Solos e Nutrição de Plantas
URI permanente para esta coleçãohttps://locus.ufv.br/handle/123456789/175
Navegar
1 resultados
Resultados da Pesquisa
Item Network Theory applied to discriminate land uses and evaluate mined areas reclamation(Universidade Federal de Viçosa, 2021-10-25) Diniz, Júnia Alencar; Soares, Emanuelle Mercês Barros; http://lattes.cnpq.br/0659723142960901Mining is an activity that deeply impacts the soil-plant system, by totally removing the vegetation cover and the superficial layers of the soil, being mandatory the recovery of these areas, back to their original condition. For this to be possible, it is necessary to know in depth the current state of the system and compare it with an appropriate reference. However, soil- plant system is highly complex, both in natural ecosystems and in agro-ecosystems degraded by mining, so we need a tool that can handle all this complexity. Network Theory has stood out in the field of complex systems as a very versatile tool and is already present in soil science, although it is still little used in the mined areas reclamation. This theory allows visualizing a complex system as a whole, transforming structural densities into clusters (communities) and enabling the identification of the elements (nodes) most relevant to the structure of relationships (links). For this reason, our aim was to investigate the potential of Network Theory to discriminate the current state of mined areas under reclamation and to quantify the similarity between these areas and their references, in two different conditions: iron mining and bauxite mining. For this, we used data on soil attributes and vegetation parameters as a basis for the construction of weighted bipartite networks, composed of two classes of nodes: area and attribute. All networks were generated with Gephi software, version 0.9.2, with layouts produced by the ForceAtlas2 energy model, which provided a very clear and intuitive interpretation of the data structure. Network Theory allowed the discrimination of areas through groupings and the identification of the most relevant attributes for their distinction. It was also possible to quantify these differences by a Similarity Index (SI) and a Relative Similarity Index (RSI), using the weights of the nodes in the complete networks and in the area- projections, respectively. For the calibration of the method, we compared our results with those obtained through Principal Component Analysis (PCA) by other authors, who worked with the same data. We conclude that Network Theory can be used to discriminate the different land uses in areas affected by mining, allowing for consistent results, comparable to those of PCA. The method developed in this work to calculate the weight of the edges in the networks proved to be very accurate, because the weights kept the proportions contained in the original data values. However, better results can be obtained by adjusting the equations to the characteristic functions of each attribute, especially to non-linear functions. We also believe that it is possible to select quality indicators using Network Theory, but this issue requires further conceptual studies. The present work can be considered an initial approach for the development of a new method of data exploration that can help in understanding the processes and in monitoring the reclamation of mined areas. Keywords: Bauxite mining. Iron mining. Soil quality. Soil-plant system. Weighted bipartite networks. PCA.