Virtual source redatuming is a data-driven interferometric approach that relies on constructive and destructive interference, and as a result it is quite sensitive to input seismic trace amplitudes. Land surveys are prone to amplitude changes that are unrelated to subsurface geology (source/receiver coupling, etc.). We have determined that such variations may be particularly damaging to construct a virtual-source signal for imaging and seismic monitoring applications, and they need to be correctly compensated before satisfactory images, repeatability, and proper relative amplitudes are achieved. We examine two methods to correct for these variations: a redatuming approach based on multidimensional deconvolution and multisurvey surface-consistent (SC) scaling. Using synthetic data, we discover that the first approach can only balance time-dependent variations between repeat surveys, e.g., compensate for variable shot scaling. In contrast, a multisurvey SC approach can compensate for shot and receiver scaling within each survey and among the surveys. As a result, it eliminates redatuming artifacts, brings repeat surveys to a common amplitude level, while preserving relative amplitudes required for quantitative interpretation of 4D amplitude differences. Applying an SC approach to a land time-lapse field data set with buried receivers from Saudi Arabia, we additionally conclude that separate SC scaling of early arrivals and deep reflections may produce better image and repeatability. This is likely due to the significantly different frequency content of early arrivals and deep reflections.