QUARTERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
Published online 24 December 2009 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/qj.538
Diabatic heating, divergent circulation and moisture
transport in the African monsoon system
Samson Hagos* and Chidong Zhang
RSMAS, University of Miami, Miami, Florida, USA
ABSTRACT: The dynamics of the West African monsoon system is studied through the diagnosis of the roles of diabatic
heating in the divergent circulation and moisture transport. The divergent circulation is partitioned into latent-heating
and non-latent-heating (the sum of surface sensible heat flux and radiative heating) driven components based on its field
properties and its relationship with diabatic heating profiles. Roles of latent and non-latent diabatic heating in the moisture
transport of the monsoon system are thus distinguished.
The gradient in surface sensible heat flux between the Saharan heat-low and the Gulf of Guinea drives a shallow
meridional circulation, which transports moisture far into the continent on the northern side of the monsoon rain band and
thereby promotes the seasonal northward migration of monsoon precipitation. In contrast, the circulation directly associated
with latent heating is deep and the corresponding moisture convergence maximum is within the region of precipitation
and thus enhances local monsoon precipitation. Meanwhile, latent heating also induces dry air advection from the north.
The seasonal northward migration of precipitation is encouraged by neither of the two effects. On the other hand, the
divergent circulation forced by remote latent heating influences local moisture distribution through advection. Specifically
by bringing Saharan air from the north, and driving moisture to the adjacent oceans, global latent heating has an overall
c 2009 Royal Meteorological Society
drying effect over the Sahel. Copyright !
KEY WORDS
African monsoon; divergent circulation; moisture transport; diabatic heating
Received 20 November 2008; Revised 21 September 2009; Accepted 8 October 2009
1.
Introduction
The West African monsoon (WAM) system is an important component of the regional hydrological cycle on
which the livelihood of a large and growing population depends. Understanding its dynamics in general and
the roles of the various diabatic processes in particular
can help the interpretation of existing uncertainties in the
monsoon response to regional and global climate variability and change. For example, the wide range of regional
precipitation change projected by the Intergovernmental
Panel on Climate Change (IPCC) models for the twentyfirst century (Cook and Vizy, 2006) calls for a thorough
comparative assessment of the influence by external forcing and feedback mechanisms on the monsoon dynamics.
The WAM system comprises surface southerly and
westerly inflows across the Guinean and western coasts
of the continent, northeasterly and easterly outflows
aloft, and seasonally migrating rainfall from the coast
into the continent during spring and summer. Figure 1
shows 1998–2007 mean April–May–June (AMJ) and
July–August–September (JAS) winds at 925 hPa and
700 hPa from the National Centers for Environmental
Prediction NCEP-DOE or NCEP II reanalysis (see section
2 for a description of the data). The precipitation seasonal
cycle is accompanied by a boundary-layer southerly
∗
Correspondence to: Samson Hagos, Pacific Northwest National Laboratory, Richland, WA99354, USA. E-mail: samson.hagos@pnl.gov
c 2009 Royal Meteorological Society
Copyright !
flow across the coast of Guinea. Further inland this
flow turns southwesterly and meets the dry northeasterly
harmattan winds to form a discontinuity often referred
to as the intertropical front (ITF). This flow transports
moisture into the continent throughout the year, but its
northward extent has a seasonal cycle and it reaches
its northernmost latitude in August (Figure 1(a) and
(b)). Immediately after that, the precipitation peak starts
its retreat southward along with the flow and reaches
its lowest value in December and January, where the
precipitation is limited to about 8◦ N and the precipitation
peak is out in the Gulf of Guinea. Above the boundary
layer, embedded in the easterlies is a northerly return
flow, which along with the boundary-layer southerlies
constitutes a shallow meridional overturning circulation
(Figure 1(c) and (d)). This shallow meridional circulation
shares the southerly surface flow with the deep Hadleylike circulation with return flow above 500 hPa, but its
ascending branch is often further to the north (Zhang
et al., 2006). Precipitation feedbacks on this circulation
include enhancement of surface latent heat flux at the
expense of sensible heat flux and radiative cooling (by
increased surface wetness and cloudiness), and changing
the surface pressure gradient (Taylor, 2008). Furthermore,
surface re-evaporation of precipitation has also been
known to substantially contribute to the moisture and
surface heat budgets over West Africa during the rainy
season (Guichard et al., 2009; Miller et al., 2009; Timouk
et al., 2009).
412
S. HAGOS AND C. ZHANG
Figure 1. Spring (April–May–June) and summer (July–August–September) precipitation (mm/day) from TRMM 3B43 and winds at 925 and
700 hPa from the NCEPII reanalysis.
The circulation responsible for moisture transport into
the continent is associated with both surface sensible
heating and atmospheric latent heating. Hagos and Cook
(2007) and Ramel et al. (2006) showed that the abrupt
northward shift of the precipitation maximum follows
enhanced moisture convergence and condensation over
land, which in turn follow the intensification and northward movement of the African heat-low. On the other
hand, seasonal cooling of sea-surface temperature (SST)
over the Gulf of Guinea (Gu and Adler, 2004; Okumura and Xie, 2004) and topographic effects of the
Atlas–Hoggar mountains (Sultan and Janicot, 2003) have
also been suggested as playing some roles in the seasonal
cycle. But a quantitative assessment of the role of diabatic
heating in moisture transport and precipitation feedback
has yet to be made. This would provide a clearer insight
into the dynamics of the monsoon and its response to perturbations in environmental conditions such as changes in
land use, global SST, aerosols etc.
Using the relationship between diabatic heating profile
and horizontal divergence as well as field properties of
the divergent circulations, this study aims to:
c 2009 Royal Meteorological Society
Copyright !
(1) distinguish the roles of latent and non-latent diabatic heating in the meridional circulation, moisture
transport and seasonal cycle of WAM precipitation, and
(2) identify the regional controls of the moisture supply
for the WAM.
2.
Data and method
2.1. Data
Data of wind, temperature, water vapour mixing ratio,
pressure and precipitation from four reanalysis products were used. They are the ERA40 (Uppala et al.,
2005), NCEP/DOE or NCEP II (Kanamitsu et al., 2002),
NCEP/NCAR or NCEP I (Kalnay et al., 1996), and
JRA25 (Onogi et al., 2007). Six-hourly data for the period
1 January 1998 to 31 December 2007 were used for
the last three reanalysis products and daily data from 1
January 1998 to 31 December 2001 for ERA40. All diagnostics described below were applied to the four datasets.
First, issues regarding precipitation in the reanalyses
need to be addressed. Figure 2 shows a comparison of the
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
413
Figure 2. Mean seasonal cycle of precipitation (mm/day) from (a) TRMM (3B42) and (b)–(e) the four reanalyses. The vertical line represents
the approximate latitude of the coastline.
seasonal cycles of precipitation from the Tropical Rainfall
Measuring Mission (TRMM) and the four reanalyses. The
TRMM precipitation data used here (version B43: Huffman et al., 2007) are monthly means on 0.25◦ × 0.25◦
grids. All reanalysis precipitation maxima reach their
northernmost positions in August and southern positions
in March or April, in agreement with TRMM precipitation. But large discrepancies exist in both the exact
latitude and strength between reanalysis and TRMM precipitation, and among the reanalyses themselves. These
discrepancies manifest the known problems in precipitation from the reanalyses. They come from the fact that
c 2009 Royal Meteorological Society
Copyright !
precipitation is a direct product of cumulus parametrization in models for data assimilation, during which corrections are made according to observations for dynamics
and certain thermodynamic variables, but not for precipitation. Thus in the subsequent diagnostics and discussions, one has to keep in mind that an inferred dynamically consistent relationship of diabatic heating and the
circulation based on the reanalyses is not free of errors
due to model flaws.
Among others, primary deficiencies of most cumulus
parametrization schemes are their inability to produce
observed mean vertical structures of diabatic heating
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
414
S. HAGOS AND C. ZHANG
that are an aggregate of different types of individual
convective systems in an area represented by a grid
point in a climate model (Zhang and Hagos, 2009).
A comparison between diabatic heating profiles from
observations and the reanalyses (Hagos et al., 2009)
reveals that the main discrepancies among the reanalyses
resides in their reproductions of low-level heating. Our
confidence in a particular result might be built upon
the degree of its consistency among the four reanalyses,
considering the discrepancies in their precipitation and
diabatic heating. On the other hand, we cannot rule out
the possibility that the consistency comes from common
problems in their parametrizations. Our results, therefore,
represent our current knowledge on the issue under study
based on the best global data currently available.
2.2.
Diabatic heating
The total diabatic heating, also known as apparent heating
source Q1 (Yanai et al., 1973), which is the sum of
latent heating, radiative heating and surface sensible heat
fluxes, is calculated as the residual of the thermodynamic
equation on pressure surfaces in a manner similar to
that by Yanai (1973), Nigam et al. (2000) and Hagos
et al. (2009):
!
"
Cp T ∂θ
∂θ
∂θ
∂θ
Q1 =
+u
+v
+ω
,
(1)
θ
∂t
∂x
∂y
∂p
where u and v are the zonal and meridional components
# $R
Cp
with T being temperaof horizontal wind, θ = T pps
ture, p pressure, ps surface pressure, Cp the specific heat
capacity of dry air, and R the specific gas constant of
dry air. Q1 is calculated using central finite difference at
the horizontal resolution (2.5◦ × 2.5◦ ), using the 6-hourly
data (daily for ERA40) at 17 pressure levels (1000, 925,
850, 700, 600, 500, 400, 300, 250, 200, 150, 100, 70,
50, 30, 20 and 10 hPa). To reduce error due to irregular
∂θ
pressure-level spacing, ∂p
is replaced by p∂ ∂θ
ln(p) in the
∂θ
finite differencing of Eq. (1). ∂t is also estimated using
central finite difference in time. This term has a relatively
small contribution to Q1 .
Once diabatic heating Q1 is calculated, its relationship
with precipitation is used to estimate the relative contributions of latent and non-latent (surface sensible heat
flux and radiation) components at every grid point. Figure #3 shows
% the$ distribution of normalized diabatic heating Q1 |Q1 | as a function of the associated reanalysis
precipitation over the WAM region (box in Figure 1(a))
for the four reanalyses. |Q1 | is the norm of Q1 in the
statistical sense, i.e. the diabatic heating at a point is
treated as a 17-dimensional vector (for 17 levels) and the
norm is the square root of its elements. Since precipitation
spans multiple orders of magnitude, the range of precipitation between 10−2 and 102 mm day−1 is partitioned into
forty bins on a logarithmic scale to construct its probability distribution function (PDF). The mean normalized
heating profile within each bin is plotted. Dashed lines
c 2009 Royal Meteorological Society
Copyright !
mark precipitation intensity of 1, 2 and 3 mm/day, respectively. The dramatic change in diabatic heating profiles
near the 1–3 mm/day precipitation range is apparent. At
low precipitation, diabatic heating is dominated by sensible heating near the surface and radiative cooling aloft.
As precipitation increases, the diabatic heating maximum
is lifted into the mid to upper troposphere, signifying the
increased importance of latent heat release by precipitation. Based on this, diabatic heating at a point at any
instant is considered to be latent heating if the associated
precipitation is greater than a threshold value and the
point is treated as a ‘latent heating point’. Otherwise the
diabatic heating is mainly composed of sensible heat flux
and radiative cooling, and for brevity, the point is referred
to as a ‘non-latent heating point’. For consistency this
threshold value is set at 2 mm/day for all reanalyses. The
sensitivity of this classification to the threshold precipitation is tested. For all reanalyses, moving the threshold
value by 1 mm/day results in a maximum change in the
categorization of 12%, 4.5%, 2% and 7% of the latent
heating for ERA40, NCEPII, NCEPI and JRA25 respectively, but it has little effect on the conclusions drawn
from this study.
The six-hourly precipitation from NCEPII, NCEPI,
and JRA25 is accumulated precipitation, while the wind
fields from which diabatic heating is calculated are
instantaneous. Possible effects of mismatch between them
are evaluated. For this purpose, the relationship between
the average of diabatic heating at two successive time
steps and precipitation for those reanalyses is examined
(Figure 4). The fact that the relationship between diabatic
heating and precipitation discussed above still holds
suggests that the lag between instantaneous wind and
accumulated precipitation has little effect.
2.3. Circulation decomposition
We now explain how a divergent circulation can be
decomposed into a component that is related to latent
heating and another one that is not. Let r i = (xi , yi ) be an
element of a set consisting of all grid points in a domain
U , where the horizontal divergence is
Di = −
∂ωi
.
∂p
(2)
This divergence introduces velocity potential (χ) at point
r k = (xk , yk ) given by
∇ 2 χki = D i =
j =NU
&
δij D j
(3)
j =1
where δij is the Kroneker Delta defined as δij = 1 if
i = j and δij = 0 otherwise, NU is the number of points
∂
∂
+ j • ∂y
is the horizontal
in domain U , and ∇ = i • ∂x
divergence operator.
The velocity potential at point r k due to horizontal
divergence in domain L, an arbitrary subset of U with
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
415
Figure 3. Normalized diabatic heating profiles vs. precipitation intensity (mm/day) and the PDF of precipitation (in %) over the WAM domain
(the box in Figure 1(a)). The dashed lines mark 1, 2 and 3 mm/day precipitation.
(Jaluria and Torrance, 2003) over the tropical domain
(42.5◦ S to 42.5◦ N), with vanishing divergent wind at the
=NU
i=N
&L j&
lateral
boundaries. Sensitivity tests showed that moving
∇ 2 χkL =
δij D j
(4) the lateral boundaries any further poleward has no sigi=1
j =1
nificant effects on the results of the study. Once the two
The set of ‘latent heating points’ and its complement, parts of velocity potential are calculated, their respective
the set of ‘non-latent heating points’, both defined in the associated divergent winds are determined by
last subsection, constitute two subsets of U : L1 = lh and
v nonlh
= ∇ χ nonlh
(6)
div
L2 = nonlh. The total velocity potential at point r k can
be written as:
and
NL elements, is then
∇ 2 χk = ∇ 2 χknonlh + ∇ 2 χklh
(5)
where
and
are the velocity potentials due
solely to all ‘non-latent heating points’ and all ‘latent
heating points’, respectively. This approach is analogous
to the calculation of the electrical potential associated
with a set of point charges (Jackson, 1999), stream
function due to point vortices (Batchelor, 2000), or
heat conduction in the presence of point heat sources
(Selvadurai, 2000; Arfken and Weber, 2005).
Equation (3) is solved for ‘non-latent heating’ and
‘latent heating’ points using Gauss–Seidel iteration
χknonlh
χklh
c 2009 Royal Meteorological Society
Copyright !
lh
v lh
div = ∇ χ .
2.4.
(7)
Moisture convergence and transport
The total horizontal moisture divergence is divided into
divergence by divergent winds and advection by nondivergent wind:
∇ • (qv) = ∇ • (qvdiv ) + vnondiv • ∇q.
(8)
The horizontal divergence of moisture associated with
divergent wind is further divided into those associated
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
416
S. HAGOS AND C. ZHANG
Figure 4. Same as Figure 3 except diabatic heating is an average of successive times to match the forecast precipitation.
with latent and non-latent heating (sensible heating and NCEPI, NCEPII and JRA25, and daily for ERA40) then
radiative cooling), respectively, as follows
averaged into monthly data. All subsequent discussions
are based on a 10-year climatology, except 4 years for
'
(
' nonlh (
∇ • (qvdiv ) = ∇ • qvlh
+
∇
•
qv
.
(9)
ERA40.
div
div
The horizontal divergence of moisture associated with
latent heating is divided into parts associated with wind
3. Results
convergence and advection;
'
(
' lh (
lh
∇ • qvlh
(10) 3.1. Mean summer circulation patterns
div = q∇ • vdiv + (vdiv • ∇)q.
Equation (10) separates the local and remote effects of
latent heating on moisture convergence. At a ‘latent heating point’, the first term on the r.h.s. of Eq. (10) represents
moisture convergence associated with the local latent
heating and depends on the local wind convergence and
mixing ratio. The second term represents the advection of
moisture by the wind associated with the latent heating
throughout the domain. Even though latent heating and
the associated convergence at a point may be zero, there
is non-zero wind at that point due to latent heating at all
other points in the domain. In the presence of moisture
gradient, this wind introduces advection.
All the above calculations are done at the maximum temporal resolution of the reanalyses (6-hourly for
c 2009 Royal Meteorological Society
Copyright !
Diabatic processes are involved in shaping the nondivergent circulation, through vorticity dynamics (Hsien
and Cook, 2007). But their primary role in the Tropics is balancing the local vertical motion and associated
divergent circulation. In order to assess their roles in
the monsoon dynamics, the divergent and non-divergent
components of the circulation and their moisture transport are separated. The meridional circulation and moisture convergence associated with total, divergent and
non-divergent circulations from the four reanalyses are
displayed in Figure 5. In all four reanalyses, there are
roughly equal surface southerly monsoon flows in the
divergent and non-divergent components over the ocean.
The divergent component of the surface monsoon flow
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
417
Figure 5. Meridional–vertical circulation, its divergent and non-divergent components, and their moisture convergence (shading, g kg−1 day−1 ) in
summer (JAS), all averaged over 15◦ W–10◦ E. The solid and dashed straight lines mark the coastline and latitude of the precipitation maximum,
respectively.
apparently penetrates much further north than its nondivergent counterpart. Both deep and shallow meridional
circulations are primarily divergent with the low-level
northerly return flow peaking at about 700 hPa, distinctly
separated from the deep return flow with its maximum at
300 hPa. Near the surface, there is moisture divergence
over the Gulf of Guinea. Over land, there are two surface
convergence zones, one centred at the precipitation region
c 2009 Royal Meteorological Society
Copyright !
(whose maximum is marked by the vertical dashed line)
and another further north. The amount of convergence in
the secondary maximum north of the precipitation varies
among the reanalyses. It is strongest in JRA25 and weakest in ERA40. The non-divergent circulation over West
Africa is primarily zonal, and since the moisture gradient is primarily meridional, the former has very little
contribution to the total moisture convergence.
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
418
S. HAGOS AND C. ZHANG
Figure 6. Total divergent meridional–vertical circulation, its components related to latent heating and non-latent heating, and their moisture
convergence (shading, g kg−1 day−1 ) in summer (JAS), averaged over 15◦ W–10◦ E. The solid and dashed straight lines mark the coastline and
latitude of the precipitation maximum, respectively.
The divergent circulation seen in Figure 5 is further
decomposed into components related to latent heating
(LHDIV) and non-latent heating (NONLHDIV) in Figure 6 as calculated by the procedure described in the last
section. Also shown in Figure 6 is the associated moisture convergence. The two components of the divergent
meridional circulation differ in their vertical structure
and their roles in moisture transport. The latent heating
component is deep and more or less symmetric about
c 2009 Royal Meteorological Society
Copyright !
the latitude of the precipitation maximum (the dashed
line). It drives moisture convergence into the region of
precipitation. It also drives low-level dry air from the
north into the precipitation region. This pattern is apparent in all reanalyses. The shallow meridional circulation
is part of the circulation associated with non-latent heating (NONLHDIV). Its subsidence branch is over the
Gulf of Guinea. It transports moisture from the Gulf of
Guinea (note the moisture divergence there) deep into the
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
419
Figure 7. Seasonal cycle of vertically integrated moisture convergence (g kg−1 day−1 ) in the WAM by TOTAL, LHDIV and NONLHDIV winds.
The contours represent associated reanalysis precipitation (mm/day).
continent, penetrating through the precipitation region. Its
centre of surface convergence is north of the precipitation
region near the heat-low, where its ascending branch is
located. From there, its low-level northerly return flow
extends all the way through the region of precipitation to
the Gulf of Guinea.
The reason for the difference between the circulation
patterns lies in the profile of vertical velocity associated
with the diabatic heating involved. The profile of latent
c 2009 Royal Meteorological Society
Copyright !
heating (and the associated vertical velocity) is middle
heavy (Figure 3) and the associated divergence is in the
upper troposphere. The shallow meridional circulation
results from strong surface sensible heating capped by
radiative cooling, resulting in a profile that favours
horizontal divergence in the lower troposphere. In the
upper troposphere, latent heating drives a symmetric
circulation with respect to the latitude of the precipitation
maximum and produces a strong upper-level southerly
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
420
S. HAGOS AND C. ZHANG
Figure 8. Seasonal cycle of moisture convergence (g kg−1 day−1 ) averaged over the WAM domain (box in Figure 1(a)).
flow north of the precipitation region. This southerly flow force moisture transport into the continent. This transis roughly cancelled by the northerly flow of the non- port compensates the dry air advection by the latentlatent heating driven circulation.
heating driven circulation and thus enables the northward
advance of the total moisture convergence. Therefore,
non-latent diabatic heating helps the seasonal northward
3.2. Seasonal cycle
migration of monsoon precipitation, while latent heatSeasonal cycles of the vertically integrated moisture con- ing helps maintain the strength of monsoon precipitavergence and its latent and non-latent (sensible heat fluxes tion.
The relative contributions of latent and non-latent
plus radiative cooling) components are shown in Figure 7
heating
vary among the reanalyses. In ERA40 latentwith the contours of the annual cycles of precipitation for
heating
driven convergence dominates, and the total
each reanalysis. The maximum of total moisture converand
latent-heating
related moisture convergence resemgence is generally to the north of the precipitation in
ble
each
other.
But
in the other reanalyses, non-latent
spring in all the reanalyses except ERA40. During this
heating
driven
convergence
dominates as demonstrated
period much of the moisture convergence over land is
by
the
peak
in
the
total
moisture
convergence north of
due to non-latent diabatic heating which is particularly
the
precipitation
maxima
throughout
the year. In NCEP I
weak in ERA40. In summer, when the non-latent heatand
II,
the
spatial
separation
of
the
moisture
convergence
ing driven convergence is weak, the regions of maximum
due
to
latent
and
non-latent
heating
is
so
large
that they
precipitation and maximum moisture convergence tend to
together
result
in
double
peaks
in
total
moisture
converbe collocated. This is expected since latent heating within
gence
in
August
and
September,
one
in
the
precipitation
the region of precipitation becomes the primary forcing
for the moisture convergence. Furthermore, latent heat- region, one to the north. In general, if non-latent heating also induces dry air advection (moisture divergence) ing forced convergence is relatively strong, the monsoon
to the north of the precipitation region and limits the precipitation tends to migrate deeper into the continent
northward expansion of the total moisture convergence (NCEPII, NCEPI and JRA25), otherwise moisture conand precipitation. The moisture convergence associated vergence is primarily forced by the local latent heating
with non-latent heating is always north of the precipita- and the northward migration of the monsoon precipitation
tion maximum. Throughout the year, sensible heat fluxes is limited (ERA40).
c 2009 Royal Meteorological Society
Copyright !
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
421
Figure 9. Seasonal cycle of precipitation (mm/day) and moisture convergence (g kg−1 day−1 ) for 2005 (solid) and 2006 (dashed).
The different roles of latent and non-latent heating in
moisture transport into the continent can also be demonstrated by the seasonal cycles in the amount of total
moisture convergence and its components averaged over
the West Africa domain (the box in Figure 1(a)). The
distinct seasonal evolutions of the moisture convergence
associated with non-latent and latent heating are striking (Figure 8). The moisture convergence associated with
non-latent heating has a semi-annual cycle, with minima in January (because of the solar cycle) and August
(because of a reduction in surface sensible-heat fluxes
due to strong precipitation, heavy cloudiness, etc.). The
moisture convergence associated with latent heating on
the other hand has a strong seasonal cycle. Before the
rainy season, latent heating from precipitation to the south
drives dry air to West Africa from the north. After convergence forced by non-latent diabatic heating has reached
its maximum, the rainy season arrives and the role of
latent heating changes dramatically. The local convergence effect (first term on the r.h.s. of Eq. (10)) of latent
heating dominates the advective effect and thus positive
c 2009 Royal Meteorological Society
Copyright !
feedback is switched on. Total moisture convergence is
enhanced even though the increase in precipitation is
accompanied by reduced moisture convergence due to
non-latent diabatic heating. The moisture convergence by
non-divergent wind is relatively small and shows little
seasonal variability.
3.3.
Special cases in 2005 and 2006
Even though interannual variability of the African
monsoon is beyond the scope of this study, a brief comparison of the seasonal cycles of moisture convergence
of the two AMMA (African Monsoon Multidisciplinary
Analysis) years is given. Figure 9 shows the precipitation
from TRMM and NCEPII reanalysis as well as the total
and the two components of moisture convergence for
the years 2005 and 2006 over West African (box in
Figure 1(a)). In the TRMM data, the most prominent
difference between the precipitation cycles of the two
years is the month of peak precipitation. It was June
in 2005 and August in 2006. In the reanalysis, on the
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
422
S. HAGOS AND C. ZHANG
Figure 10. July–August–September convergence (CONVLH, left panels, shading) and advection (ADVLH, right panels, shading) components
of moisture convergence (g kg−1 day−1 ) by latent heating. The contours are vertically averaged mixing ratio (g kg−1 ).
c 2009 Royal Meteorological Society
Copyright !
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
423
Figure 11. Schematic diagrams summarizing the divergent circulations (arrows), their moisture transport (shading of the arrows) and convergence
(ovals) associated with non-latent and latent heating in the West African monsoon system. Amount of moisture transport increases from white
to dark shading.
other hand, the precipitation peak is in July in 2005
and September in 2006. Therefore once again, direct
interpretation of the reanalysis moisture convergence
to explain the observed seasonal cycle of precipitation
could lead to an erroneous conclusion. However, insight
into some of the mechanisms through which interannual
variability is introduced in the monsoon cycle can
be gained by considering the seasonal cycle of the
components of moisture convergence discussed above.
c 2009 Royal Meteorological Society
Copyright !
Consider the 2005 and 2006 total moisture convergence
from the reanalysis (Figure 9(c)). Of particular interest
is the difference between May, June and July months
of the two years. The reason for the particularly large
(about 25%) difference during this period transition is
related to the distinct roles of latent heating discussed in
the last subsection. During the dry period latent heating
elsewhere advects dry air and suppresses convergence,
while in the wet period local latent heating enhances
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
424
S. HAGOS AND C. ZHANG
convergence. Figure 9(d) shows the evolution of the
moisture convergence associated with latent heating. In
April 2005, the negative (remote diverging) effect of
latent heating prevails in both years. This remains the
case for 2006 until late June. But in 2005, the role of
latent heating has become positive (local converging) by
early June and hence the large difference in latent-heating
driven moisture convergence and the total moisture
convergence during that month between the two years.
One has to keep in mind that the changing role of
latent heating early in the rainy season only amplified
differences. This difference is ultimately caused by the
processes that led to the early arrival of precipitation
and the associated latent-heating driven convergence in
2005 in comparison to 2006. One plausible candidate for
this is the larger non-latent heating moisture convergence
in the spring of 2005 compared to that of 2006 (Figure 9(e)). Particularly in May, the difference is about
25%. Nevertheless, latent-heating feedback through moisture convergence appears to play an important role in the
year-to-year variations of the moisture convergence cycle,
especially during the early months of the rainy season.
3.4.
Influences of local and remote latent heating
As noted in section 3.2, latent heating plays two roles
in moisture transport. Locally it favours moisture convergence into the region of precipitation and meanwhile
drives the large-scale circulation that may advect remote
dry air towards the precipitation region from the north. In
this section, these two roles are compared. Moisture convergence associated with latent heating is decomposed
according to Eq. (10) into one term corresponding to
moisture convergence at a point driven by local latent
heating (CONVLH) and the second term (ADVLH) representing the total advection related to all latent heating
in the global Tropics.
Figure 10 shows the July–August–September climatological means of the two components of moisture convergence associated with latent heating. As expected,
local precipitation–convergence feedback is most effective over regions with high precipitation and high moisture. Those are the western coast of Africa and the
Cameroon highlands. More remarkable, however, is the
response of the Sahel region to remote latent heating.
Because it straddles the Sahara region, the Sahel is a
region of strong moisture gradients (Figure 10 contours).
A circulation pattern with a wind component against this
gradient advects dry air from the Sahara into the Sahel.
Those are the circulation patterns forced by remote latent
heating. In other words, the aggregate of circulations
forced by convective activities throughout the Tropics
mixes the dry Sahara air with the Sahel air to inhibit
moisture convergence.
4.
Discussion
Latent and non-latent (surface sensible heat flux and radiation) heating processes play vastly different roles in
c 2009 Royal Meteorological Society
Copyright !
the dynamics of the African monsoon system because
of their associated vertical velocity and horizontal divergence profiles. Their different roles are summarized in
Figure 11. Northward moisture transported from the Gulf
of Guinea by the latent-heating driven deep circulation
converges inside the monsoon rain band and thereby helps
maintain the strength of monsoon precipitation. Its dry
air advection into the monsoon rain band from the north,
on the other hand, does not help the seasonal northward
migration of monsoon precipitation. This negative feedback is largely compensated by the moisture transport
of the shallow circulation driven by non-latent diabatic
heating, which penetrates through the monsoon rain band
and converges north of it. In this sense, the non-latent
heating driven shallow circulation plays an indispensable
role in the seasonal advance of monsoon precipitation in
boreal spring. This result suggests that if a GCM has difficulties in reproducing the WAM seasonal cycle with its
rain band reluctant to migrate deep inland, its simulated
shallow meridional circulation might be too weak.
Throughout the year, the near-surface circulation
driven by non-latent (sensible) heating gradient between
the Saharan heat-low and the cold tongue of the Gulf of
Guinea supplies moisture to West Africa. This moisture
supply has a semi-annual cycle with minima in January
and August and maxima in May and October. The moisture transport by the latent-heating driven circulation, in
contrast, has a clear seasonal cycle. In the dry season, the
circulation driven by regional latent heating advects dry
air from the north and suppresses local moisture convergence. This distinction of the roles of various components
of heating has a potential to improve our understanding of
the nature of interannual variability, as demonstrated by
comparison of moisture convergence in 2005 and 2006.
In those cases, stronger moisture convergence by sensible heating fluxes in spring of 2005 appear to have led to
earlier arrival of precipitation and greater latent-heating
feedback, in comparison to that of 2006.
An analysis of the regional dependence of the local
and remote influences of latent heating shows that effects
of local moisture convergence are the strongest over
regions of high precipitation, such as the western coast
of Africa and the Cameroon highlands. The remote
advective effects dominate over the region of strong
moisture gradient, e.g. the Sahel. The results of this study
provide further thermodynamic and dynamical dimension
to the emerging consensus that the Sahel droughts of the
1980s are indeed related to warm SSTs and the enhanced
convective activities over the tropical southern oceans
(Giannini et al., 2003; Lu and Delworth, 2005; Hagos
and Cook, 2008 and references therein). This mechanism
might have played a role in the observed decline in the
Sahel precipitation in response to warming and enhanced
convection over the adjacent oceans.
While the processes discussed in this study are fairly
consistent in all reanalyses used, the magnitude and contribution of each process in the monsoon dynamics vary
among them. A case in point is the relatively weak nonlatent heating moisture convergence in ERA40 and a lack
of inland migration of the associated precipitation when
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)
DIABATIC HEATING, DIVERGENCE AND MOISTURE
425
Precipitation Analysis (TMPA): Quasi-global, multiyear, combinedcompared to those of the other reanalyses. In light of the
sensor precipitation estimates at fine scales. J. Hydrometeorol. 8:
uncertainties in the future of the monsoon in the twenty38–55.
first century, the tools presented in this study can be used Jackson JD. 1999. Classical electrodynamics, 3rd edition. John Wiley
to evaluate climate models on their faithful representation
& Sons: New York.
nd
of diabatic processes and their roles in moisture trans- Jaluria Y, Torrance KE. 2003. Computational heat transfer, 2 edition.
Taylor & Francis.
port. Specifically, vertical profiles of diabatic heating are Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L,
closely related to the circulation patterns. Therefore, evalIredell M, Saha S, White G, Woollen J, Zhu Y, Leetmaa A,
Reynolds R, Chelliah M, Ebisuzaki W, Higgins W, Janowiak J,
uation of modelled and observed diabatic heating profiles
Mo KC, Ropelewski C, Wang J, Jenne R, Joseph D. 1996. The
and their implications for moisture transport and surNCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc.
face–air interaction may be a good starting point.
77: 437–471.
Acknowledgements
This study was supported by NASA TRMM/GPM
Project Award Number, NNX07AD41G. NCEP Reanalyses Derived data are provided by the NOAA/OAR/ESRL
PSD, Boulder, Colorado, USA, from their Web
site at http://www.cdc.noaa.gov/. ERA-40 data are
provided by the European Centre for MediumRange Weather Forecasts, from their web site at
http://www.ecmwf.int/. JRA-25 data are provided by the
Japan Meteorological Agency (JMA) at their web site at
http://jra.kishou.go.jp/index en.html and the TRMM data
provided by TRMM Science Data Information System
(TSDIS).
References
Arfken GB, Weber HJ. 2005. Mathematical methods for physicists, 5th
edition. Elsevier Academic Press.
Batchelor GK. 2000. An introduction to fluid dynamics. Cambridge
University Press.
Cook KH, Vizy EK. 2006. Coupled model simulations of the
West African monsoon system: Twentieth- and twenty-first-century
simulations. J. Climate 19: 3681–3703.
Giannini A, Saravanan R, Chang P. 2003. Oceanic forcing of Sahel
rainfall on interannual to interdecadal time scales. Science 302:
1027–1030.
Gu G, Adler RF. 2004. Seasonal evolution and variability associated
with the West African monsoon system. J. Climate 17: 3364–3377.
Guichard F, Kergoat L, Mougin E, Timouk F, Baup F, Hiernaux P,
Lavenu F. 2009. Surface thermodynamics and radiative budget in
the Sahelian Gourma: Seasonal and diurnal cycles. J. Hydrol. 375:
161–177.
Hagos SM, Cook KH. 2007. Dynamics of the West African monsoon
jump. J. Climate 20: 5264–5284.
Hagos SM, Cook KH. 2008. Ocean warming and late-twentieth-century
Sahel drought and recovery. J. Climate 21: 3797–3814.
Hagos SM, Zhang C, Tao W-K, Lang S, Takayabu YN, Shige S,
Katsumata M, Olson B, L’Ecuyer T. 2009. Estimates of tropical
diabatic heating profiles: Commonalities and uncertainties. J.
Climate, submitted.
Hsieh J-S, Cook KH. 2007. A study of the energetics of African easterly
waves using a regional climate model. J. Atmos. Sci. 64: 421–440.
Huffman GJ, Adler RF, Bolvin DT, Gu G, Nelkin EJ, Bowman KP,
Hong Y, Stocker EF, Wolff DB. 2007. The TRMM Multisatellite
c 2009 Royal Meteorological Society
Copyright !
Kanamitsu M, Ebisuzaki W, Woollen J, Yang S-K, Hnilo JJ, Fiorino M,
Potter GL. 2002. NCEP–DOE AMIP-II reanalysis (R-2). Bull. Am.
Meteorol. Soc. 83: 1631–1643.
Lu J, Delworth TL. 2005. Oceanic forcing of the late 20th
century Sahel drought. Geophys. Res. Lett. 32: L22706,
DOI:10.1029/2005GL023316.
Miller RL, Slingo A, Barnard JC, Kassianov E. 2009. Seasonal contrast
in the surface energy balance of the Sahel. J. Geophys. Res. 114:
D00E05, DOI:10.1029/2008JD010521.
Nigam S, Chung C, DeWeaver E. 2000. ENSO diabatic heating
in ECMWF and NCEP–NCAR reanalyses, and NCAR CCM3
simulation. J. Climate 13: 3152–3171.
Okumura Y, Xie S-P. 2004. Interaction of the Atlantic equatorial cold
tongue and the African monsoon. J. Climate 17: 3589–3602.
Onogi K, Tsutsui J, Koide H, Sakamoto M, Kobayashi S, Hatsushika H,
Matsumoto T, Yamazaki N, Kamahori H, Takahashi K, Kadokura S,
Wada K, Kato K, Oyama R, Ose T, Mannoji N, Taira R. 2007. The
JRA-25 reanalysis. J. Meteorol. Soc. Jpn 85: 369–432.
Ramel R, Gallée H, Messager C. 2006. On the northward shift of the
West African monsoon. Clim. Dyn. 26: 429–440.
Selvadurai APS. 2000. Partial differential equations in mechanics. Vol
2: The biharmonic equation and Poisson’s equation. Springer Verlag.
Sultan B, Janicot S. 2003. The West African monsoon dynamics. Part
II: The ‘preonset’ and ‘onset’ of the summer monsoon. J. Climate
16: 3407–3427.
Taylor CM. 2008. Intraseasonal land–atmosphere coupling in the West
African monsoon. J. Climate 21: 6636–6648.
Timouk F, Kergoat L, Mougin E, Lloyd CR, Ceschia E, Cohard J-M, de
Rosnay P, Hiernaux P, Demarez V, Taylor CM. 2009. Response of
surface energy balance to water regime and vegetation development
in a Sahelian landscape. J. Hydrol. 375: 178–189.
Uppala SM, Kållberg PW, Simmons AJ, Andrae U, Da Costa
Bechtold V, Fiorino M, Gibson JK, Haseler J, Hernandez A,
Kelly GA, Li X, Onogi K, Saarinen S, Sokka N, Allan RP,
Andersson E, Arpe K, Balmaseda MA, Beljaars ACM, Van De
Berg L, Bidlot J, Bormann N, Caires S, Chevallier F, Dethof A,
Dragosavac M, Fisher M, Fuentes M, Hagemann S, Hólm E,
Hoskins BJ, Isaksen L, Janssen PAEM, Jenne R, Mcnally AP,
Mahfouf J-F, Morcrette J-J, Rayner NA, Saunders RW, Simon P,
Sterl A, Trenberth KE, Untch A, Vasiljevic D, Viterbo P, Woollen J.
2005. The ERA-40 re-analysis. Q. J. R. Meteorol. Soc. 131:
2961–3012.
Yanai M, Esbensen S, Chu J-H. 1973. Determination of bulk properties
of tropical cloud clusters from large-scale heat and moisture budgets.
J. Atmos. Sci. 30: 611–627.
Zhang C, Hagos SM. 2009. Bi-modal structure and variability of largescale diabatic heating in the Tropics. J. Atmos. Sci., in press.
Zhang C, Woodworth P, Gu G. 2006. The seasonal cycle in the lower
troposphere over West Africa from sounding observations. Q. J. R.
Meteorol. Soc. 132: 2559–2582.
Q. J. R. Meteorol. Soc. 136(s1): 411–425 (2010)