In practical applications, the accuracy of optical remote sensing data is often reduced because of the influence of atmospheric effect. The sensitivity of Shortwave Infrared Perpendicular Water Stress Index (SPSI) to atmospheric effect is evaluated by using the data from the Moderate Resolution Imaging Spectroradiometer (MODIS), the corresponding ground reflectivity product and the atmospheric transfer model simulation. The result shows that the atmospheric effect can make the reflectivity in near infrared and shortwave infrared bands increase linearly. However, since it has less influence on the typical triangular distribution and the geometric relationship among the pixels in near infrared and shortwave infrared spectral space, the influence of atmospheric effect on soil moisture inversion can be compensated to a certain extent. Therefore, the uncertainty brought by atmospheric effect to SPSI and relative soil moisture can be ignored in practical applications and the operational process can be simplified.