Horseshoes and spiral waves: capturing the 3D flow induced by a low-mass planet analytically

TL;DR

This paper develops an analytical approach to model the 3D flow induced by a low-mass planet in a disc, improving the accuracy of 2D models by incorporating vertical structure effects and matching 3D simulation results.

Contribution

It introduces a vertical averaging method that reproduces 2D fluid equations from 3D adiabatic flow, providing a simple prescription for planetary potential treatment in 2D models.

Findings

Accurate co-orbital flow solution derived

Vertical structure predictions match 3D simulations

Lindblad torque estimates are 2-3 times lower than previous 2D results

Abstract

The key difficulty faced by 2D models for planet-disc interaction is in appropriately accounting for the impact of the disc's vertical structure on the dynamics. 3D effects are often mimicked via softening of the planet's potential; however, the planet-induced flow and torques often depend strongly on the choice of softening length. We show that for a linear adiabatic flow perturbing a vertically isothermal disc, there is a particular vertical average of the 3D equations of motion which exactly reproduces 2D fluid equations for arbitrary adiabatic index. There is a strong connection here with the Lubow-Pringle 2D mode of the disc. Correspondingly, we find a simple, general prescription for the consistent treatment of planetary potentials embedded within '2D' discs. The flow induced by a low-mass planet involves large-scale excited spiral density waves which transport angular momentum radially away from the planet, and 'horseshoe streamlines' within the co-orbital region. We derive simple linear equations governing the flow which locally capture both effects faithfully simultaneously. We present an accurate co-orbital flow solution allowing for inexpensive future study of corotation torques, and predict the vertical structure of the co-orbital flow and horseshoe region width for different values of adiabatic index, as well as the vertical dependence of the initial shock location. We find strong agreement with the flow computed in 3D numerical simulations, and with 3D one-sided Lindblad torque estimates, which are a factor of 2 to 3 times lower than values from previous 2D simulations.