Coherence scanning interferometry （CSI） is widely employed for the measurement of microstructure morphology due to its non-contact nature， low detection cost， high efficiency， and robust stability. However， when measuring the bottoms of high aspect ratio microstructures using CSI， challenges arise， such as the inability of the detection light to focus and diminished detection capabilities resulting from the sample's structure. This phenomenon is attributed to aberrations induced by the modulation of light by the sample. By conducting simulations， the aberration modulation characteristics of microstructures was elucidated， thereby providing initial values for aberration correction and compensation， which enhanced the measurement capabilities of CSI. The numerical calculation methods grounded in vector diffraction efficiently characterized complex modulation phenomena， including occlusion， diffraction， multiple reflections， and scattering of detection light， which were instigated by high aspect ratio microstructures. Notably， these methods require substantial data， exhibit low computational efficiency， and pose challenges in establishing three-dimensional models. To mitigate these issues， a sample-induced aberration simulation approach that leverages both near-field and far-field methodologies was proposed in this paper. This approach employed the Finite Difference Time Domain （FDTD） technique for full-wave simulation to compute the modulation process of high aspect ratio microstructures on the detection light field within a three-dimensional framework. The near-field distribution of the detection light modulated by the microstructure was obtained， followed by transmission to the far field using the band-limited scaling angle spectrum method to assess the modulation aberration. Verification of the proposed method's accuracy was achieved through experimental measurements of modulation aberration across various samples. Simulation results further demonstrate that， under identical simulation conditions， the proposed method can enhance computational speed more than twice.