The combination of a low-density geochemical survey, multispectral data obtained with Unmanned Aerial Vehicle-Remote Sensing (UAV-RS), and a machine learning technique was tested in the search for a statistically robust prediction of contaminant distribution in soil and vegetation, for zones with a highly variable pollutant load. To this end, a novel methodology was devised by means of a limited geochemical study of topsoil and vegetation combined with multispectral data obtained by UAV-RS. The methodology was verified in an area affected by Hg and As contamination that typifies abandoned mining-metallurgy sites in recent decades. A broad selection of spectral indices were calculated to evaluate soil-plant system response, and four machine learning techniques (Multiple Linear Regression, Random Forest, Generalized Boosted Models, and Multivariate Adaptive Regression Spline) were tested to obtain robust statistical models. Random Forest (RF) provided the best non-biased models for As and Hg concentration in soil and vegetation, with R2 and rRMSE (%) ranging from 0.501 to 0.630 and from 180.72 to 46.31, respectively, and with acceptable values for RPD and RPIQ statistics. The prediction and mapping of contaminant content and distribution in the study area were well enough adjusted to the geochemical data and revealed superior accuracy for As than Hg, and for vegetation than topsoil. The results were more precise than those obtained in comparable studies that applied satellite or spectrometry data. In conclusion, the methodology presented emerges as a powerful tool for studies addressing soil and vegetation pollution and an alternative approach to classical geochemical studies, which are time-consuming and expensive.
In the context of soil pollution, plants suffer stress when exposed to extreme concentrations of potentially toxic elements (PTEs). The alterations to the plants caused by such stressors can be monitored by multispectral imagery in the form of vegetation indices, which can inform pollution management strategies. Here we combined geochemistry and remote sensing techniques to offer a preliminary soil pollution assessment of a vast abandoned spoil heap in the surroundings of La Soterraña mining site (Asturias, Spain). To study the soil distribution of the PTEs over time, twenty-seven soil samples were randomly collected downstream of and around the main spoil heap. Furthermore, the area was covered by an unmanned aerial vehicle (UAV) carrying a high-resolution multispectral camera with four bands (red, green, red-edge and near infrared). Multielement analysis revealed mercury and arsenic as principal pollutants. Two indices (from a database containing up to 55 indices) offered a proper correlation with the concentration of PTEs. These were: CARI2, presenting a Pearson Coefficient (PC) of 0.89 for concentrations >200 mg/kg of As; and NDVIg, PC of −0.67 for >40 mg/kg of Hg. The combined approach helps prediction of those areas susceptible to greatest pollution, thus reducing the costs of geochemical campaigns.
Soil organic carbon (SOC) is essential for maintaining ecosystem health, and its depletion is widely recognized as a key indicator of soil degradation. Activities such as mining and wildfire disturbances significantly intensify soil degradation, leading to quantitative and qualitative declines in SOC. Accurate SOC monitoring is critical, yet traditional methods are often costly and time-intensive. Advances in technologies like Unmanned Aerial Vehicles (UAVs) and satellite remote sensing (SRS) now offer efficient and scalable alternatives. Combining UAV and satellite data through machine learning (ML) techniques can improve the accuracy and spatial resolution of SOC monitoring, facilitating better soil management strategies. In this context, this study proposes a methodology that integrates geochemical data (SOC) with UAV-derived information, upscaling the UAV data to satellite platforms (GEOSAT-2 and SENTINEL-2) using ML techniques, specifically random forest (RF) algorithms. The research was conducted in two distinct environments: a reclaimed open-pit coal mine, representing a severely degraded ecosystem, and a high-altitude region prone to recurrent wildfires, both characterized by extreme environmental conditions and diverse soil properties. These scenarios provide valuable opportunities to evaluate the effects of soil degradation on SOC quality and to assess the effectiveness of advanced monitoring approaches. The RF algorithm, optimized with cross-validation (CV) techniques, consistently outperformed other models. The highest performance was achieved during the UAV-to-SENTINEL-2 upscaling, with an R2 of 0.761 and an rRMSE of 8.6%. Cross-validation mitigated overfitting and enhanced the robustness and generalizability of the models. UAV data offered high-resolution insights for localized SOC assessments, while SENTINEL-2 imagery enabled broader-scale evaluations, albeit with a smoothing effect. These findings underscore the potential of integrating UAV and satellite data with ML approaches, providing a cost-effective and scalable framework for SOC monitoring, soil management, and climate change mitigation efforts.
17 March 2009 in Environmental Modeling & Assessment
This research focuses on the identification of industrial wastelands in Langreo and the strategies employed for their regeneration. The breakdown of the industrialisation production model, characterised by the concentration of mining and iron and steel activities, introduced economic, social and landscape changes. For all these reasons, Langreo is a clear example of a shrinking city, as it presents the characteristic features that defi ne these cities, such as the loss of population, the cessation of productive activities, processes of environmental and landscape deterioration and urban transformations that are manifested by the presence of industrial wastelands and degraded spaces. The process of regeneration of these spaces is characterised by the lack of planning instruments, the absence of a comprehensive strategy for the whole area and its implementation is therefore the result of the individual initiative of each agent involved.