Some arguments have been about over the representativeness of government-run air quality monitoring stations among scholars and non-governmental organizations (NGOs). However, it is not a simple problem to move monitoring stations because of continuity of data and high cost. So it is necessary to complement the monitoring data if it do not represent the ambient air quality properly. The purpose of this study was to evaluate the representativeness of some monitoring stations using passive $NO_2$ samplers and to find a complementary method from linear regression. Two stations were chosen for the evaluation: Shinlim Station was one of the most controversial stations in Seoul and Banpo Station had the best reputation. Air qualities were surveyed at seven points around each monitoring station with consideration of land use and distance. The ratios of the average $NO_2$ levels of the areas to these at the monitoring stations were 1.59 for Shinlim Station and 1.03 for Banpo Station. The differences between the average $NO_2$ levels and those at the monitoring stations were 10.75 ppb for Shilim Station and 0.34 ppb for Banpo Station. The correlation coefficients between the two levels were 0.7668 for Shinlim and 0.7662 for Banpo. The average coefficients of determination $(R^2)$ were 0.61 for Shinlim and 0.61 for Banpo. The Shinlim Station could not represent the air quality of Shinlim-Dong good because it is located in a green area at an outskirt of Shinlim-Dong. But the Banpo Station located in a central residential area of Banpo-Dong showed a fair representativeness. However, air quality turned out to be different with land use such as residential area, green area or road: the air quality data from a monitoring station located at a certain land use should not be interpreted as representing the air quality at any sites around the station. Equations to predict the average $NO_2$ levels of each area from the data from the monitoring stations were presented based on linear regression.