A curious symmetry in the distribution of tropical precipitation

A curious symmetry in the distribution of tropical precipitation is that it rains as much over land as it does over the ocean, in both cases about 3 mm/day. This seems to suggest that the properties of the surface, being land or ocean, do not affect the partitioning of precipitation between the two — a surprising thought given that the land, in contrast to the ocean, cannot directly tap into an unlimited water reservoir. In a paper published this month in AGU Advances, Dr. Cathy Hohenegger and Prof. Bjorn Stevens exploit this land-ocean symmetry to better understand the role of land in conditioning the distribution of precipitation in the tropics.

In their study, Hohenegger and Stevens first show that the distribution of tropical precipitation is deceptively simple. To a very good degree of approximation, conceptualizing tropical precipitation by a rain belt, whose width and seasonal migration vary by longitude, reproduces the observed distribution of precipitation. By exploiting this conceptualization, the authors show that for a zonally uniform rain belt, i.e., one whose properties remain the same over land and over ocean, tropical continents would receive only 86% as much rain as the ocean. Observations reveal much larger percentages, with values generally indistinguishable from 100%. To maintain the observed equitable land/ocean distribution of rainfall, the width of the rain belts and their seasonal migration are amplified over land.

As the land is drier than the ocean, this behavior is indicative of a negative feedback between surface water storage and precipitation. Dry regions don’t only draw more precipitation from the ocean, they draw more than would be expected if they weren’t moisture limited, i.e., indistinguishable from ocean. State-of-the-art climate models behave differently, by favoring precipitation over wetter surfaces, so that precipitation is persistently suppressed over land, making land more susceptible to drying. The inability of climate models to represent the effect of the presence of land on a basic property of the climate system calls into question their ability to answer more subtle questions, such as how regional precipitation is expected to change with warming. As precipitation over land is, relative to what would be expected from its mere geometry, relatively favored, this suggests that the rainfall over tropical land may be more resilient to perturbations of the land surface, like deforestation, than many existing climate models seem to suggest.

Animation from a coupled global storm-resolving simulation run with a grid spacing of 5 km, without continent outlines. If the land significantly affects the distribution of precipitation, its outlines should be guessable merely from the global dance of clouds and winds: where is the land?