Plant root–soil water interactions are fundamental to vegetation–water relationships. Soil water availability and distribution impact the temporal–spatial dynamics of roots and vice versa. In the Loess Plateau (LP) of China, where semi-arid and arid climates prevail and deep loess soil dominates, drying soil layers (DSLs) have been extensively reported in artificial forestland. While the underlying mechanisms that cause DSLs remain unclear, they hypothetically involve root–soil water interactions. Although available root growth models are weak with respect to simulating the rooting depth, this study addresses the hypothesis of the involvement of root–soil water interactions in DSLs using a root growth model that simulates both the dynamic rooting depth and fine-root distribution, coupled with soil water, based on cost–benefit optimization. Evaluation of field data from an artificial black locust (Robinia pseudoacacia L.) forest site in the southern LP positively proves the model's performance. Further, a long-term simulation, forced by a 50-year climatic data series with varying precipitation, was performed to examine the DSLs. The results demonstrate that incorporating the dynamic rooting depth into the current root growth models is necessary to reproduce soil drying processes. The simulations revealed that the upper boundary of the DSLs fluctuates strongly with infiltration events, whereas the lower boundary extends successively with increasing rooting depth. Most infiltration was intercepted by the top 2.0 m layer, which was the most active zone of infiltration and root water uptake. Below this, the percentages of fine roots (5.0 %) and water uptake (6.2 %) were small but caused a persistently negative water balance and consequent DSLs. Therefore, the proposed root–water interaction approach succeeded in revealing the intrinsic properties of DSLs; their persistent extension and the lack of an opportunity for recovery from the drying state may adversely affect the implementation of artificial afforestation in this region as well as in other regions with similar climates and soils.