This study performs both field tests and numerical simulations to investigate the detection effect of microtremor arrays under different noise source distributions. Field tests are conducted to compare the detection effects of different arrays in the presence and absence of obvious noise sources. A numerical analysis is conducted to simulate the different noise source distribution conditions and evaluate the corresponding detection effects of the different arrays. The results indicate the following observations: (1) for the circular and triangular arrays, accurate dispersion curves can be extracted, even when obvious noise sources exist near the site, earning these arrays priority in microtremor surveys; (2) for the cross-shaped arrays, accurate dispersion curves can be obtained in the absence of obvious noise sources; however, their results are affected when there are noise sources; therefore, the usage of the cross-shaped arrays must be determined based on the ambient noise conditions and the surface characteristics around the site; and (3) for the linear arrays, different orientations may lead to varying exploration results; hence, they must be used with caution in microtremor surveys. In conclusion, different noise source distributions have significant effects on the microtremor exploration results; thus, the appropriate array types must be selected according to the noise source distribution. It is suggested that circular and triangular arrays must be prioritized; cross-shaped arrays must be utilized depending on the situation; and linear arrays must be used cautiously.