Abstract. Payment of Environmental Services (PES) projects have grown our perception on the dependency of nature, yet the quantification and valuation of Environmental Services (ES) are key to evidence-base management. The modelling of the effect of deforestation in the water and sediment budgets of rural catchments usually prescribes baseline scenarios with fully pristine vegetation. Such comparison however is hardly associated with the landscape conditions where observational data is collected to provide model calibration. For example, the existence of unpaved roads and small water retention basins (containments) are potential controls of runoff and erosion in small catchments. This work shows the impact of roads and barraginhas (small sediment retention basins nearby the roads) on the water and sediment fluxes in a 12 km² catchment area in Extrema city, Brazil, which provides water for the Cantareira System's water reservoirs that supply about 50 % for the Sao Paulo megacity, and enrolled in the Water Producer Program (Water National Agency) as the first Brazilian PES project. Simulations with the InVEST model using high spatial resolution (5 m × 5 m) included the description of unpaved roads and barraginhas. Calibration used river discharge and sediment flux estimated from water turbidity measurements. Sediment concentration was estimated both with the observation and simulations, and annual comparisons seemed reasonable for mean annual estimates. Unpaved roads produced sediment export 5 times higher compared to a scenario with no roads, and potentiated the effect of barraginhas on sediment reduction. This study showed the benefit from understanding effects of representation of the landscape particularities in modelling such as the roads, which apart from affecting calibration, are important issue for providing efficient modelling of the effect of the Best Management Practices in the landscape scale. We also demonstrated usefulness of our research methodology and its possible applications on simulations of sediment concentration and streamflow in InVEST with few samples of observed data and thus quantify the impacts of land use change on hydrology in any other watershed.