<p>This study investigated the impact of using different digital elevation models (DEMs) and gravity field representation strategies on regional gravity and height anomaly recovery by the spectral enhancement method (SEM), which combines 2160-degree global geopotential model (GGM) functionals with residual terrain modeling (RTM) corrections. Seven global DEMs with 1″ and 3″ resolutions were compared, and the effect of moving average (MA) reference topography resolution was evaluated. In addition to the traditional representation strategy of the SEM (SEM_D) that directly computes GGM functionals and RTM corrections on evaluation points, two alternative strategies were introduced: SEM_S (direct computation of GGM functionals and interpolation of RTM corrections from one grid model) and SEM_I (interpolation of both components from two grid models). Key factors including grid source, grid resolution, and grid combination were systematically analyzed. Numerical tests conducted in the Colorado and Auvergne test areas demonstrated that DEM selection has a minor influence on recovering gravity anomalies and no effect on height anomalies, with 3″ DEMs preferred. An MA reference topography resolution of 6’-7’ is recommended, rather than the commonly used 5’. Grid source selection shows negligible influence on both SEM_S and SEM_I. When grid resolution is at least 15″, SEM_S and SEM_I provide gravity anomaly solutions comparable to SEM_D, while 1’ grid is sufficient for height anomalies. SEM_I using a grid combination scheme of 5’ (GGM) + 15″ (RTM) yields better gravity anomaly solutions than SEM_D and SEM_S, whereas the scheme of 5’ + 1’ is advised for height anomalies. Replacing RTM corrections linked to the MA reference topography by those from the SRTM2gravity model improves gravity anomaly results in both SEM_S and SEM_I. Additionally, downsampling the 3″ SRTM2gravity model to 15″ still provides comparable gravity anomaly solutions, and in SEM_I, a grid combination scheme of 1’ + 3″ is suggested.</p>