Análise da dispersão de substâncias conservativas provenientes de acidentes com cargas perigosas em pequenos e médios cursos de águas naturais

Abstract

Road accidents with dangerous goods are potential pollution originators conservative constituents in natural waterways. As an assistant tool for the assessment of environmental impact caused by the release of conservative soluble pollutants in them, quality models of water diffusion-advection are able to provide reliable predictions of the contamination levels downstream of the spill. However, such a model requires parameters obtained in the field to validate itself, among which stands out the longitudinal dispersion coefficient DL, which reflects the greater or lesser ease a pollutant is to be transported as it is dissolved in rivers. Through the application of field techniques with the use of saline and fluorescent tracers, this study determined the longitudinal dispersion coefficient for six natural waterways in the Planning Resources Unit Management UPGRH from the Paranaíba River watershed, which are intercepted at the state highway BR MG 050 MG. Its quantification was performed by using five direct methods applied to field data generated in 12 tests conducted in six sections of small and medium waterways of the sub-watersheds of the Bom Jardim stream, Uberabinha River and Jordão River. These values were compared with the DL values obtained by applying 15 empirical formulas and compared with the values obtained with the methods. Based on the results obtained, it can be stated that the method of moments was the one which overestimated the DL the most, as it showed the highest values of Reason Discrepancy( Rd ), followed by the method of Crown Concentration. Furthermore, the Chatwin method stated Rd < 0 in the majority of the results, demonstrating that this method is, among all, the one which would underestimate the most the results of DL. The method of peak of concentration was the one which presented most Rd values close to zero, demonstrating that, compared with the routing method, it is the most similar one. Note that the empirical formulas of Devens (2006) and Liu (1977) were the one which stated values of Rd closer to zero in most of the results. Besides, particularly the method of moments, produces very inaccurate results, since the long time concerning the tails of the curves significantly affect calculations of variances. Using different values of DL, it was noted that in scenarios of contamination, the Cxt curves showed distinct envelopes that significantly interfere with the interpretation of results.