P3.9
LOW-FREQUENCY RAINFALL VARIABILITY IN NORTHWEST AFRICA
AND THE SAHELIAN AND SUDANIAN ZONES
1
2
2
2
3
2
Jon M. Schrage , A.H. Fink , T. Bruecher , M. Christoph , P. Knippertz , and P. Speth
1
2
Dept. of Environmental
and Atmospheric Sciences
Creighton University
2500 California Plaza
Omaha, NE 68178, USA
As seen in Fig. 1, broad regions of the
Sahel had above normal precipitation from 19501970, but have seen unrelenting dry conditions
since 1970. The longer climate records in Benin
(Fig. 2) also exhibit this trend, with another period
of wet (dry) conditions from 1920-1935 (19351950), although the time series for the entire Sahel
(Fig.1) show considerably more year-to-year
persistence.
Much of West Africa experiences
significant interannual and interdecadal variability
of rainfall. The German IMPETUS (“An Integrated
Approach to the Efficient Management of Scarce
Water Resources in West Africa”) project has
compiled and quality-checked precipitation data
from networks of rain gauges in various climate
regions of West Africa from a variety of sources.
Particular attention is given to the climates of
Benin and Morocco, as these are the locations of
the field campaigns for the IMPETUS project.
Annual mean,
1921-2002
Standardised
precipitation
anomaly, 1922 - 2001
South-Benin/Guinea Coast (6°N - 9°N) and Sudan region (9°N - 12,5°N)
3
2000
1995
1990
0
1985
1980
1975
1970
1965
1960
1955
Base period: 1950-1990
Figure 1. Annual precipitation index anomalies for
the Central and Western Sahel (top). Data
processing: 1950-1998, Chris Landsea (NOAA
AOML), 1999- present, IGM Cologne.
0
15
-1
10
-2
-3
5
Base period: 1922-2001
0
2000
10
1950
-3
15
20
1990
0
1
1980
20
25
1970
5
1
2
1960
-2
25
Sudan
1950
West Sahel
Guinea Coast
1940
Central Sahel
Precipitation index
a)
-1
2
30
b)
1930
30
number of available stations
3
number of available stations
Jun.-Sept., 1955-2002
June-September, 1950-2002
Precipitation index
Dept. of Atmospheric
and Oceanic Sciences
University of Wisconsin
1225 W. Dayton Street
Madison, WI 53706 USA
INTRODUCTION
1920
1.
3
Institute of Geophysics
and Meteorology
University of Cologne
Kerpener Strasse 13
50923 Cologne, Germany
Figure 2. Annual precipitation index anomalies for
the Guinea Coast and Sudanian climate regions of
Benin. Data source: National Meteorological
Service, Benin. Data processing: IGM Cologne
a.)
25
1
20
0
15
-1
10
-2
5
2000
1990
1980
1970
1960
1950
1940
1930
1910
1920
0
1900
-3
year
Year (September to August)
Precipitation index
3
2
30
b) Mediterranean region
25
1
20
0
15
-1
10
-2
5
2000
1990
1980
1970
1960
1950
1940
1930
1910
1920
0
1900
-3
number of available stations
(max. 4)
Precipitation index of the Mediterranean region
(calculated by the climate normal period from 1961 to 1990)
b.)
year
Year (September to August)
c.)
Precipitation index
3
2
30
c) Region south of the Atlas
25
1
20
0
15
-1
10
-2
5
2000
1990
1980
1970
1960
1950
1940
1930
1910
1920
0
1900
-3
number of available stations
(max. 5)
Precipitation index of the region south of the Atlas
(calculated by the climate normal period from 1961 to 1990)
year
Year (September to August)
d.)
35°N
30°N
10°W
5°W
Figure 3. Time series of annual (SeptemberOctober, hydrological year) precipitation index
anomalies for three climate zones of northwest
Africa: (a) the Atlantic—region I (ATL), (b) the
Mediterranean Coast—region II (MED), and (c)
South of the Atlas Mountains—region III (SOA).
For more details, see Knippertz et al. (2003a)
Climate regions are indicated in panel (d).
conditions from the mid 1980s through the mid
1990s.
2.
Relation to NAO and ENSO indices
Of primary interest to this study is the
nature of the association between the precipitation
in each climate region and two year indices of
global climate variability: NAO-G and the Niño3
sea surface temperatures (SSTs). Knippertz et al.
(2003a) demonstrated that rainfall anomalies in
the Atlantic region can be understood from
meridinal shifts in the Atlantic storm track in
December, January, and February (DJF).
Therefore, they are strongly correlated to indices
of the North Atlantic Oscillation (NAO). It will be
shown that about 50% of the rainfall variance in
the Moroccan Atlantic region during DJF can be
explained by the NAO-G index calculated from the
standardized mean sea-level pressure differences
between Iceland and Gibraltar (NAO-G). The
same index explains less than 9% of the rainfall
variability in the other two regions. The lacking
NAO influence can be understood from the major
rainfall producing circulation patterns; Knippertz et
al. (2003a) showed that rainfall along the
Mediterranean Coast is largely related to weak
lows in the western Mediterranean Sea, whereas
precipitation events occurring south of the Atlas
Mountains in the transition seasons are often
related to tropical-extratropical interactions
(Knippertz et al., 2003b). Figure 4 illustrates the
25-year running correlations between NAO-G and
the precipitation indices for the three climate
regions of Morocco. The strong negative
correlation between the NAO-G and Moroccan
rainfall in the Atlantic climate zone persisted
th
throughout the 20 century.
0,4
25-year running mean correlation coefficient of the precipitation index of the
ATL, MED and SOA region with the NAO-G-index (DJF) for 20. century
0,4
0,2
0,2
0
-0,2
0
-0,2
95%
-0,4
-0,6
-0,4
99%
-0,6
MED
ATL
1920
1930
-0,8
SOA
1990
1980
1970
1960
1950
-1
1940
-1
1900
For northwestern Africa (Fig. 3), the
variability is highly interannual along the Atlantic
coast, while highly interdecadal in the
Mediterranean region. Despite this difference in
character, however, both regions exhibit
considerably reduced rainfall totals since the late
1970s, with the exception of a few wet years in the
mid-1990s along the Atlantic coast. The region
south of the Atlas mountains experienced wet
1910
-0,8
correlation coefficient
30
a) Atlantic
correlation coefficient
Precipitation index
2
number of available stations
(max. 28)
Precipitation index of the Atlantic region
(calculated by the climate normal period from 1961 to 1990)
3
year
Figure 4. Twenty-five year running correlations
between NAO-G index and December-JanuaryFebruary (DJF) precipitation in the Atlantic
(green), Mediterranean (red), and South of Atlas
Mountains (blue) Moroccan climate zones.
Knippertz (2003c) also investigated the
influence of ENSO on rainfall in northwest Africa.
A correlation analysis between the seasonal
precipitation indices and the DJF Niño3-index
revealed a small negative correlation for spring
(March–May, MAM) and a small positive
correlation for autumn (September–October,
SON). However, as evident in Fig. 3 (a slightly
modified version of Fig. 6.2 in Knippertz, 2003c)
the MAM correlations were not stable in the
course of the 20th century. Especially between the
mid-1930s and 1960 the influence of ENSO on
spring rainfall north of the Atlas Mountains was
insignificant.
25-year running mean correlation coefficient of the precipitation index of the
ATL, MED and SOA region with the winter Nino3 - index for 20. century
-0.2
0
-0.2
95%
-0.4
-0.4
99%
-0.6
ATL
1990
1980
1970
1960
-0.8
1950
1910
1900
MED
1940
SOA
-0.8
1930
-0.6
correlation coefficient
0.2
0
1920
correlation coefficient
0.2
year
Figure 5. As in Fig. 4, except for one-season
lagged correlation between Niño-3 SSTs and
Moroccan climate zones.
As can be seen from Fig 6 rainfall in the
Sahelian regions were significantly (95% level)
correlated with the DJF Niño3-index after the
drought commenced about 1970. In contrast,
ENSO seems not to influence the rainfall along the
Guinea Coast (Fig. 6). In the latter region lowfrequency variations of rainfall are largely
determined by anomalies of sea-surface
temperatures in the Gulf of Guinea and the tropical
Atlantic.
25-year running mean correlation coefficient of the precipitation index of the
Central Sahel and West Sahel with the Nino3 - index for the 20. century
0.2
0
-0.2
0
-0.2
95%
-0.4
-0.4
99%
1995
1990
-0.8
1985
1980
Guinea Coast
1970
1960
1950
West Sahel
1975
-0.6
Central Sahel
-0.8
1965
-0.6
1955
correlation coefficient
0.2
correlation coefficient
a.)
year
Figure 6. Twenty-five year running lagged
correlation between Niño-3 SSTs and precipitation
totals for three West African climate zones.
3.
Concluding remarks
Our preliminary analysis of the role of
teleconnections for rainfall north and south of the
Sahara yielded two noteworthy results. Firstly, a
discernible influence of ENSO on springtime
rainfall in subtropical northwest Africa and on
monsoon rainfall in the Sahel was restricted to the
periods after the 1960-1970s and, at least for
Northwest Africa, to before 1930. What may have
caused these changes over time in the
teleconnection patterns? In a recent paper,
Knippertz et al. (2003d) show different correlation
patterns between the Niño3-index and the mean
sea-level pressure (MSLP) over the North Atlantic
Ocean for the periods 1900-1925, 1931-1956 and
1962-1987 that may help to explain the timevarying influence of ENSO on European and
Northwest African springtime precipitation. In this
context, it is interesting to note the results of
Goldenberg et al. (2001) with respect to the
decadal fluctuations of the interhemispheric SST
dipole in the Atlantic Ocean. From 1930–1970 the
North Atlantic Ocean was anomalously warm,
while the South Atlantic Ocean was colder than
normal. Our preliminary analyses suggests that
during this period, an ENSO influence on rainfall is
not evident from the precipitation time series at
neither margin of the Saharan desert. It is
speculated that during these decades the forcing
from the Atlantic region was dominant and/or the
strength of the ENSO events was generally
weaker.
References:
Goldenberg, B. S., C.W. Landsea, A.M. Mestas-Nuñez and W.
M. Gray, 2001: The recent increase in Atlantic hurricane
activity: causes and implications. Science, 293, 474-479.
Knippertz, P., M. Christoph and P. Speth, 2003a: Long-term
precipitation variability in Morocco and the link to the largescale circulation in recent and future climates. Meterol. Atmos.
Phys., 83, 67-88.
Knippertz, P., A. H. Fink, A. Reiner and P. Speth, 2003b: Three
late summer/early autumn cases of tropical-extratropical
interactions causing precipitation in northwestern Africa. Mon.
Wea. Rev., 131 (1), 116-135.
Knippertz, P., 2003c: Niederschlagsvariabilität in
Nordwestafrika und der Zusammenhang mit der großskaligen
atmosphärischen Zirkulation und der synoptischen Aktivität.
Mitteilungen aus dem Institut für Geophysik und Meteorologie
der Universität zu Köln (Eds. M. Kerschgens, F. M. Neubauer,
M. Pätzold, P. Speth, B. Tezkan), 152, 136 S (In German).
Knippertz, P., A. H. Fink, A. Reiner and P. Speth, 2003d:
Decadal changes in the link between El Nino and springtime
North Atlantic Oscillation and European-North African Rainfall.
Int. J. of Climatol., 23, 1293-1311.