Severity of Double-ITCZ Tied to Land Surface Temperature in Atmospheric Models

Rainfall on Earth is asymmetric about the equator. Much more rain falls in the NH than in the SH with the most intense rain locating within a narrow belt about 6 degree north of the equator, namely, the intertropical convergence zone (ITCZ). The asymmetry of rainfall is underestimated in global climate models (GCMs), which commonly simulate a pair of symmetric precipitation bands, one in each hemisphere. This spurious double-ITCZ is perhaps the most significant and most persistent bias of GCMs. It can not only directly affect the tropical climate, such as the Hadley circulation and El Nino, but also impact mid-latitude weather and climate through atmospheric teleconnection. Despite effort from modelling centers around the world, there has been little progress in reducing this bias over the past two decades.

The double-ITCZ problem has been extensively studied in the context of the atmosphere-ocean system. Yet, there is still no consensus regarding the source of the bias. In this study, we demonstrate that land surface temperature, which is often considered as passive to ocean changes and thus widely overlooked in the literature, plays an important role in shaping tropical climate through extratropic-tropic interaction. We first show that the inter-model variability in severity of the double-ITCZ bias is closely related to land surface temperature biases, which can be further traced back to those in the Atmosphere Model Intercomparison Project (AMIP) simulations. In particular, cooler land temperature is related to a more sever double-ITCZ, with warmer SST coupled with weakened southeasterly trades. By perturbing land temperature in models, we demonstrate that cooler land can indeed lead to a more severe double-ITCZ bias by inducing the above coupled SST-trade wind pattern. Thus, land proves not passive as previously thought. Realizing this land effect substantially narrows the spectrum of root causes for the double-ITCZ problem and provides practical guide for reducing the bias in model development.