Accurate methane quantification from oil and gas sites is critical for reliable reporting and climate assessments. This study tested downwind models using point sensors. A non-standard eddy covariance (EC) failed due to instrumentation issues, while the Gaussian plume inverse model (GPIM) and backward Lagrangian stochastic (bLs) model gave more reliable results. The bLs was the most accurate for single sources, with the best performance at 15 min averaging and 5° wind sectors.