It can’t satisfy the requirement of correction for response non-uniformity at systematic levelif only if the array detector for spatial heterodyne spectrometer is corrected. Traditional methods, such as irradiation with uniform source and column-flat-fielding, are not suitable for spatial heterodyne spectrometer. The article expounds convection arm-blocking method for spatial heterodyne spectrometer briefly at first. This method leads to kinds of mismatches including pixel and sub-pixel level shift and rotation in a single arm data after gluing gratings. The effect of registration accuracy of flatfielding coefficients has been analyzed for the experimental breadboard. The result shows that the registration accuracy of flatfielding coefficients needs to be better than 0.1 pixel for the breadboard. The shift at pixel level is calculated by solving the rotational degree by using logarithm-polar coordinate and phase correlation method for the requirement of registration. The shift at sub-pixel level is estimated with DFT based on matrix multiplication. The flow path of flatfielding method at systematic level is concluded. The integral condition of interferometer after actual gluing is modulated by adjusting the positions of gratings slightly. The flatfielding flow path is applied to the data acquired from the modulated interferometer after gluing. Then, the result is compared with the spectrum after the correction with totally matched signal arm data. The final result shows that the spectral deviation is 0.6% between the two spectra compared with the spectral deviation of 4.1% without correction. The accuracy of recovered spectrum after correction has been improved markedly. This can be the foundation for the follow data processing.