Combining results from different monitoring systems in a statistically robust manner

The EU Directive on Soil Monitoring and Resilience (SML) asks EU Memberstates to design a national soil monitoring scheme to assess soil health in their country and report the results to EU. The spatial sampling design needs to adhere to “The sampling scheme shall be a stratified random sampling optimised on the best available information on the variability of soil descriptors, and the stratification shall be based on the soil units established in accordance with Article 4(2). Sampling points related to measurements referred to in Article 9(4) may be taken into account partly or completely in the sampling scheme, regardless of their design.” as stated in Annex II Part A of the SML. This means that it is allowed to use existing monitoring campaigns and designs in a new design for the SML. This can be desirable for efficiency reasons, to maintain an existing monitoring frequency, allow trend analysis etc. A question that then needs to be answered is how points or results from different spatial sampling designs can be combined in a statistically valid and meaningful manner. To assess this, an inventory of national and regional soil monitoring campaigns in the Netherlands was performed, including their sampling design, number of points, locations, etc. This was classified to types of sampling design. A set of rules was set to choose the most appropriate statistical method, including an evaluation of appropriate statistical inference of data from various soil campaigns. A test with actual monitoring campaigns was performed to evaluate the performance of compiled methods. The results of this exercise are used to test different options or variants for designing the soil sampling design for SML implementation in the Netherlands in an assignment for the responsible ministry in the Netherlands.