Naslov (eng)

Soil mosaic: A multi-modelling approach to understand the complex interactions that determine the spatial heterogeneity of soil health - An example from Serbia

Autor

Perović, Veljko
Mrvić, Vesna
Čakmak, Dragan
Sikirić, Biljana
Dinić, Zoran
Mitrović, Miroslava
Pavlović, Pavle

Opis (eng)

Abstract: In this study, an integrated and explainable modelling framework is applied to assess soil health and its spatial heterogeneity in Eastern Serbia. The approach combines Network Analysis (NA), Self-Organizing Maps (SOM) and the Geodetector Model (GDM) to identify key soil indicators, reveal spatial patterns and quantify the influence of environmental factors. The main objectives were: (i) to develop a scalable and transferable methodology to assess soil health, (ii) to identify the most influential natural and anthropogenic factors and (iii) to create a Soil Health Disturbance Index (SHDI) to visualise the impact at ecosystem level. A total of 2,555 soil samples were analysed using fourteen key indicators, while nine environmental variables were considered as potential influencing factors. The network analysis identified pH, cadmium (Cd) and potassium (K) as the most central elements within the soil system. SOM clustering revealed five different spatial patterns associated with dominant soil processes, while GDM results emphasised soil type (Q = 0.26) and parent material (Q = 0.24) as the strongest individual influencing factors. In particular, interactive effects such as soil type ∩ soil erosion (Q = 0.47) indicated a significant nonlinear amplification of soil health disturbances. The SHDI results showed that 26.9 % of the region is categorised as highly disturbed, mainly in urban and industrial areas. Overall, the proposed framework enhances the diagnostic capacity for soil health assessment and offers practical support for sustainable soil management and evidence-based spatial planning.

Jezik

engleski

Datum

2025

Licenca

© All rights reserved

Predmet

Soil health, Natural factors, Anthropogenic factors, Spatial heterogeneity

Deo kolekcije (1)

o:508

Institut za zemljište