This study aims to develop a novel method for measuring diffusion flow velocities based on temperature gradient inversion, which is particularly effective in density-driven flow fields. The proposed method addresses the challenges faced by traditional acoustic and optical Doppler velocimeters in low-particle-concentration environments, where achieving accurate measurements of ultra-low-speed diffusion flows is difficult. The experimental setup involves generating a temperature gradient in a controlled water tank and using an optical Doppler velocimeter for flow velocity calibration. Temperature and flow velocity data are simultaneously monitored and collected. The flow velocity is then inverted using a temperature gradient model. The experimental results are analyzed, and a relationship between temperature gradient and flow velocity is established through data fitting. The experimental results demonstrate that the proposed method achieves a velocity measurement resolution of 0.1 mm/s, with the error between the inverted flow velocity and the actual flow velocity being within 6%. The temperature gradient changes and flow velocity exhibit a highly consistent trend, indicating a proportional relationship. The study also employs COMSOL Multiphysics software for simulation to validate the experimental findings, confirming the accuracy and reliability of the derived fitting formula for flow velocity inversion. The study successfully proposes a measurement method that leverages temperature gradient inversion to measure diffusion flow velocities. This method offers a new technical approach for measuring diffusion flow velocities in particle-scarce or clear water environments, complementing existing measurement techniques. It holds promise for applications in environments where traditional velocimetry methods are limited, such as in clear or particle-free waters.