Abstract
Environmental pollution by heavy metals and persistent organic pollutants across the North East of England reflects a century of coal mining, petroleum processing, and intensive horticulture superimposed on reclaimed and repurposed landscapes. Traditional point-based soil sampling often fails to capture geochemical variability across large land areas, creating a need for rapid, landscape-scale assessment methods. This study assesses pollution across three North-East England sites: an allotment, a former petroleum facility, and Weetslade Country Park. Using UAV multispectral imaging, ED XRF, metabolomics, and metagenomics, it integrates remote sensing and laboratory analyses to evaluate pollution concentration and environmental impacts from historical coal mining and petroleum processing.A total of 31 surface soil samples were collected from two sub-sites within Greenside, subdivided into KEN (20 samples, continuous allotment use) and LOR (11 samples, mixed gardening and livestock). Additionally, 45 soil samples were collected from Teesside Terminal, Bed 3 of a former coal washing facility, including 36 from Site A (petroleum reprocessing) and 9 from Site B (former storage depot). At Weetslade Country Park, 137 samples were obtained with 72 from the Top site and 65 from the Bottom. All surface soils (0 -10 cm depth) were collected at an estimated mass of 100 g per sample with a stainless steel hand trowel, and precise GPS coordinates was recorded for accuracy and traceability. Soil samples underwent ED XRF and HILIC LC MS analysis for metals and metabolites, alongside CEC, pH, organic matter, and metagenomics profiling. UAV multispectral imaging with vegetation indices (NDVI, SAVI, GNDVI, RE NDVI, TSAVI, MSAVI) provided near laboratory spectral data. Image stitching and spatial analysis were performed using Agisoft Metashape Professional and ArcGIS software.
Energy dispersive X ray fluorescence (ED XRF) data for 28 elements were analysed using Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS DA), identifying 12 key elements (Al, Si, K, V, Mn, Fe, Cu, As, Sr, Ba, Ce, Ni). Quantification, corrected via certified reference materials, showed strong agreement with standards. Soil physicochemical properties were similar between sites, except total organic carbon (TOC), which were higher at LOR. KEN site exhibited moderate variability, with Si at 164,564 ± 46,675 μg/g and Fe at 68,395 ± 5,943 μg/g, while Cu and Ni averaged 98.1 ± 43.8 μg/g and 68.5 ± 5.2 μg/g. LOR site showed higher, more uniform Si (232,400 μg/g) and Fe (127,636 μg/g). The study revealed moderate to strong correlations (R² ≈ 0.3–0.6) between vegetation indices and metals, notably Si, Fe, Cu, and Ni with NDVI, GNDVI, and Red Edge NDVI capturing strong metal signals. Regression analyses (R² > 0.7) showed plant metabolites correlated with chlorophyll-sensitive indices, soil metabolites with soil-adjusted indices, indicating healthier vegetation in LOR than KEN. At petroleum site A and B, mean elemental concentrations (μg/g) showed elevated Ba, V, Pb, Zn, Sr, and Zr, with lower As, Ni, and Y. Metabolomics identified 91 metabolites, including PAHs, plasticisers, PFOS, and pesticides, indicating fossil fuel and industrial influence. VIP scoring highlighted 25 key metabolites. Remote sensing revealed Site A with moderate vegetation health, while Site B exhibited higher vegetation index (VI) values, suggesting comparatively healthier cover. Combined elemental, metabolomic, and UAV MSI data indicate distinct pollution profiles between sites, with Site A reflecting more diverse industrial pollutants and B exhibited relatively higher VI values, indicating comparatively healthier vegetation cover.
At Weetslade Country Park, the top site exhibited significantly higher lead (Pb) concentrations (93.9 ± 8.2 μg/g) than the bottom (58.4 ± 12.0 μg/g), reflecting the immobility of legacy coal-mining contaminants. Conversely, Na and Mg were higher in the bottom site (Na: 2050 ± 430 μg/g; Mg: 4270-7153 μg/g) than in the top (Na: 1204 μg/g; Mg: 3865-5465 μg/g). Soil pH also differed, with the top site being acidic (4.0 ± 0.6) and the bottom more neutral (6.2 ± 0.8). Metabolomic reveals seven natural product derived compounds in the top site, such as 7,2’-dihydroxyflavanone, chrysoeriol, and D-glucose, etc. indicating higher biological activity, whereas the bottom was dominated by industrial compounds (dibutyl sebacate and diethyl phthalate), reflecting its brownfield legacy. Microbial communities were also distinct: Actinomycota dominated the top site, suggesting fertile conditions, while Betaproteobacteria prevailed in the bottom, linked to pH buffering. Pseudomonas species were abundant in both, indicating active bioremediation, and the presence of thermodesulfobacteria and thermophilic fungi highlighted the site’s colliery history and adaptation to geothermally stressed subsurface conditions.
Overall, this integrated UAV multispectral, and laboratory analytical approach demonstrates a rapid, cost effective, and scalable method for environmental diagnostics on legacy polluted land. By combining elemental, metabolomic, and microbial profiling with remote sensing, the framework enables robust assessment of pollution patterns, ecological health, and land use history. Future incorporation of multispectral time series monitoring and expanded molecular markers will enhance the ability to track pollutants dynamics, guide bioremediation efforts, and support sustainable land-management strategies for post-industrial landscapes.
| Date of Award | 28 May 2026 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Justin Perry (Supervisor), Edmond Sanganyado (Supervisor), Kate Randall (Supervisor) & Kate Nicholson (Supervisor) |
Keywords
- Correlations between vegetation indices and metals
- Legacy coal-mining contaminants
- Distributions and impact of microbial organisms
- Image stitching and spatial analysis
- Geographical distribution of metals, soil properties and organic metabolites
Cite this
- Standard