Spatial prediction of potentially toxic element content in agricultural soil using digital soil mapping approaches, health risk and source distribution assessment

Prince Chapman Agyeman

Doctoral thesis

Abstract:

The thesis presented contains consolidated publications spanning from 2018 to 2022 that are partitioned into sections and intercorrelated based on the objectives and the proposed hypothesis that follow scientific protocols. The thesis uses a dataset from the Frydek Mistek district to apply various models to the intent prediction and mapping of potentially toxic elements (PTEs) in soil. Additional auxiliary datasets employed in the thesis are from Sentinel 2, Landsat 8, the digital elevation model, soil chemical properties, and visible near-infrared spectroscopy to improve the prediction of PTEs in agricultural soil. PTE namely Pb, Sb, Mn, Cr, Cd, Cr, Ni, and Mn were studied in conjunction with a series of learning algorithms and auxiliary datasets in the prediction and the mapping of PTE levels in the agricultural soils of the Frýdek-Místek district. In addition, a source distribution assessment and a health risk assessment of the study area were carried out. The Department of Soil Science and Soil Protection at the Czech University of Life Sciences (CZU), Prague, provided oversight for the entire thesis, and grant providers and co-authors are recognized in the relevant publications. Nevertheless, the study also used different PTEs datasets, including legacy datasets and preferential sampling datasets, to predict PTEs in agricultural soil on a national scale (Czech Republic). To our knowledge, no other study has investigated this, possibly due to the uncertainty of the fusion of legacy data and preferential sampling datasets, as well as ensembles, in the prediction of PTE in agricultural soil on a national scale. The manuscript for this study is currently under review. The use of preferential sampling data and legacy data in conjunction with ensemble models improved PTE prediction on a national scale. The study will facilitate the use of preferential sampling data in known polluted areas to supplement legacy data and improve prediction on a national scale. However, the composition of papers based on scientific findings obtained from various papers in this thesis suggests that there is no single modeling approach that is deemed to be the best in all studies. Below are the papers that constitute the thesis. Agyeman, P. C., Khosravi, V., Kebonye, N. M., John, K., Borůvka, L., & Vašát, R. (2022). Using spectral indices and terrain attribute datasets and their combination in the prediction of cadmium content in agricultural soil. Computers and Electronics in Agriculture, 198, 107077. Agyeman, P. C., Kingsley, J. O. H. N., Kebonye, N. M., Ofori, S., Borůvka, L., Vašát, R., & Kočárek, M. (2022). Ecological risk source distribution, uncertainty analysis, and application of geographically weighted regression cokriging for prediction of potentially toxic elements in agricultural soils. Process Safety and Environmental Protection, 164, 729-746. Agyeman, P. C., John, K., Kebonye, N. M., Borůvka, L., Vašát, R., Drábek, O., & Němeček, K. (2021). Human health risk exposure and ecological risk assessment of potentially toxic element pollution in agricultural soils in the district of Frýdek-Místek, Czech Republic: a sample location approach. Environmental Sciences Europe, 33(1), 1-25. Agyeman, P. C., Kebonye, N. M., Khosravi, V., John, K., Borůvka, L., & Vašát, R. (2023). Optimal zinc level and uncertainty quantification in agricultural soils via visible near-infrared reflectance and soil chemical properties. Journal of Environmental Management. Agyeman, P. C., John, K., Kebonye, N. M., Khosravi, V., Borůvka, L., & Vašát, R. (2023). Prediction of the concentration of antimony in agricultural soil using data fusion, terrain attributes combined with regression kriging. Environmental Pollution. Agyeman, P. C., Borůvka, L., Kebonye, N. M., Khosravi, V., John, K., Drabek O., Tejnecky V., Quantification of the optimal cadmium level in agricultural soil using legacy data, preferential sampling, Sentinel 2, La

Abstract:

ndsat 8 coupled with ensemble model. Journal of Environmental Management (Under review)

Keywords

Potential toxic elements, Digital soil mapping, Health risk, Source distribution, Remote sensing, Terrain attributes, Data fusion

Language used: English

Date on which the thesis was submitted / produced: 8. 12. 2022

Thesis defence

• Supervisor: prof. Dr. Ing. Luboš Borůvka

Czech University of Life Sciences Prague

Faculty of Agrobiology, Food and Natural Resources

Doctoral programme / field:
Agricultural Specialization / Exploitation and Conservation of Natural Resources