The evolution of soil depths is investigated by modeling the interaction between soil production and surface erosion within a landform evolution model. An enhanced version of the landform evolution model SIBERIA that incorporates a soil evolution module is used to simulate evolving landforms and soils depths over geologic timescales. The spatial and temporal evolution of soil depths are examined at the hillslope scale. Though it is widely accepted among the geomorphology community that soil water enhances chemical, physical and biological weathering processes, its effect has not been explicitly included in published models of soil production. The main scientific questions that we address are (1) what are the implications of incorporating soil moisture dependency in the soil production function and (2) what type of soil production dynamics is needed to generate a bedrock topography that has a different spatial pattern from that of the ground surface. A range of physics for the soil production model is explored. The effect of soil moisture is included using the wetness index obtained from drainage analysis of either surface elevations or the bedrock topography. The results show that the various soil production functions that incorporate either a wetness index or subsurface flow depth based on the bedrock topography give rise to soils that self-organize with well-defined spatial patterns and bedrock elevations with spatial organization significantly different from that of the surface. The model that incorporates the influence of subsurface water on soil production is able to naturally generate a soil production rate with a maximum value for a nonzero soil depth and overcomes an inconsistency of previously published "humped" soil production models.