RESUMO
This study investigated the collapse of B1 mine-tailings dam that occurred in 25 January 2019 and severely affected the Brumadinho region (Minas Gerais state, Brazil) socially, economically and environmentally. As regards water resources, the event impacted the Paraopeba River in the first 155.3 km counted from the dam site, meaning nearly half the main water course downstream of B1. In the impacted sector, high concentrations of tailings-related Al, Fe, Mn, P in river sediment-tailings mixtures and water were detected, as well as changes to the reflectance of riparian forests. In the river water, the metal concentrations raised significantly above safe levels. For caution, the water management authorities declared immediate suspension of Paraopeba River as drinking water source to the Metropolitan Region of Belo Horizonte (6 million people), irrespective of representing nearly 30% of all supply. In this study, the main purpose was to assess potential links between tailings distribution, river water composition and reflectance of forest vegetation, which worked out as latent variables in regression models. The latent variables were represented by numerous physical and chemical parameters, measured 4 times in 22 sites during the dry period of 2019. The modeling results suggested the release of aluminum and phosphorus from sand fractions in the mine tailings as major cause of water contamination. The NDVI changes were interpreted as environmental deterioration. Changes in redox potential may have raised manganese concentrations in surface water further affecting the forest NDVI. Distance from the B1 dam and dissolved calcium appear to attenuate deterioration. Overall, the regressions allowed robust prognoses of environmental deterioration in the Paraopeba River under low flow conditions. More importantly, they can be transposed to similar dam ruptures helping environmental authorities to decide upon measures that can bring the affected rivers to pre-rupture conditions.
Assuntos
Monitoramento Ambiental , Poluentes Químicos da Água , Brasil , Humanos , Análise dos Mínimos Quadrados , Água , Poluentes Químicos da Água/análiseRESUMO
Cattle grazing is a major source of income across the globe, and therefore conservation of pastures is vital to society. Pasture conservation requires the full understanding of factors contributing to their degradation, which is facilitated through panoramic analyses capable to handle all factors and capture their relationships at once. In this study, Partial Least Squares - Path Modeling (PLS-PM) was used to accomplish that task. The study area was the Environmental Protection Area of Uberaba River Basin (525â¯km2), located in the state of Minas Gerais, Brazil, and extensively used for livestock pasturing (51%). The selected (15) contributing factors comprised soil characteristics (e.g., organic matter, phosphorus content), runoff indicators (e.g., percentage of sand and clay in the soil), environmental land use conflicts (deviations of actual from natural uses), stream water quality parameters (e.g., oxidation-reduction potential-ORP, turbidity), and pasture conservation indicators (extent of degraded pasture within a pre-defined buffer). These measured variables were assembled into 5 conceptual (latent) variables to form the PLS-PM model, namely Groundcover, Pasture Conservation, Surface Runoff, Environmental Land Use Conflicts and Water Quality. The results elected Groundcover as prominent contributor to Pasture Conservation, because of its largest regression (path) coefficient (ßâ¯=â¯0.984). The most influent measured variable was organic matter. Surface Runoff (ßâ¯=â¯-0.108) and Environmental Land Use Conflicts (ßâ¯=â¯-0.135) contribute to pasture degradation. The role of conflicts is, however, limited to predefined areas where the deviations of actual from natural uses are more expressive. Pasture Conservation contributes unequivocally to improved Water Quality (ßâ¯=â¯0.800), expressed as high ORP. The PLS-PM model was free from multi-collinearity problems and model fits (R2) were high. This gives us confidence to implement conservation measures and improved management techniques based on the PLS-PM results, and to transpose the model to other areas requiring pasture quality improvements.
RESUMO
Se compara la precisión en la recuperación de parámetros del Análisis de Estructura de Covarianza (ACOV) y el Modelo de Rutas mediante Mínimos Cuadrados Parciales (PLS-PM), en un modelo simple con variables manifiestas simuladas con escala ordinal de cinco puntos. Se utiliza un diseño experimental, manipulando el método de estimación, tamaño muestral, nivel de asimetría y tipo de especificación del modelo. Se valora la media de las diferencias absolutas para el modelo estructural. ACOV presenta estimaciones más precisas que PLS-PM, en distintas condiciones experimentales. Cuando se utiliza un tamaño muestral pequeño, ambas técnicas son igualmente precisas. Se sugiere utilizar ACOV frente a PLS-PM. Se desaconseja fundamentar la elección de PLS-PM frente a ACOV en la utilización de una muestra pequeña.
The accuracy on parameter recovery is compared between Structure Covariance Analysis (ACOV) and Partial Least Squares Path Modeling (PLS-PM), with simulated ordinals data with 5 points, in a simple model. An experimental design is used, controlling the estimation method, sample size, skewness level and model specification. Mean absolute differences are used to assess accuracy for the structural model. ACOV provided more accurate estimates of the structural parameters than PLS-PM in different experimental conditions. With a small sample size, both techniques are equally accurate. Using ACOV against PLS -PM is suggested. PLS choosing ACOV instead based on the use of a small sample size is not recommended.