<p>Urban pollution leaks threaten people’s lives and cause environmental pollution. However, there is a lack of a lightweight algorithm that can be embedded in portable devices to estimate the location and dynamic release strength of the pollution source rapidly and accurately. This study introduced a spatiotemporally separated source term estimation method that integrated the Gaussian equation with adjoint theory, known as the lightweight adjoint method (LAM). The adjoint method was initially utilized to identify the high probability spatial region of the pollution source. The regularization method was then used to determine the temporal release characteristics of potential sources in the region. Finally, the residual function was used to determine the real source location. The adjoint method based on the validated CFD model was employed to compare with the LAM in Mock Urban Setting Test (MUST). The results show that the LAM can rapidly estimate the location and dynamic strength of different pollution sources in the urban environment. It was particularly effective for source localization under the neutral atmospheric condition, achieving a maximum localization error of 22.1 m. Additionally, it was more suitable for calculating source intensity for the pulse source, with the maximum mean absolute percentage error (MAPE) of 15.1%.</p>