Weakened Interannual Variability in the Tropical Atlantic

10 Climate variability in the Tropical Atlantic is complex with strong 11 ocean-atmosphere coupling, where the sea surface temperature (SST) 12 variability impacts the hydroclimate of the surrounding continents. We 13 observe a decrease in the variability of the Tropical Atlantic after 1970 14 in both CMIP6 models and observations. Most of the Tropical Atlantic 15 interannual variability is explained by its equatorial (Atlantic Zonal 16 Mode, AZM) and meridional (Atlantic Meridional Mode, AMM) modes 17 of variability. The observed wind relaxation after 1970 in both the equa- 18 torial and Tropical North Atlantic (TNA) plays a role in the decreased 19 variability. Concerning the AZM, a widespread warming trend is observed 20 in the equatorial Atlantic accompanied by a weakening trend of the 21 trade winds. This drives a weakening in the Bjerknes Feedback by deep- 22 ening the thermocline in the eastern equatorial Atlantic and increasing 23 the thermal damping. Even though individually the TNA and Tropical 24 South Atlantic (TSA) show increased variability, the observed asym- 25 metric warming in the Tropical Atlantic and relaxed northeast trade 26 winds after the 70s play a role in decreasing the AMM variability. This 27 conﬁguration leads to positive Wind-Evaporation-SST (WES) feedback, 28 increasing further the TNA SST, preventing AMM from changing phases 29 as before 1970. Associated with it, the African Sahel shows a positive 30 precipitation trend and the Intertropical Convergence Zone tends to 31 shift northward, which acts on maintaining the increased precipitation. the

of the variability of this mode is due to thermodynamic processes (68±23% 80 of the variance) (Nnamchi et al, 2015), but other studies (e.g. Lübbecke et al, 81 2018, and references therein) conclude that despite the significant contribu-82 tion of thermodynamics to the variability, the dynamics and in particular the 83 Bjerknes Feedback plays the main role. 84 The AZM is tightly coupled to West African rainfall variability. The leading 85 EOF of precipitation during JJA displays an anomaly pattern that is the 86 strongest along the northern coast of the Gulf of Guinea (Chang et al, 2006). 87 A warming in the equatorial Atlantic in this season leads to a weaker West 88 African Monsoon by decreasing the sea level pressure gradient between the 89 equator and the Saharan heat low (Losada et al, 2010). A decrease in rainfall 90 variability over the coast of Guinea has also been associated with a weakening 91 of the zonal mode (Tokinaga and Xie, 2011). 92 The second mode of Atlantic SST variability is known as the Atlantic Weakened interannual variability in the Tropical Atlantic Some studies have addressed changes in Atlantic SST variability that could 124 be related to mean state changes and weakening of the feedback mechanisms. 125 It has been shown that the interannual variability of Atlantic SST zonal gra-126 dient has decreased by 48+/-13(11)% in variance for 1960-1999 (

214
The majority of the CMIP6 models represent the Tropical Atlantic variability level for at least 60% of the models ( Fig. 1a

235
The precipitation pattern associated with each mode is also represented in      represented by ERSSTv5 (Fig. 2c). However, as shown in Fig. 2c, the CMIP6 307 AMM variability underestimates that of ERRSSTv5. Similarly, it has also been 308 found that AMM variability represented by CMIP6 underestimates that of  anomaly over the TNA also shows relaxation up to 2015, with wind stress 331 anomalies from the southwest (Fig. 3c), the same period in which we observe 332 a decrease in AMM variability (Fig. 1d). The wind relaxation over the TNA and precipitation over land (Fig. 4). 361 We show that there is a relative agreement between CMIP6 MMM and Tropical Atlantic (Fig. 4a and b) plays a role in decreasing AMM long-term 387 variability (Fig. 1d).

388
The warming is associated with the southwest wind stress trend (Fig. 4a   389 and b) located mostly over the equatorial region of the Northern Hemisphere tion over the Sahel (Fig. 4a,b and 3d). It was expected a negative trend in the 431 NEB precipitation, however, it was neither consistent among CMIP6 models 432 nor for the observational data set ( Fig. 3e and 4). The precipitation trend and 433 ITCZ northward shift are also consistent with the AMM reduced variability.

434
Given the constant anomalous warming and wind relaxation in the Northern

435
Hemisphere related to the decreased AMM variability, the ITCZ shifts north-436 ward following the increased SST (Fig. 4c). As consequence, the precipitation 437 pattern is also displaced northward, resulting in increased precipitation over 438 the African Sahel (Fig. 4a) and decreased precipitation over the NEB (not 439 consistent among CMIP6 models).

440
The ITCZ tends to follow the warmer oceanic region, but the results change