis of available data in Flanders. The aim of this section is to gain insight in the underlying causes of good or bad water quality by drawing up statistical models that predict the water quality and then analyse the model structure and performance of the models. 
  • Part 3: Targeted statistical analysis of available data. The aim of this section is to increase the understanding of the underlying causes of good or bad water quality by continuing to build on the main conclusions of the global statistical analysis. 
  • Part 4: Recommendations for the implementation of the research results. The aim of this section is to transfer the results of the statistical analysis to recommendations on measures and instruments that should improve water quality.