To generate a significant echo a particle should have a diameter at least equal to the quarter of the ultrasound beam wavelength (i.e. for 4 MHz in water a minimum of 93 *m of diameter is required).
On the other hand, it is known that a particle as a reflector follows the flow motion faithfully, when the density is nearly equal to the fluid itself and the size is smaller than 100 *m (in water). From these facts, the particle is expected to be larger in size as a reflector and at the same time to be smaller as a tracer.
Ideally, the reflecting particle should have as different acoustic impedance from the measured liquid, as possible (or as practical). Since the reflecting particle size is usually much smaller than the ultrasound wave length, signal is formed by reflections from many particles, and measured profile is affected by its concentration. When concentration of particles is smaller, some points of profile may not be measured during a single US pulse. This does not mean that the accuracy in velocity value becomes lower, but that the instantaneous profile has some points missing. These points have zero value because of no reflection, or are set to zero by the algorithm when too weak a reflection is detected. When the UVP is to be applied to a configuration with low concentration of particles, the average velocity profile can be reconstructed from many profiles stored on a disk file. This can been successfully done for the stationary flow.
Usually natural particle contamination of water is sufficient for UVP Monitor measurement. If media is very clean or if you want to improve measurement, it is possible to introduce tracing particles. A little mud in hydraulic models, or stirring of bottom slurry is usually enough. Other particles (hollow glass spheres, nylon or polystyrene powder) can be used as well.