Spectral broadening in pulsed-wave Doppler caused by the transit-time effect deteriorates the frequency resolution and may cause overestimation of maximum velocities in high-velocity blood flow regions and for large beam-to-flow angles. Data-adaptive spectral estimators have been shown to provide improved frequency resolution, especially for small ensemble lengths, but offer little or no improvement when the transit-time effect dominates. In this work, a method is presented that combines a data-adaptive spectral estimation method, the power spectral Capon, and 2-D tracking Doppler to enable improved frequency resolution for both high and low velocities. For each velocity, a time signal is extracted by tracking scatterers over time and space to decrease the transit-time effect, and power spectral Capon is used for spectral estimation. The method is evaluated using simulations, flow phantom recordings, and recordings from healthy and stenotic carotid arteries. Simulation results showed that the spectral width was decreased by 60% compared to 2-D tracking Doppler for velocities around 2.3 m/s using 12 time samples. The reduction was estimated to be 66% using the flow phantom results for 0.85-m/s mean velocity. A 5-dB SNR gain was observed from the in vivo results compared with Welch's method. Computer simulations confirm that in the presence of velocity gradients or out-of-plane motion, the proposed method can be used to reduce spectral broadening by requiring shorter observation windows.