Accurate measurements of P/S ratios at local-regional scale are often challenged by near-distance failure and/or sparse monitoring. During June 2014 - Aug. 2016 (iMUSH project), the XD array was deployed around the Mount St. Helens (MSH, hear and after) within ~150 km of the summit. The array contains ~80 three-component broadband stations, forming a nearly uniform monitoring system that recorded 23 active shots and 407 of M>1 earthquake within the 2-year period. The determined local magnitudes for the shots range from 0.9 to 2.3. Most earthquakes are deep (>10 km) and the shots are 10m below the surface. Due to the complexity of near surface structure, waveforms of the shots have both strong P and S energy on the transverse components, leaving them identical to earthquakes. This study takes advantage of this integrate dataset to systematically evaluates potential contributing factors to the P/S ratio measurements, including: source type (i.e., explosive vs. shear), distance, depth, and station location (i.e., site effect, coverage). We also compared between different frequency ranges, window choices, methods and component choices (i.e., RTZ vs. LQT) that differentiate the shots from the earthquakes. We observed no clear dependence for P-to-S ratio over depth, azimuth nor magnitude, while other parameters could be optimized to isolate contributions from the source. For example, the P-to-S ratios increase with frequency from 4Hz to 18 Hz and shots show much larger increasements than the earthquakes. We suggest that the near-source challenges could be relieved by: 1) use narrow windows to capture early phase and avoid overlap between S-coda and surface wave, 2) use higher frequency range to enhance body wave signal-to-noise ratio and allow for enough period within narrow phase windows, 3) rotate from the RTZ system to LQT system to maximize P wave energy and/or 4) include the transverse (T) component in P-phase energy calculations. With these optimizations, our averaged ratios are consistent throughout all distances (6-160 km) and clearly separate the shots from the earthquakes. In addition, regardless of the source type, the P-to-S ratios also vary with station location, which could be attributable to site effect and local structures.