(a) A 1D synthetic seismogram is formed by simply convolving an embedded waveform with a reflectivity function (also called a stickogram because it is usually plotted as a series of spikes indicating the sign and magnitude of the reflectivity at successive interfaces, the variable usually being two-way traveltime). The embedded waveform is sometimes an assumed waveform (such as a Ricker wavelet) and sometimes a waveform resulting from analysis of actual seismic data (the embedded wavelet, also called the equivalent wavelet). The reflectivity function sometimes involves primary reflections only, sometimes selected multiples are added, sometimes all multiples are added. Sometimes earth-filtering effects (divergence and other attenuation effects, including frequency-dependent absorption) are also simulated. The reflectivity is usually that calculated for normal incidence from velocity and density data, but often only velocity changes are considered because density changes are unknown (or else some relationship between density and velocity is assumed).
(b) While a 1D synthetic seismogram is a single-channel convolution (in effect involving vertical travel in the assumed model only and horizontally continuous layering), often the model is varied and successive 1D traces are displayed side by side to simulate a seismic section. It is used to compare with an actual seismogram to aid in identifying events or to predict how variations in the model might appear on a seismic section.
(c) A two-dimensional synthetic seismogram allows for wave effects including reflections from dipping reflectors, diffractions, etc. Usually only 2D effects are included but occasionally true 3D effects are included. Often only coincident source and receiver are modeled, but sometimes offset-dependent effects are included, occasionally including head waves, surface waves, and other wave modes. It sometimes involves stratigraphic modeling.
(d) A 1.5D synthetic seismogram includes AVO effects. The making of a synthetic seismogram is an example of direct modeling.