Calculating the correlations between tailings ponds and the hydraulic characteristics of floods may be important for determining the directions for prevention strategies. The current paper examines the article “Floods from Tailings Dam Failures” by M. Rico, G. Benito, and A. Diez-Herrero (2008). The authors systematize all the available information regarding the occurrence of tailings dam failures. In their calculations they use correlations between the hydraulic characteristics of floods and tailings ponds parameters.
The authors have determined that tailings’ volume stored is closely correlated with the volume of tailings’ outflow (Rico, Benito & Diez-Herrero 2008, p. 79). It demonstrates that numerous characteristics are interrelated, even though it is difficult to establish direct logical relationships between them. Rico, Benito, and Diez-Herrero (2008) have constructed the envelope curve showing the maximum distance negatively affected by the spill. In this way, it is possible to specify the expected level of threats that may allow selecting corresponding counterbalancing measures. The authors have also constructed a number of regression models that demonstrate average relationships between the dependent and independent variables.
The determined regression equations demonstrate general relationships between parameters. The degree of statistical significance shows the reliability of the obtained equations. At the same time, Rico, Benito, and Diez-Herrero (2008) correctly point out that “the application of the described regression equations for prediction purposes needs to be treated with caution and with support of on-site measurement and observations” (p. 79). It means that even highly statistically significant regression equations may be inapplicable in particular situations if some additional unknown factors exist.
The article is well-structured. In the beginning, the authors provide the definitions of dams and tailings dams as well as other related characteristics in order to explain the essence of research for the general public. The authors have constructed a large database of all available statistical information. The process of systematization and creation of the given database is complex as a large fraction of the entire data set is significantly biased.
Rico, Benito, and Diez-Herrero (2008) have statistically proved that dam height and reservoir volume are two main factors influencing “the magnitude of dam failure hydrographs” (p. 80). The authors have supported their conclusions by a number of statistical equations and regression coefficients. Rico, Benito, and Diez-Herrero (2008) explicitly state that these equations may be used for prediction purposes. Thus, their study has direct practical applicability. They propose to use regression equations primarily for assessing key sources of risks at dams.
This proposal is relevant, but the average level of risks may not closely correlate with the degree of risk in each particular case. Therefore, the majority of proposals and equations cannot be directly applied. It is necessary to take into account the conditions and characteristics of a given case in order to adjust the regression equation accordingly. The estimated statistical significance of various models and coefficients allows understanding the exact degree of reliability of the use of a given indicator.
In general, the study is provided at a high level. All authors’ claims and proposals are supported by corresponding statistical evidences. Moreover, they efficiently combine statistical models with theoretical explanations and practical recommendations. It seems that the amount of theoretical explanations could be higher as the correlation analysis of tailings dam failures does not demonstrate the exact pattern for correct interpretation.
The authors do not use any specific theoretical framework in their analysis but arrive at reliable conclusions and recommendations through evaluating a large number of indirect empirical facts. Another important point is the recognition of some limitations in the study. The authors state that regression equations require additional on-site support and point out that prediction will demonstrate only the average level of risks.
Therefore, the main value of this research is its practical significance as it may be used for prevention purposes. At the same time, the authors do not explicitly specify what response strategies should be applied in various situations. For example, it is possible to predict the main sources of risks on the basis of their models; however, they do not state how these risks may be addressed. On the one hand, it may be possible to use independent variables from the regression equations as the predictors for interventions that may be organized. On the other hand, the developed equations do not adequately deal with the specific conditions of each particular case. Therefore, the effect of interventions may significantly differ from the expected one derived from the model. It is also reasonable to reduce the existing variations of outcomes in the regression models. It may be achieved if an additional classification is introduced. Then, it is reasonable to create a regression model for each group of events separately.
Thus, this research may be more relevant for macro-policies while it may generate highly unstable results at the micro-level. The government and other social institutions that operate at the global level may use the information about the main sources of risks for designing their policies to minimize these risks. Even though it is problematic to predict what exact regions may primarily benefit from these policies, it is evident that the entire situation in the society may substantially improve. The companies and organizations at lower levels may experience difficulties with the use of formal models as they do not generate the reliable outcomes in all possible cases. Therefore, the future research in this area should concentrate on adjusting the model to the micro-level.