Geomorphologic risk zoning to anticipate tailings dams' hazards: A study in the Brumadinho's mining area, Minas Gerais, Brazil

Abstract

The use of tailings dams in the mining industry is recurrent and a matter of concern given the risk of collapse. The planning of tailings dam's emplacement usually attends construction design criteria and site geotechnical properties, but often neglects the risk of installing the depositional facilities in potentially unstable landscapes, namely those characterized by steep slopes and(or) high drainage densities. In order to help bridging this gap, the present study developed a framework model whereby geomorphologic vulnerability is assessed by a set of morphometric parameters (e.g., drainage density; relief ratio; roughness coefficient). Using the Ribeirão Ferro-Carvão micro-basin (3265.16 ha) as test site, where six dams currently receive tailings from the mining of iron-ore deposits in the Brumadinho region (Minas Gerais, Brazil) and one has collapsed in 25 January 2019 (the B1 dam of Córrego do Feijão mine of Vale, S.A.), the risk of dam instability derived from geomorphologic vulnerability was assessed and alternative suitable locations were highlighted when applicable. The results made evident the location of five dams (including the collapsed B1) in high-risk regions and two in low-risk regions, which is preoccupying. The alternative locations represent 58 % of Ribeirão Ferro-Carvão micro-basin, which is a reasonable and workable share. Overall, the study exposed the fragility related with tailings dams' geography, which is not restricted to the studied micro-basin, because dozens of active tailings dams exist in the parent basin (the Paraopeba River basin) that can also be vulnerable to geomorphologically-dependent hydrologic hazards such as intensive erosion, valley incision or flash floods. Attention to this issue is therefore urgent to prevent future tragedies related with tailings dams' breaks, in the Paraopeba River basin or elsewhere, using the proposed framework model as guide.