Topography, or depth variation, of certain interfaces in the solid Earth can provide important insights into the dynamics of our planet interior. Although the intermediate- and long-range topographic variation of the 660-kilometer boundary between Earth’s upper and lower mantle is well studied, small-scale measurements are far more challenging. We found a surprising amount of topography at short length scale along the 660-kilometer boundary in certain regions using scattered P'P' seismic waves. Our observations required chemical layering in regions with high short-scale roughness. By contrast, we did not see such small-scale topography along the 410-kilometer boundary in the upper mantle. Our findings support the concept of partially blocked or imperfect circulation between the upper and lower mantle.

One interpretation of […] is very weak small-scale topography on a sharp 410-km discontinuity. This interpretation implies that the 410-km discontinuity is a *pure phase-transition boundary*, which only has topographic variations at large and intermediate scales mostly caused by smoothly changing thermal structures.

Small-scale topography of the 660-km boundary, or a less likely thin layer of volumetric heterogeneities, would best be explained with a *chemical origin* (and not only a phase transition). A gravitationally stable garnetite layer above the 660-km interface, due to oceanic crust accumulation from currently subducting and/or ancient slabs, is possible. Some regions lack small-scale topography of the 660-km interface, implying a globally discontinuous chemical layer. Our observations support simulations that describe subducting slabs as transient features of the transition zone, which eventually penetrate into the lower mantle. They also support a picture of partially blocked upper- to lower-mantle circulation.