This paper describes new method for selecting optimal field points in Inverse-Numerical Acoustic analysis (INA), and its application to construction of a sound source model for diesel engines. INA identifies the surface vibration of a sound source by using acoustic transfer functions and actual sound pressures measured at field points located near the sound source. When measuring sound pressures with INA, it is necessary to determine the field point arrangement. Increased field points leads to longer test and analysis time. Therefore, guidelines for selecting the field point arrangement are needed to conduct INA efficiently. The authors focused on the standard deviations of distance between sound source elements and field points and proposed a new guideline for optimal field point selection in our past study. In that study, we verified the effectiveness of this guideline using a simple plate model. In the present study, we used a new guideline for optimal selection of the field points and constructed a diesel engine sound source model, which we installed on an agricultural machine. During sound pressure measurement at the field points, the microphone spacing and distance were decided by using the new guideline. Moreover, the number of the field points was changed by analysis frequency. It was confirmed that it is possible to construct an efficient sound source model using the new guidelines. In addition, the surrounding sound pressure level of the engine was predicted using the sound source model, and confirm the accuracy of the model.