Modelagem processual de qualidade da água na Bacia Hidrográfica do Rio São Lourenço

Abstract

The objective of this study was to evaluate the performance of mesoscale process modeling of nitrogen with the Soil and Water Assessment Tool (SWAT) in the São Lourenço river basin and to discuss the main sources of the nutrient. The watershed was parameterized with land use and land cover data from the WWF-Brazil survey (2009), soils from SEPLAN-MT (2000) and a digital elevation model SRTM. Meteorological inputs including precipitation were from the ERA5-Land global grid. For estimates of diffuse sources, adjustments were made to the model according to the typical forms of management in planting and regional livestock. As punctual sources, estimates of the population in grid (IBGE) were inserted considering rates of sewage treatment efficiency, collections and punctual releases of agricultural and industrial activities. Calibration and validation of the models were carried out from four flow stations (ANA), five sedimentological stations (ANA/CPRM) and six water quality stations (UFMT). The adjustments made to the model allowed us to discuss the relevance of the densification of the monitoring network in time and space to simulate N inputs in a contribution basin upstream of the Pantanal plain, in addition to contributing to the discussion about the polluting sources that influence the quality water in the study area. To evaluate the modeled data, the metrics used were the nash sutcliffe efficiency coefficient, coefficient of determination (R²) and residual mass coefficient. For the flow estimates, there was an evolution in the model's response to the observed variations (compared to previous works in the region), however, with the occurrence of daily peaks that compromised the NSE. The nitrate estimates showed unsatisfactory performance metrics, but they showed a good spatial representation of the nutrient flow in the region, considering the Conama resolution 357/2005, all sub-basins presented values lower than the maximum established for class 2 rivers (10 mg/l N). The highest loads and concentrations were found in regions with intense urban occupation (Rondonópolis city) and, in the background, in areas of contribution with predominance of intensive agriculture. The lack of data for automated model calibration and the heterogeneous physical conditions of BCH São Lourenço proved to be limiting for the model to provide satisfactory estimates that could follow the observed spatial variations.