<p>The Night Light Satellite (DMSP-OLS) collects data on artificial lights emitted from Earth's surface at night. With a long-term dataset spanning 20 years, its data remains widely utilized. However, variations in sensor types, flight conditions, and the absence of internal calibration have caused significant fluctuations in digital values within the same year. To address this issue, researchers have developed models to enhance image comparability. One key solution is intermediate calibration, which in this study is categorized into two main approaches: fixed reference area methods and stable pixel methods. This study evaluates existing calibration techniques and their effectiveness over a 20-year period in the Tehran metropolitan area, with an SNDI of 9.66, using 20 models, 13 based on fixed reference area methods and 7 on stable pixel methods. These models were selected based on prior research and the study's primary objectives. Key accuracy assessment criteria include the visual evaluation of line plot convergence, reduction in SNDI values, and alignment with GDP data. Among the fixed reference area models, the most effective approaches involve using the Sicilian reference area and the F18 satellite (2010) with quadratic regression, achieving an SNDI value of 9.09 and a correlation coefficient of 0.93 with GDP. For stable pixel methods, the best-performing model removes outlier bright pixels and applies quadratic regression, resulting in an SNDI value of 9.06 and a correlation coefficient of 0.94 with GDP. When comparing these two leading approaches, both provide similar accuracy; however, fixed reference area methods are simpler to implement, whereas stable pixel methods are more automated and minimize operator error. The choice of the most suitable calibration method depends on the study’s objectives, available resources, spatial scale, and the socio-economic characteristics of the region.</p>