17
In the tropical Atlantic, a number of seasonal processes in
the coupled atmosphere-ocean system produce a climate
state that allows the Bjerknes feedback to operate during
early boreal summer. Although we explain the processes in a
sequential manner below, note that clear causalities are hard
to establish in a coupled system. Different aspects of the phenomenon – here: the northward movement of the ITCZ and
the development of the Atlantic cold tongue – cannot be disentangled from each other. Neither does the ITCZ move
north because of the cold tongue development, nor does the
cold tongue develop because the ITCZ moves north. Rather,
both phenomena co-occur as manifestations of the same coupled phenomenon.
One key ingredient of the equatorial Atlantic seasonal
cycle is the northward migration of the marine ITCZ (Xie
and Philander 1994). In boreal spring, the ITCZ is in its
southernmost position. The trade wind regimes of both hemispheres converge close to the equator and produce weak
equatorial surface winds. When the ITCZ moves north in late
boreal spring, the southern hemisphere trade winds cross the
equator. Starting in March–April, surface winds intensify
(illustrated by an increase in magnitude in Fig. 7a) and contribute to enhanced equatorial upwelling.
The spring strengthening of western equatorial zonal surface winds enhances the zonal gradient in upper ocean heat
content. Strong easterly winds push the surface waters more
efficiently towards the western warm pool, thinning out the
warm surface layer in the eastern ocean basin and transporting the cooling signal westward. As a result, cold subsurface
water lodges closer to the ocean surface. This background
state requires very little subsurface water to be mixed into
the surface layer to produce a substantial cooling. The western equatorial zonal spring winds “precondition” the eastern
equatorial Atlantic for the formation of the cold tongue (e.g.,
Merle 1980; Okumura and Xie 2006; Grodsky et al. 2008;
Hormann and Brandt 2009; Marin et al. 2009).
In concert with the development of the first seasonal cooling signals in May and June, the West African monsoon sets
in (e.g., Okumura and Xie 2004; Brandt et al. 2011b; Caniaux
et al. 2011; Giannini et al. 2003). From an atmospheric perspective, the monsoon onset is characterized by accelerating
southeasterly surface winds in the Gulf of Guinea in late
boreal spring. The strengthening meridional component of
these winds enhances upwelling slightly to the south of the
equator, and downwelling slightly to the north. The intensified upwelling provides additional initial cooling to the eastern equatorial region by mixing colder subsurface water into
the warm surface layer. From the ocean perspective, on the
other hand, cooling SSTs in the eastern equatorial Atlantic
intensify the southerly winds in the Gulf of Guinea, which in
turn contributes to the northward migration of convection
and rainfall associated with the West African monsoon
(Okumura and Xie 2004).
Lastly, oceanic processes contribute to the formation of
the cold tongue. A number of studies found that vertical mixing at the base of the surface layer – where temperature gradients are strongest – seasonally varies in strength (e.g.,
Hazeleger and Haarsma 2005; Jouanno et al. 2011; Hummels
et al. 2013, 2014). A likely explanation for this is that the
intensities of the westward surface current and the eastward
equatorial undercurrent vary over the course of the year.
When the relative velocities of the two currents are strong,
the vertical velocity shear at their boundary increases,
10
and
frictional processes mix colder subsurface water into the
warm surface layer. Figure 9 illustrates both the spring state
of the tropical Atlantic and the basic processes that produce
the first cooling signals in early boreal summer.
The net effect of these interacting processes – the northward migration of the ITCZ and the associated strengthening
of the southern hemisphere trade winds on the equator, the
thinning of the of eastern equatorial surface layer, the
enhanced upwelling along the equator and especially in the
cold tongue region, and the increased mixing at the base of
the surface mixed layer – is that the first cold anomalies
develop in the eastern equatorial Atlantic in late April. The
atmosphere in turn reacts to the cold anomalies, and the
Bjerknes feedback sets in. Starting in May, it lends additional growth to the cold tongue (Burls et al. 2011). In
August, the seasonally active Bjerknes feedback loop breaks
down (Dippe et al. 2017) and a more moderate warming sets
in. In the absence of the Bjerknes feedback the cold tongue
can no longer be maintained and dissolves, due to mixing
processes in the ocean and surface heat exchange with the
atmosphere.
Many models struggle to simulate a seasonally active
Bjerknes feedback that is comparable to observations in both
strength and seasonality. Richter and Xie (2008) pointed out
that model performance with respect to the Atlantic Bjerknes
feedback is quite diverse between models that participated in
the Coupled Model Intercomparison Project, Phase 3
(CMIP3, Meehl et al. 2007). Likewise, Deppenmeier et al.
(2016) found systematic weaknesses in the CMIP5 models.
For example, many models displace the Atlantic warm pool
towards the central equatorial Atlantic (Chang et al. 2007;
Richter and Xie 2008; Liu et al. 2013). This displacement is
a consequence of the westerly wind bias in the western equatorial Atlantic (Wahl et al. 2011; Richter et al. 2012, 2014b).
Figure 7 illustrates for the KCM that the spring winds are
much weaker in the model than in observations. Consequently,
the surface wind stress is not sufficient to pile up warm sur10 “Velocity shear” is a different term for “velocity gradient”. A flow is
sheared when different layers of the flow have different velocities.
Depending on the magnitude of the shear and the viscosity of the fluid,
the shear produces local turbulence and mixing due to frictional processes within the fluid. If no turbulence occurs, the flow is called
“laminar”.
Can Climate Models Simulate the Observed Strong Summer Surface Cooling in the Equatorial Atlantic?
In the tropical Atlantic, a number of seasonal processes in
the coupled atmosphere-ocean system produce a climate
state that allows the Bjerknes feedback to operate during
early boreal summer. Although we explain the processes in a
sequential manner below, note that clear causalities are hard
to establish in a coupled system. Different aspects of the phenomenon – here: the northward movement of the ITCZ and
the development of the Atlantic cold tongue – cannot be disentangled from each other. Neither does the ITCZ move
north because of the cold tongue development, nor does the
cold tongue develop because the ITCZ moves north. Rather,
both phenomena co-occur as manifestations of the same coupled phenomenon.
One key ingredient of the equatorial Atlantic seasonal
cycle is the northward migration of the marine ITCZ (Xie
and Philander 1994). In boreal spring, the ITCZ is in its
southernmost position. The trade wind regimes of both hemispheres converge close to the equator and produce weak
equatorial surface winds. When the ITCZ moves north in late
boreal spring, the southern hemisphere trade winds cross the
equator. Starting in March–April, surface winds intensify
(illustrated by an increase in magnitude in Fig. 7a) and contribute to enhanced equatorial upwelling.
The spring strengthening of western equatorial zonal surface winds enhances the zonal gradient in upper ocean heat
content. Strong easterly winds push the surface waters more
efficiently towards the western warm pool, thinning out the
warm surface layer in the eastern ocean basin and transporting the cooling signal westward. As a result, cold subsurface
water lodges closer to the ocean surface. This background
state requires very little subsurface water to be mixed into
the surface layer to produce a substantial cooling. The western equatorial zonal spring winds “precondition” the eastern
equatorial Atlantic for the formation of the cold tongue (e.g.,
Merle 1980; Okumura and Xie 2006; Grodsky et al. 2008;
Hormann and Brandt 2009; Marin et al. 2009).
In concert with the development of the first seasonal cooling signals in May and June, the West African monsoon sets
in (e.g., Okumura and Xie 2004; Brandt et al. 2011b; Caniaux
et al. 2011; Giannini et al. 2003). From an atmospheric perspective, the monsoon onset is characterized by accelerating
southeasterly surface winds in the Gulf of Guinea in late
boreal spring. The strengthening meridional component of
these winds enhances upwelling slightly to the south of the
equator, and downwelling slightly to the north. The intensified upwelling provides additional initial cooling to the eastern equatorial region by mixing colder subsurface water into
the warm surface layer. From the ocean perspective, on the
other hand, cooling SSTs in the eastern equatorial Atlantic
intensify the southerly winds in the Gulf of Guinea, which in
turn contributes to the northward migration of convection
and rainfall associated with the West African monsoon
(Okumura and Xie 2004).
Lastly, oceanic processes contribute to the formation of
the cold tongue. A number of studies found that vertical mixing at the base of the surface layer – where temperature gradients are strongest – seasonally varies in strength (e.g.,
Hazeleger and Haarsma 2005; Jouanno et al. 2011; Hummels
et al. 2013, 2014). A likely explanation for this is that the
intensities of the westward surface current and the eastward
equatorial undercurrent vary over the course of the year.
When the relative velocities of the two currents are strong,
the vertical velocity shear at their boundary increases,
10
and
frictional processes mix colder subsurface water into the
warm surface layer. Figure 9 illustrates both the spring state
of the tropical Atlantic and the basic processes that produce
the first cooling signals in early boreal summer.
The net effect of these interacting processes – the northward migration of the ITCZ and the associated strengthening
of the southern hemisphere trade winds on the equator, the
thinning of the of eastern equatorial surface layer, the
enhanced upwelling along the equator and especially in the
cold tongue region, and the increased mixing at the base of
the surface mixed layer – is that the first cold anomalies
develop in the eastern equatorial Atlantic in late April. The
atmosphere in turn reacts to the cold anomalies, and the
Bjerknes feedback sets in. Starting in May, it lends additional growth to the cold tongue (Burls et al. 2011). In
August, the seasonally active Bjerknes feedback loop breaks
down (Dippe et al. 2017) and a more moderate warming sets
in. In the absence of the Bjerknes feedback the cold tongue
can no longer be maintained and dissolves, due to mixing
processes in the ocean and surface heat exchange with the
atmosphere.
Many models struggle to simulate a seasonally active
Bjerknes feedback that is comparable to observations in both
strength and seasonality. Richter and Xie (2008) pointed out
that model performance with respect to the Atlantic Bjerknes
feedback is quite diverse between models that participated in
the Coupled Model Intercomparison Project, Phase 3
(CMIP3, Meehl et al. 2007). Likewise, Deppenmeier et al.
(2016) found systematic weaknesses in the CMIP5 models.
For example, many models displace the Atlantic warm pool
towards the central equatorial Atlantic (Chang et al. 2007;
Richter and Xie 2008; Liu et al. 2013). This displacement is
a consequence of the westerly wind bias in the western equatorial Atlantic (Wahl et al. 2011; Richter et al. 2012, 2014b).
Figure 7 illustrates for the KCM that the spring winds are
much weaker in the model than in observations. Consequently,
the surface wind stress is not sufficient to pile up warm sur10 “Velocity shear” is a different term for “velocity gradient”. A flow is
sheared when different layers of the flow have different velocities.
Depending on the magnitude of the shear and the viscosity of the fluid,
the shear produces local turbulence and mixing due to frictional processes within the fluid. If no turbulence occurs, the flow is called
“laminar”.
Can Climate Models Simulate the Observed Strong Summer Surface Cooling in the Equatorial Atlantic?