A battery-powered in situ Underwater Miniature PIV (UWMPIV) has been developed and deployed for field studies. Instead of generating high-energy laser pulses as in a conventional PIV system, the UWMPIV employs a low-power Continuous Wave (CW) laser (class IIIb) and an oscillating mirror (galvanometer) to generate laser sheets. In a previous version of the UWMPIV, the time between exposures of a pair of particle images, Dt, could not be reduced without loss of illumination strength. This limitation makes it unsuitable for high-speed flows. In this paper, we present a technique to solve this problem by adopting two CW lasers with different wavelength and two CCD cameras in a second-generation UWMPIV system. Several issues including optical alignment, non-uniform distribution of Dt due to the varying speed of the scanning beam and local flow velocities are discussed. The timing issue is solved through a simple calibration procedure that involves the reconstruction of maps of laser beam arrival time. Comparison of the performance between the new method and a conventional PIV system is presented. Measurements were performed in a laboratory open-channel flume. Excellent agreement was found between the new method and the standard PIV measurement in terms of the extracted vertical profiles of mean velocity, RMS fluctuation, Reynolds stress and dissipation rate of turbulent kinetic energy.