Modelo de evaluación de la contaminación descargada en humedales producto de la escorrentía urbana: metales pesados en sedimentos viales

Abstract

This study presents the results of an assessment model for heavy metal pollution in the Capellanía Wetland, focusing on its correlation with total solids. Through principal component analysis, it was identified that total solids were a significant variable in relation to heavy metals, leading to the development of five linear regression models for each specific metal (chromium, copper, nickel, lead, and zinc). Subsequently, a dynamic modeling software was used to establish the connection between the wetland's hydrological behavior and the results of the linear regression models. This enables the quantification of heavy metal concentrations within the wetland, as this variable is not commonly measured in routine water quality assessments. Integrating hydrological dynamics and regression models allows for a comprehensive understanding of heavy metal pollution in the wetland. Using results from granulometric tests conducted in various locations, linear regressions were developed using R® software, obtaining coefficient of determination values ranging from 0.48% to 0.99% of explained percentage. These findings highlight the models' capacity to explain the relationship between total solids and heavy metals in the wetland. In addition to considering these relationships, the model also incorporates other relevant variables, such as the area of the wetland's contributing basin, its area, volume, depth, inflow and outflow rates, average annual precipitation, resuspension rate, sedimentation rate, and hydraulic retention time, all crucial for dynamic modeling carried out with Vensim® software, where an estimated 25.4 tons of heavy metals enter the wetland each year. To validate the model, two methods were employed. The first method involved using Vensim® software, applying the Monte Carlo method to assess sensitivity. The second method consisted of a direct comparison between real data obtained from secondary documentation and the modeled data. This comparison yielded the following statistical metrics: mean squared errors (MSE) and mean absolute errors (MAE), ranging from 0.0001 to 0.003 and from 0.6% to 3.9%, respectively. These values indicate the model's precision in predicting pollution and its variability. The combination of these validation approaches allowed for a comprehensive evaluation of the model's performance and its ability to accurately represent the dynamics of heavy metal pollution in the wetland.