The solid-state mantle flow is an important factor that controls the mass and heat transfer in the solid Earth. This study aims to provide a simple picture of three-dimensional (3-D) mantle flow patterns in the sub-arc region of subduction zones based on the results of 3-D steady-state numerical models with varying subduction parameters. Here, the mantle wedge flow pattern is evaluated based on the azimuthal directions of the mantle inflow from the back-arc and the down-dip outflow. The outflow direction generally parallels the subduction direction, but the inflow direction relative to the outflow direction depends on the local subduction obliquity – the angle between the subduction direction and the strike-normal axis of the subducting slab. A change in the strike of the slab leads to a change in the obliquity and thus the inflow direction. Such change is common along curved margins as the strike of the slab tends to follow that of the margin, or vice versa. Along convex-arc-ward margins, the mantle inflow is deflected towards the region of lowest obliquity but with reduced vigor due to lower dynamic pressure gradients that partly drive the flow, resulting in a cooler mantle wedge. Along concave-arc-ward margins, the mantle inflow is deflected away from the region of lowest obliquity but with increased vigor, resulting in a hotter mantle wedge. These effects increase with decreasing radius of curvature. Along-margin change in the dip of the subducting slab also affects the inflow direction through its impact on the strike of the slab, but its effect is relatively small. We express the azimuthal inner angle between the inflow and outflow directions as a function of obliquity and apply the function to predict sub-arc mantle inflow directions in the circum-Pacific and neighboring regions. Within and among these margins, the inner angle varies over its full range of 0–180°. Most of the margins that are 1000s of kilometers in length are either straight or curved concave-arc-ward with large radii of curvature, for which small or gradual along-margin changes in the mantle inflow direction and the mantle wedge temperature are predicted. A large drop in the mantle wedge temperature by up to a couple of hundred degrees is predicted at short convex-arc-ward segments, such as at the Kuril-Japan and Bonin-Mariana junctions. The fringes of flat slab segments are curved with small radii of curvature, likely resulting in sharp lateral changes in the inflow direction and the mantle wedge temperature.

固态地幔流是控制固体地球中传质和传热的重要因素。本研究旨在根据具有不同俯冲参数的 3-D 稳态数值模型的结果，提供俯冲带亚弧区的 3-D (3-D) 地幔流动模式的简单图片。在这里，地幔楔流型是根据地幔从弧后流入和下倾流出的方位角方向来评估的。流出方向通常与俯冲方向平行，但流入方向相对于流出方向取决于局部俯冲倾角——俯冲方向与俯冲板片走向法线轴之间的角度。板坯走向的变化导致倾斜度的变化，从而导致流入方向的变化。这种变化沿弯曲边缘很常见，因为板坯的走向往往跟随边缘的走向，反之亦然。沿着凸弧向边缘，地幔流入向最低倾角区域偏转，但由于较低的动态压力梯度，部分驱动流动，从而导致较冷的地幔楔，因此活力减弱。沿着凹弧向边缘，地幔流入从最低倾角区域偏转，但强度增加，导致更热的地幔楔。这些影响随着曲率半径的减小而增加。俯冲板片倾角的沿边缘变化也通过其对板片走向的影响来影响流入方向，但其影响相对较小。我们将流入和流出方向之间的方位角内角表示为倾斜度的函数，并应用该函数来预测环太平洋和邻近地区的亚弧地幔流入方向。在这些边缘之内和之间，内角在其 0-180° 的整个范围内变化。1000 公里长的边缘大部分是直的或弯曲的凹弧形，曲率半径大，预测地幔流入方向和地幔楔温度的沿边缘变化很小或逐渐变化。预计在短凸弧段，例如千岛-日本和博宁-马里亚纳交界处，地幔楔温度将大幅下降数百度。平板段的边缘以小曲率半径弯曲，