Multiple Equilibria in Weak Temperature Gradient Simulations over a Land Surface

Abstract:

Precipitation over tropical land is influenced by a three-way interaction between convection, large-scale atmospheric dynamics, and the land surface. Limited-area models run under the weak temperature gradient (WTG) approximation, which provides a simple parameterization of large-scale tropical dynamics, allow these interactions to be studied at reasonable computational expense while explicitly resolving convection. Recent work highlights the promise of of WTG-constrained convection-permitting models by showing that they can reproduce the seasonal cycle of rain in the Amazon with fewer biases than global climate models. However, basic dynamical properties of WTG-constrained atmospheres over interactive surfaces remain poorly understood. In particular, should we even expect them to support a unique time-mean precipitation rate for a given set of boundary conditions? Or can they support multiple equilibria in precipitation, as is known to be the case over fixed-temperature ocean surfaces? Here, we show that WTG-constrained convection-permitting simulations with interactive surface temperatures and soil moisture can support both precipitating and non-precipitating equilibria for a single set of boundary conditions. Which equilibrium is reached depends on both the initial atmospheric moisture and the initial soil moisture. Including small-scale surface heterogeneity increases the range of initial conditions that reach a precipitating equilibrium: if precipitation is initially suppressed and the surface dries out unevenly, convection is often eventually triggered over locally dry soil. These results show that the "climates" of WTG-constrained simulations over land are not functions of boundary conditions alone, and highlight the importance of small-scale land-atmosphere interactions in determining the evolution to a stable equilibrium.