GIS-based advanced spatial analysis of total organic carbon and potentially toxic elements in European agricultural and Irish soils
Xu, Haofan
Xu, Haofan
Loading...
Publication Date
2022-03-01
Keywords
Type
Thesis
Downloads
Citation
Abstract
With the increasing availability of data in environmental geochemistry, one of the biggest challenges is to extract useful knowledge and interpretable information from large and diverse data sources. The unprecedented volume and complexity of datasets make it difficult to rely on traditional tools for data analysis, which requires the applications and development of GIS-based spatial techniques. In this thesis, four advanced spatial analysis and machine learning (ML) techniques: (1) hot spot analysis (Get-is Ord Gi*); (2) Geographically weighted regression (GWR); (3) K-means clustering analysis; and (4) Geographically Weighted Pearson Correlation Coefficient (GWPCC) were deployed to investigate the spatial patterns and to extract hidden information in large-scale datasets. The total organic carbon (TOC) and potentially toxic elements (PTEs) were studied based on datasets of GEochemical Mapping of Agricultural Soil (GEMAS) of EuroGeoSurveys and Tellus of Geological Survey of Ireland. On the one hand, the TOC contents are receiving increasing attention in agricultural soils as an important indicator of soil nutrient, not only due to their close relationship with soil fertility, but also with carbon dioxide (CO2) in the atmosphere. On the other hand, the advanced spatial techniques played important roles to evaluate concentrations and spatial variation of PTEs affected by multiple influencing factors from natural and anthropogenic sources. These studies provide demonstrations of applications of these advanced analytical techniques as possible solutions to the challenges of data analytics in the big data era. (1) The hot spot analysis was performed on a total of 2,108 agricultural soil samples based on GEMAS data and revealed an overall negative correlation between TOC and pH, which was in line with the general relationship between these two variables. However, a ‘special’ feature of co-existence of comparatively low TOC and pH values was also identified in north-central Europe. It has been found that these ‘special’ patterns are strongly related to the high concentration of quartz (SiO2) in the coarse-textured glacial sediments in north-central Europe. (2) The GWR further explored the spatially varying relationships between TOC and pH based on the GEMAS data, with more than 50% original negative relationship changed to positive at the continental level. The significant positive correlations clustered in central-eastern Europe, while negative correlations were observed mainly in northern Europe. Mixed relationships occurred in southern Europe. Such results further highlighted the influences of the extensive occurrence of quartz-rich soils and climate factors on the ‘special’ positive correlations. In addition, anthropogenic inputs also interfered the relationships in the mixed southern European areas. (3) The integration of hot spot analysis and K-means clustering analysis was applied to investigate the spatial patterns for 15 PTEs and associations with their controlling factors based on the Tellus data under the complicated geological background of Northern Ireland (NI). The spatial clustering patterns for the 15 PTEs from hot spot analysis and hidden patterns of 6,862 soil samples from K-means clustering were consistent with each other, highlighting the dominant control of peat and basalt in the topsoil of Northern Ireland. (4) The GWPCC found that the relationships between lead (Pb) and aluminium (Al) are spatially varying, with both positive and negative correlations in the topsoil of northern half of Ireland based. The ‘special’ negative correlations were observed in more than 35% of the whole study area, mainly clustered in the north-eastern and western Ireland. The positive correlations were observed in the midlands. Mixed relationships of both negative and positive correlations occurred in the eastern coastal areas. The majority of negative correlation patterns showed clear association with blanket peat, which can be attribute to long-distance transportation of Pb from atmospheric deposition. The main scientific contributions to the advancements in environmental geochemical studies of this research include the following: (1) identified a ‘special’ feature of positive relationship of low TOC contents and low pH values in the north-central Europe; (2) introduced the topic of ‘spatially varying relationships between TOC and pH’ which provide added value and clarification to the understanding of the controversy of their complicated relationship in the literature; (3) provided latest understanding and classification of 15 PTEs in the topsoil of NI to enhance the current knowledge of their controlling factors under the complicated geological background; (4) proved and observed the spatially varying relationships between Pb and Al which are associated with atmospheric deposition and anthropogenic activities. Overall, these novel findings indicated that the spatial techniques have strong efficiency in processing large-scale datasets, providing demonstration and evidence for the application of GIS-based advanced spatial analysis on identification of the hidden spatial patterns for TOC and PTEs in the topsoil and to associate them with related influencing factors. These analytical results enhanced the current knowledge for soil management and risk assessment, and can be applied in environmental studies elsewhere.
Funder
Publisher
NUI Galway