Content uploaded by Nussaȉbah B. Raja
Author content
All content in this area was uploaded by Nussaȉbah B. Raja on Apr 24, 2018
Content may be subject to copyright.
Vol.:(0123456789)
1 3
Meteorology and Atmospheric Physics
https://doi.org/10.1007/s00703-018-0604-7
ORIGINAL PAPER
Trend analysis ofannual precipitation ofMauritius fortheperiod
1981–2010
NussaïbahB.Raja1· OlguAydin2
Received: 22 May 2017 / Accepted: 15 April 2018
© Springer-Verlag GmbH Austria, part of Springer Nature 2018
Abstract
This study researched the precipitation variability across 53 meteorological stations in Mauritius and different subregions
of the island, over a 30-year study period (1981–2010). Time series was investigated for each 5-year interval and also for
the whole study period. Non-parametric Mann–Kendall and Spearman’s rho statistical tests were used to detect trends in
annual precipitation. A mix of positive (increasing) and negative (decreasing) trends was highlighted for the 5-year interval
analysis. The statistical tests nevertheless agreed on the overall trend for Mauritius and the subregions. Most regions showed
a decrease in precipitation during the period 1996–2000. This is attributed to the 1998–2000 drought period which was
brought about by a moderate La Niña event. In general, an increase in precipitation levels was observed across the country
during the study period. This increase is the result of an increase in extreme precipitation events in the region. On the other
hand, two subregions, both located in the highlands, experienced a decline in precipitation levels. Since most of the reservoirs
in Mauritius are located in these two subregions, this implies serious consequences for water availability in the country if
existing storage capacities are kept.
1 Introduction
Precipitation in the tropics is projected to undergo signifi-
cant changes in terms of pattern, magnitude, and intensity
as a result of global climate change (Collins etal. 2013).
Small island developing states (SIDS) such as Mauritius,
generally more sensitive and vulnerable to climate change,
have a lower adaptive capacity as compared to mainland
countries (Kelman 2014; Mavrogenis etal. 2014). In the
recent years, precipitation-related natural hazards such as
floods, extreme precipitation, landslides, and drought have
become a recurrent phenomenon in Mauritius. In addition,
Mauritius depends primarily on precipitation and its distri-
bution for both agricultural and individual use. It is, there-
fore, imperative to track the changing precipitation trends
in the region, so the effects of the changing climate can be
monitored and understood.
Dore and Singh (2013) computed precipitation projec-
tions in Mauritius and showed that the annual mean total
precipitation is due to decline in the future. However, this
study was based on one single region of Mauritius and,
therefore, did not consider the island of Mauritius as a
whole. Despite being a small island, Mauritius experiences
some significant localised and microclimatic variations
(Padya 1989; Staub etal. 2014). The precipitation distribu-
tion in Mauritius is influenced by the varying topographic
features, coastal proximity, prevailing winds, synoptic scale
processes, and ocean–atmosphere interactions (Dhurmea
etal. 2009; Fowdur etal. 2014). Thus, changing precipita-
tion trends in different microclimatic regions in Mauritius
should be analysed individually for a clearer picture.
Several statistical methods, both parametric and non-
parametric, are used to detect trends in hydrometeorologi-
cal time series (Chen etal. 2007; Dahmen and Hall 1990;
Zhang etal. 2006). While parametric statistical tests are usu-
ally more powerful, non-parametric statistical tests are more
appropriate for non-normally distributed data that are often
found in hydrometeorological time series. The Mann–Ken-
dall test is the most common non-parametric test used by
researchers in studying hydrometeorological time series
Responsible Editor: J.-T. Fasullo.
* Olgu Aydin
oaydin@ankara.edu.tr; drolguaydin@gmail.com
1 GeoZentrum Nordbayern, University Erlangen-Nürnberg,
91054Erlangen, Germany
2 Faculty ofHumanities, Department ofGeography, Ankara
University, 414 Nolu Oda, Sıhhıye, 06100Ankara, Turkey
N.B.Raja, O.Aydin
1 3
(Longobardi and Villani 2009; Wang etal. 2012; Yang etal.
2012a, b; Yanming etal. 2012). Less common is Spearman’s
rho which is used to detect monotonic trends in hydrome-
teorological data (Yue etal. 2002). In many studies, Spear-
man’s rho is used in combination with the Mann–Kendall
test for comparison purposes (Ahmad etal. 2015; Kahya and
Kalaycı 2004; Li etal. 2008; Shadmani etal. 2012; Yaning
etal. 2009; Yue etal. 2002).
The purpose of this study is to evaluate the changing
temporal trends of precipitation on the island of Mauritius,
located east of Africa. This study proposes to investigate the
long-term precipitation variability across 53 stations in Mau-
ritius, over a 30-year study (1981–2010). Two non-paramet-
ric rank-based statistical tests, namely, the Mann–Kendall
test and Spearman’s rho, will be used to detect the mono-
tonic trends in the time series data of the annual mean total
precipitation of Mauritius. The different modes of oscilla-
tions in the western Indian Ocean will also be reviewed and
employed to explain the changing precipitation trends in
Mauritius.
2 Modes ofvariability inthewestern
tropical Indian Ocean
The ocean–atmosphere interactions in the tropical Indian
ocean play an important role in shaping regional and local
climates in the south-western tropical Indian Ocean (Schott
etal. 2009). Although rainfall patterns in Mauritius have not
been intensively analysed, there have been several studies
investigating changing rainfall patterns in the neighbour-
ing East African countries. It is believed that the climate in
Mauritius may be affected by these same phenomena, despite
having a distinct climate as compared to the countries located
on the African mainland (Senapathi etal. 2010). These pro-
cesses exhibit differing characteristics of climate variability
with time scales from semi-annual to several years and dec-
ades. A summary of current knowledge of the significant
processes in the south-western tropical Indian Ocean, affect-
ing the local climate of Mauritius, is provided in this section.
2.1 Annual andsemi‑annual variability
The annual cycle, arising from complex ocean–atmos-
phere dynamics, is the most dominant component of
low-frequency variability over the Indian Ocean (White
2001; Brandt etal. 2002). Mauritius, located at the edge of
the tropics, is located within the belt of trade winds. The
island is dominated by two wind regimes: the South-East
Trade Winds (westerlies) and the North West Monsoon
Winds (easterlies) (Padya 1989; Fowdur etal. 2014). The
westerlies prevail almost all year round, but are strongest
during the austral summer months when the Inter-tropical
Convergence Zone (ITCZ) migrates south of Mauritius.
The annual cycle of trade winds in the southern tropical
Indian Ocean generates Rossby waves propagating west-
ward in the open ocean (Masumoto and Meyers 1998;
Birol and Morrow 2001). Instead, Quartly etal. (2006)
and Chelton etal. (2007) considered these anomalies to
be a series of westward eddies moving at the phase speed
of baroclinic Rossby waves. Furthermore, ocean processes
in the Indian Ocean are forced not only by local winds
but also remotely from the Pacific into the tropical Indian
Ocean through the Indonesian Throughflow, confirming
the complexity of ocean dynamics in this region.
The easterlies are usually observed during the austral
winter, mostly from April until June. These develop dur-
ing transition periods between the two monsoons: from the
Northeast to the Southwest Monsoon in April–June and vice
versa in November–December, characterised by an equato-
rial jet known as the Wyrtki Jet and upwelling Rossby waves
(Schott etal. 2009). The formation of the jet is accompanied
by thermocline uplifting in the west and sinking in the east
of the Indian Ocean Basin, resulting in easterly propagat-
ing waves. This leads to warmer sea surface temperatures
(SSTs) in the region during the austral summer, leading to
increased evaporation and moisture availability, and hence
enhanced rainfall.
2.2 Inter‑annual variability
There has been growing interest with regard to the inter-
annual variability in the tropical ocean in the past decade
since the occurrence of a pronounced basin-wide pattern of
SSTs in the equatorial Indian Ocean—a phenomenon named
the Indian Ocean Dipole (IOD) (Saji etal. 1999). The recog-
nised modes of inter-annual variability of the Indian Ocean
are: the El Nino Southern Oscillation (ENSO), the bien-
nial signal or oscillation as a result of the IOD (Meehl etal.
2003; Allan etal. 2003; Feng and Meyers 2003) and dec-
adal variability (Ashok etal. 2004; Annamalai etal. 2005).
The latter two signals are also referred to in the literature as
quasi-biennial signal and quasi-decadal oscillation (Allan
etal. 2003; White and Tourre 2007). A multidecadal vari-
ability in the Indian Ocean has also been observed, although
the mechanics behind this process have not been much stud-
ied (Tierney etal. 2013).
2.2.1 Indian ocean dipole mode
The coupled ocean–atmosphere phenomenon, IOD, is char-
acterised by an anomalous cold SST in the southeastern
equatorial Indian Ocean and anomalous warming of the
western equatorial Indian Ocean (Saji etal. 1999). It is
strongly connected to the annual cycle and reaches a peak
during the austral spring in September–October. A positive
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
IOD is characterised by anomalously cold SSTs in the east
and warm SST in the west. The zonal winds are anomalous
easterlies, blowing from east to west towards warmer waters,
which in turn causes excessive rain along the eastern Afri-
can coast, including Mauritius, and drought in Australia.
A negative IOD has the opposite effect, with anomalously
warm SSTs in the western Indian Ocean, resulting in more
westerly winds and enhanced rainfall in Australia. The zonal
wind anomaly created by this phenomenon along the equator
is responsible for the upper-ocean evolution that is unique
to the IOD and has a time scale of approximately 2years,
hence the term quasi-biennial signal which is used to refer to
the anomaly triggered by the IOD (Feng and Meyers 2003).
From Fig.1a, it is observed that during negative phases, a
decrease in precipitation is more likely observed in Mauri-
tius. However, during positive phases, no trend in precipita-
tion is seen.
2.2.2 El Niño Southern oscillation
El Niño influences the Indian Ocean through the atmos-
pheric connection, namely, the Southern Oscillation (Yu and
Rienecker 2000). This leads to large-scale sea-level pressure
fluctuations and affects the atmospheric zonal circulations,
namely, the Walker Circulation in the Indian Ocean. Warm
ENSO extremes are generally characterised by warmer than
usual SSTs off the East African coast and are associated with
enhanced rainfall during October–November (Webster etal.
1999). During the extreme phases of La Niña, the opposite
is observed, i.e., below-average precipitation throughout the
East African region, including in Mauritius (Fig.1b). ENSO
cycles between its extreme phases with a period of between
approximately 3 and 7years. The ENSO accounts up to 30%
of the variability in SST and regional climate in the Indian
Ocean (Saji etal. 1999).
2.2.3 Quasi‑decadal variability
Some distinct ENSO episodes of longer duration than the
typical inter-annual events have also been observed in the
Indian Ocean (Allan etal. 2003). These episodes have been
referred to as “protracted” El Niño and La Niña episodes and
cannot be defined as longer lived counterparts of the typical
inter-annual ENSO events. These “protracted” events exhibit
the presence of a quasi-decadal signal, similar to ENSO
events. It is important to note that these protracted events
are a combination of quasi-biennial, low-frequency modes
and other quasi-decadal components. Both “protracted”
El Niño and La Niña episodes are characterised by times
during their existence when they appear to be terminating,
only to recover again and often then be followed by sev-
eral other periods of waxing and waning before they finally
end. These intra-episode fluctuations result from periods of
strong biennial and low-frequency signal intensities when
more inter-annual ENSO characteristics dominate. During a
protracted La Niña episode during 1998–2000, some parts of
Eastern Africa received well above and below average areas
during short period of times. There was also a substantial
anticyclonic anomaly situated in the central south-western
Indian Ocean, although cyclonic anomalies usually observed
during the austral summer were not observed. As such, pro-
tracted episodes are defined by periods with mixed climatic
patterns and signatures, and at times, even opposite to the
usual influences of typical ENSO events.
2.3 Multidecadal variability
The Atlantic Multidecadal Oscillation (AMO) is influen-
tial on the decadal timescale for the long rains, particularly
for the precipitation cessation in the month of May in the
Eastern African Region (Argent etal. 2016). The influence
on the region is propagated through AMO-induced station-
ary Rossby waves, which in turn causes the placement of
the high and low pressures influencing the flow over the
region and the corresponding rising and sinking motion
Fig. 1 Scatterplot of annual precipitation anomalies in Mauritius
(1981–2010) vs a SST anomalies in the Equatorial Indian Ocean and
b SST anomalies in the North Atlantic
N.B.Raja, O.Aydin
1 3
influences the locations that receive the most precipitation.
It is also believed that the AMO may impact the strength
of the Indian monsoons (Wang etal. 2009). During a posi-
tive phase, this teleconnection consists of an increase in
the temperature gradient of the troposphere in the northern
hemisphere, which, in turn, may induce a late withdrawal
of the Indian summer monsoon, thus causing an increase of
the seasonal mean Indian monsoon rainfall. A strong mon-
soon is associated with the cooling of the whole western
Indian Ocean starting in the Arabian Sea, influencing the
climate through reduced precipitation (Krishnamurthy and
Kirtman 2003). In addition, Argent etal. (2016) suggest that
AMO-induced teleconnections also influence the climate in
Eastern Africa on the inter-annual scale by reinforcing or
countering other modes of variability in the region. Tierney
etal. (2013) also showed that Indian Ocean SSTs are also a
dominant influence on East African precipitation on multi-
decadal and longer timescales. The changes in SSTs weaken
or strengthen the Walker Circulation, causing multidecadal
wet or dry conditions, respectively.
3 Materials andmethods
All statistical computations were carried out using the R
3.1.3 (R Core Development Team 2016) using the following
packages: gstat (Pebesma 2004), Kendall (Mcleod 2011),
and trend (Pohlert 2016). The ArcGIS 10.2 software (ESRI
2013) was used to prepare the data and represent results on
map figures.
3.1 Study area
The island of Mauritius is located in the south-western
Indian Ocean, east of Madagascar (Fig.2). The island, of
volcanic structure, is 61km long and 46km wide, with
a surface area of 1844km2 (Fowdur etal. 2014). It has a
complex topography, consisting of mountains, river val-
leys, and plains. A distinctive feature of the island is its
central plateau area, which lies at an altitude of 300–600m
but with peaks of up to 800m, representing a former cal-
dera. This elevated area is closer to the west coast as com-
pared to the east coast and gently slopes towards coastal
lowlands, which are protected by coral reefs, except for
the southern coast. The climate is tropical maritime with
two seasons: summer, also the rainy season, which is
dominated by the occurrence of tropical cyclones between
November and April, and a drier winter, which is domi-
nated by the south-east trade wind and frontal systems
for the remaining part of the year (Nigel and Rughooputh
2010). The inter-annual variation in precipitation totals
depends on the passage of cyclones, occurring mostly
during January and March. This period accounts for 40%
of the total annual precipitation, as a result of the south-
ward migration of the inter-tropical convergence zones
towards subtropical latitudes and the occurrence of tropi-
cal cyclones (Staub etal. 2014). During the drier season,
anticyclonic conditions prevail, interrupted by the occa-
sional cold front.
Fig. 2 Location of Mauritius and distribution network of meteorological stations across the island
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
3.2 The data
3.2.1 Climate
According to the World Meteorological Organisation
(WMO), a long-term period of at least 30years is considered
appropriate for estimating climate factors such as precipi-
tation. This study spans 30years, from 1981 to 2010, and
includes total annual mean precipitation data from 53 mete-
orological stations located throughout Mauritius (Fig.2),
obtained from the Mauritius Meteorological Services
(2016). Based on topographical features, coastal proximity,
and mean precipitation values, the island is further divided
into six subregions: south-western coast (Region 1), eastern
highlands (Region 2), Black River National Park (Region
3), Eastern coast (Region 4), Northern coast (Region 5), and
north-western highlands (Region 6), as shown in Fig.3. A
clear spatial variability of precipitation in Mauritius can be
observed from Fig.3. The central part of the island, cor-
responding to the location of the Central Plateau, receives
the most annual precipitation. On the other hand, the north
and western part of the island receives less precipitation than
anywhere else in Mauritius. Total mean annual precipitation
is approximately 1400mm on the eastern coast (Region 4),
4000mm on the central plateau (Regions 2, 3, and 6), and
600mm on the drier south-western coast (Region 1). This
large spatial contrast in precipitation is caused by the rain
shadow effect on the interior and orographic forcing of the
South-East Trade Winds.
3.2.2 Historical review oftheIndian Ocean 1981–2010:
ENSO, IOD, andAMO
The list of years for ENSO, IOD, and AMO events is avail-
able both in the literature and in the online. However, due
to the classification of the different modes of oscillations,
the list of years normally does not agree in all details (Rao
etal. 2002; Meyers etal. 2007; Ummenhofer etal. 2009;
Yamagata etal. 2013). Several phases of the above-men-
tioned modes of oscillations occur at the same time (Fig.4).
It is observed that ENSO and positive IOD episodes mostly
occur during the same period. Allan etal. (2001) also show
evidence that supports the dependence, rather than the inde-
pendence, of ENSO and IOD. In addition, a positive AMO
Fig. 3 Annual precipitation
of Mauritius for the period
1981–1990 and homogene-
ous precipitation regions of
Mauritius
N.B.Raja, O.Aydin
1 3
episode persists during 1995–2010. As such, these oscil-
lations should be taken into consideration when analysing
climate variability. Table1 also shows the classification of
the years of ENSO and IOD events and identifies phases of
positive/negative IOD that occur at the same time as El Niño
and La Niña. This classification corresponds to the phases
observed in Fig.4. As for AMO, a negative phase persisted
until 1995, after which a positive phase is observed. A “pro-
tracted” La Niña episode is also observed in 1999, lasting
until 2004.
3.3 Preliminary tests
The homogeneity of the precipitation time series was
assessed by applying three tests: the standard normal homo-
geneity test (SNHT) (Alexandersson and Moberg 1997;
Asong etal. 2015), Buishand’s range (BR) test (Buishand
1982) and the Pettit test (Pettitt 1979), at a 5% significance
level for each station. A homogenous dataset implies that
measurements were taken during a certain period of time
with the same instruments (Kang and Yusof 2012). The
precipitation time series is considered homogeneous if the
critical values of SNHT, BR, and Pettitt statistics are less
than 7.747, 1.50, and 133.
Non-random climate variations are usually represented
in the form of persistence, trend, and periodic fluctuations.
The first step towards the physical explanation of the climate
variations is to determine their serial correlation, or auto-
correlation statistical significance (Rodriguez-Puebla etal.
1998). All the precipitation time series data were first tested
for the presence of autocorrelation coefficient at lag – 1, r1,
at a 5% significance level using a two-tailed test:
(1)
r
1=
n−1
i=1
Xi−
̄
X
Xi+1−
̄
X
n
i=1
X
i
−
̄
X
2
.
Fig. 4 SST Anomalies of ENSO
(based on the Equatorial Pacific
SSTs), IOD (based on the
Equatorial Indian Ocean SSTs),
and AMO (based on the North
Atlantic SSTs). Data obtained
from: ESRL (2017)
Table 1 Classification of years
of positive/negative IOD along
with El Niño and La Niña
events (after Ummenhofer etal.
2009)
This is a revised version of classification provided by Meyers etal. (2007), altered and updated to fit the
study period. “Protracted” ENSO episodes are in italic, as per Allan etal. (2003)
Negative IOD Neutral Positive IOD
El Niño 2010 1986, 1987, 2009 1982, 1991, 1997
Neutral 1980, 1985, 1989, 1992 1990, 1993, 1995, 2001, 2002, 2003, 2005,
2006, 2008
1994, 2004, 2006
La Niña – 1981, 1984, 1988, 1996, 1998, 2000, 2007 1999
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
The 5% critical values of the autocorrelation at any lag
d(d≠0)
are calculated as
where
T
is the sample size.
Cr
, in this study, is calculated to
be 0.358 for lag – 1.
As shown in Table2, all the precipitation series were
found to be homogenous and the autocorrelations at lag − 1
(
r1
) do not show evidence of a significant persistence. As
such, no homogenisation or pre-whitening methods were
required in this study.
(2)
C
r=±
1.96
√T
−
d
,
The preliminary analysis for this study also included com-
puting the mean, standard deviation, coefficient of skew-
ness, coefficient of kurtosis, and coefficient of variation in
the annual precipitation time series for each station. Table3
presents these parameters for the 30-year time period studies
(1981–2010). The mean annual precipitation varied between
619.2mm at Albion in the western region of the country
(Region 1) and 3924.5mm at Bois Cheri in the region of
the Black River National Park (Region 3). The coefficient
of skewness varied from −0.103 to 1.460; kurtosis varied
between −1.119 and 3.958. For the time series data to be
considered normally distributed, the coefficient of skew-
ness and kurtosis must be equal to 0 and 3, respectively.
Table 2 Results of homogeneity tests for mean annual precipitation for the period 1981–2010
For homogenous series,
T0
<
7.47
,
R
∕
√
n<
1.50
, and
Xk
<
133
,
r1
<
0.37
Station Homogeneity tests Station Homogeneity tests
SNHT Buishand’s
range
Pettitt test Autocor-
relation
coefficient
SNHT Buishand’s
range
Pettitt test Autocor-
relation
coefficient
T0
R
∕
√n
Xk
r1
T0
R
∕
√n
Xk
r1
Plaisance 3.636 0.569 64 −0.22 Constance 2.425 0.603 127 0.19
Vacoas 2.535 0.737 55 −0.11 Etoile 5.364 0.629 76 −0.10
Pamplemousses 3.376 0.636 84 −0.04 Mare La Chaux 3.246 0.607 79 −0.05
Fuel 5.676 0.734 119 0.00 Queen Victoria 4.244 0.747 59 −0.03
Labourdonnais 2.119 0.649 57 −0.07 Grosse Roche 3.885 0.703 105 0.17
Mon Choisy 3.518 0.689 58 −0.09 Piton du Milieu 6.740 0.641 83 0.19
Mon Piton 7.234 0.617 47 0.23 Alma 2.583 0.675 126 −0.03
Mon Loisir 3.775 0.642 102 −0.03 Arnaud 2.107 0.671 60 −0.09
California 1.833 0.699 77 −0.13 Petrin 2.516 0.621 58 −0.15
Notre Dame 5.153 0.615 80 0.17 Belle Rive 3.375 0.666 63 −0.15
Riche Terre 2.119 0.658 105 −0.12 Reduit 2.234 0.605 59 −0.33
Medine 1.357 0.750 79 −0.17 Union Park 1.819 0.580 39 −0.18
Albion 6.253 0.717 57 −0.13 Bois Cheri 4.507 0.568 47 −0.04
Barkly 4.116 1.064 121 −0.11 Britannia 3.208 0.645 93 −0.26
Clarens 7.104 0.712 60 0.20 Fort William 1.360 0.637 69 −0.12
Petite Riv. Noire 1.629 0.630 129 −0.16 M. D. Alma 2.583 0.675 54 −0.03
Yemen 2.818 0.697 51 −0.10 Quatre Bornes 2.651 0.666 60 −0.23
Case Noyale 3.129 0.555 55 −0.08 Riche en Eau 3.947 0.554 83 −0.03
Beaux Songes 4.380 0.789 49 −0.09 Sans Souci 4.936 0.664 91 −0.04
Le Morne 1.871 0.649 99 −0.10 Industrie Cour 6.133 0.502 97 0.08
Plaine Cham-
pagne
1.593 0.682 57 −0.22 Q. Militaire 5.072 0.753 89 −0.08
Bel Ombre 2.970 0.641 57 −0.07 Terracine 3.566 0.542 103 −0.07
Luchon 5.808 0.647 68 −0.08 Beau Rivage 2.164 0.707 99 −0.14
Ferney 6.957 0.617 98 0.34 Mon Vernon 6.044 0.528 83 0.03
Pte Aux Feuilles 1.651 0.725 120 −0.16 Henrietta 4.254 0.626 83 0.04
Savannah 1.931 0.593 65 −0.20 Mon Loisir
Rouillard
3.715 0.661 91 −0.07
Clemencia 7.361 0.634 49 0.16
N.B.Raja, O.Aydin
1 3
Table3 indicates that the data are positively skewed and not
normally distributed. Hence, the usage of non-parametric
statistical tests is appropriate and justified. The coefficient
of variation, also a measure of dispersion around the mean,
was also calculated to analyse the spatial variability of
annual precipitation for each station. This coefficient varied
between 16.6% (Union Park) and 51.5% (Barkly). The aver-
age variation of the precipitation over the whole region of
Mauritius was 27.5%.
3.4 Statistical trend analysis
Before the statistical trend analysis is carried out, the time
series for Mauritius and that for each of the subregions will
be analysed. A LOWESS curve (Cleveland 1979, 1984; Hel-
sel and Hirsch 1992) will also fitted to the time series based
on the total annual mean precipitation data. A 5-year inter-
val was chosen to explore the temporal trends in the data,
based on the study of Nicholson (2000) who states that the
most prominent time scales of variability average to approxi-
mately 5years in the Western Indian Ocean. Therefore,
5-year averages of total precipitation were calculated using
yearly averages for the following time intervals: 1981–1985,
1986–1990, 1991–1995, 1996–2000, 2001–2005, and
2006–2010. The trend analysis will also be carried out for
the whole study period 1981–2010.
3.4.1 Mann–Kendall
The Mann–Kendall method (Kendall 1948; Mann 1945)
was applied to detect statistically significant trends in pre-
cipitation for (1) individuals stations, (2) different regions
of Mauritius, and (3) Mauritius as a whole. The Mann–Ken-
dall statistics
S
, variances
V(S)
, and the standardized test
statistics
Zmk
(Mann–Kendall Coefficient) are computed as
follows:
Table 3 Summary of total mean annual precipitation statistics for study stations for the period 1981–2010
Cs
skewness,
Ck
, kurtosis,
Cv
coefficient of variation
Station Mean Standard
deviation
Cs
Ck
Cv
Station Mean Standard
deviation
Cs
Ck
Cv
Plaisance 1647.3 419.1 1.231 2.510 0.254 Constance 1501.7 457.4 0.637 0.226 0.305
Vacoas 1966.5 420.0 0.484 −0.267 0.214 Etoile 2650.3 665.7 1.175 0.846 0.251
Pamplemousses 1295.7 373.9 1.112 1.578 0.289 Mare La Chaux 1365.7 386.8 0.973 1.197 0.283
Fuel 2070.0 512.2 0.256 −0.492 0.247 Queen Victoria 2134.7 533.4 0.139 −0.065 0.250
Labourdonnais 1342.1 366.3 0.415 0.040 0.273 Grosse Roche 3341.9 803.7 0.530 −0.619 0.240
Mon Choisy 993.3 275.4 0.934 0.401 0.277 Piton du Milieu 3079.1 801.8 0.799 0.429 0.260
Mon Piton 1585.2 594.7 1.079 0.980 0.375 Alma 2741.7 657.0 0.559 −0.474 0.240
Mon Loisir 1334.8 390.8 0.974 0.365 0.293 Arnaud 3671.3 641.8 0.070 −0.398 0.175
California 2293.4 516.4 0.703 0.275 0.225 Petrin 3679.4 657.9 0.498 −0.148 0.179
Notre Dame 1369.9 461.8 1.009 0.711 0.337 Belle Rive 3100.3 712.4 1.253 2.230 0.230
Riche Terre 1035.8 328.1 0.694 0.459 0.317 Reduit 1426.4 458.9 1.040 0.557 0.322
Medine 759.6 245.8 0.361 −0.341 0.324 Union Park 3456.2 572.4 0.264 0.110 0.166
Albion 619.2 291.8 0.885 −0.159 0.471 Bois Cheri 3924.5 709.7 0.484 0.056 0.181
Barkly 800.7 412.0 1.052 0.854 0.515 Britannia 2513.6 477.9 0.628 0.399 0.190
Clarens 771.5 339.4 0.799 0.819 0.440 Fort William 729.8 302.7 1.460 2.773 0.415
Petite Riv. Noire 755.6 237.9 0.419 0.438 0.315 M. D. Alma 2741.7 657.0 0.559 −0.474 0.240
Yemen 882.4 329.7 0.703 −0.683 0.374 Quatre Bornes 1418.3 374.7 0.617 −0.177 0.264
Case Noyale 1005.7 269.9 −0.051 −0.409 0.268 Riche en Eau 2161.6 553.8 0.625 0.595 0.256
Beaux Songes 958.6 332.2 0.425 −0.560 0.346 Sans Souci 3658.1 732.2 0.183 −0.752 0.200
Le Morne 893.3 271.5 0.098 −0.292 0.304 Industrie Cour 1648.1 525.2 1.079 1.142 0.319
Plaine Champagne 2941.2 619.6 0.443 0.321 0.211 Providence Q. Militaire 3057.4 699.1 0.031 −0.862 0.229
Bel Ombre 1322.6 288.8 −0.103 0.685 0.218 Terracine 1637.8 345.0 −0.096 0.567 0.211
Luchon 3515.9 802.0 0.211 −0.099 0.228 Beau Rivage 1429.5 374.4 0.629 0.917 0.262
Ferney 2361.4 725.6 −0.099 −1.119 0.307 Mon Vernon 2876.0 634.6 1.448 3.958 0.221
Pte Aux Feuilles 1702.2 476.0 0.551 0.360 0.280 Henrietta 1888.5 447.5 −0.009 −0.453 0.237
Savannah 1688.5 399.0 0.651 0.698 0.236 Mon Loisir Rouillard 1271.1 356.8 0.934 0.659 0.281
Clemencia 2264.3 566.7 0.078 −0.716 0.250
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
(3)
S
=
n−1
∑
i=1
n
∑
j=i+1
sgn
(
Xj−Xi
)
sgn
Xj−Xi
=
+1 if
Xj−Xi
>0
0 if
Xj−Xi
=0
−1 if
Xj−Xi
<
0,
(4)
V
(S)=1
18
[
n(n−1)(2n+5)−
q
∑
p
=
1
tp
(
tp−1
)
(2tp+5)
],
where
Xi
and
Xj
are time series observations in chronological
order,
n
is the length of the time series,
tp
is the number of
times for the
p
th value, and
q
is the number of tied values.
Positive
Zmk
values indicate an upward trend in the time
series; negative
Zmk
values indicate a negative trend.
(5)
Z
mk =
⎧
⎪
⎨
⎪
⎩
S−1
√V(S)
S+1
√V(S)
0 if S=0
S+1
√
V(S)
if S<
0
Fig. 5 Total mean annual
precipitation (black dots) and
LOWESS trend curves (red
line) for the region of Mauritius
during the period 1981–2010
Table 4 Summary of
precipitation characteristics
and trend statistics in 5-year
intervals for the period
1981–1990
*Significant trend at 5% significance level of two-tailed test
Period Mean Minimum Maximum Mann–Kendall Spearman’s rho
1981–1985 1979.3 1349.2 2910.5 0.00* −0.10*
1986–1990 2011.3 1457.3 2769.6 −0.20* −0.30*
1991–1995 1923.3 1532.8 2501.7 0.60 0.70*
1996–2000 1609.8 1106.4 1892.2 0.40* 0.40
2001–2005 2006.8 1711.3 2206.2 1.00 1.00*
2006–2010 1973.4 1713.0 2268.0 0.00* −0.10*
Entire study period 1917.3 1106.4 2910.5 0.09* 0.08*
N.B.Raja, O.Aydin
1 3
3.4.2 Spearman’s rho
The Spearman’s rho test (Lehmann etal. 2006; Sneyers
1990) was applied as a comparison with the Mann–Kendall
test. The test assumes that the time series data are inde-
pendent and identically distributed. The Spearman’s rho test
statistics
Rsp
and standardized statistics
Zsp
are defined as
where
Di
is the rank of the
i
th observation and
n
is the total
length of the time series data. Positive values of
Zsp
indicate
an increasing trend across the time series; negative values
represent the decreasing trends.
(6)
R
sp =1−
6
n
i=1
Di−i
2
n
n2−1
,
(7)
Z
sp =Rsp
√
n−2
1−R2
sp
,
4 Results
4.1 Trend analysis overthewhole region
The annual precipitation time series and the fitted LOW-
ESS curve for the whole island are illustrated in Fig.5. A
steady increase in precipitation is observed for the period
1981–1990 after which there was a dip in total annual mean
precipitation reaching a record low in 1999 at 1106mm.
An increase in precipitation is apparent as from the year
2000. Similar results are observed when precipitation char-
acteristics and trend analysis results are analysed (Table4).
Comparing trend analysis results, the performances of
Mann–Kendall and Spearman’s rho are relatively simi-
lar except for the 1981–1985 and 2005–2010 intervals,
where the Mann–Kendall statistic suggests no change in
terms of precipitation levels, but Spearman’s rho reveals a
decrease in total annual mean precipitation. It is noted that
the highest precipitation levels were experienced during
1986–1990, although maximum precipitation for the whole
Fig. 6 Histogram of total mean
annual precipitation (mm) for
the period a 1981–1990 and b
2001–2010
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
study period was observed during 1981–1985. The low-
est precipitation levels were observed during 1996–2000.
Mean precipitation levels at the beginning and end of the
study period were relatively similar. Likewise, the intervals
2001–2005 and 2006–2010 were characteristically similar
in terms of mean, minimum, and maximum precipitation
levels. Despite precipitation levels, a significant decrease
was experienced during 1981–1990. Precipitation trends
significantly increased again as from post-1990. A clear
discrepancy is seen between pre-1990 and post-2000. While
both periods saw a stabilisation of the precipitation, pre-
1990 data show a more erratic pattern and a larger range
as compared to post-2000, where the range is smaller and
precipitation levels are steadier. Moreover, the histograms
for pre-1990 and post-2000 precipitation data are positively
and negatively skewed, respectively (Fig.6). Before 1990,
annual precipitation lied mostly below 2000mm. In con-
trast, annual precipitation was mostly above 2000mm after
2000. An increase is observed over the whole study area
during the whole study period.
4.2 Trend analysis acrossdierent subregions
The annual precipitation time series and respective fitted
LOWESS curves for each region are illustrated in Fig.7.
Table5 presents the mean precipitation at each 5-year
Fig. 7 Total mean annual precipitation (black dots) and LOWESS trend curves (red line) for the subregions of Mauritius during the period
1981–2010
N.B.Raja, O.Aydin
1 3
interval for the period 1981–2010. Different patterns are
observed for different subregions, except for the period
1996–2000 which suffered the lowest mean precipitation
values across all the subregions. Mixed results are observed
when the result for Mann–Kendall and Spearman’s rho is
examined. An overall increasing trend is observed for the
whole study period, with two distinct intervals recognised:
1981–1995 and 1995–2010.
Region 1 The time series shows a slight decrease during
the period 1981–2000, after which precipitation levels stead-
ily increased until 2010. This pattern is also suggested by the
Mann–Kendall statistics. On the other hand, the Spearman’s
rho test showed a significant decrease in precipitation levels
until 2005. Both tests agree that overall; precipitation levels
increased since 1981 with a rate of change of 0.99mm/year.
Region 2 Precipitation levels gradually decreased
between 1981 and 1990, followed by a sharp decline dur-
ing 1990–1995, according to the time series. While the
Spearman test agrees with the graph during this period,
the Mann–Kendall statistic suggests an increase during this
period until 2010. Spearman’s rho records a decrease in
precipitation levels until 2005, after which an increase in
observed. On the other hand, Mann–Kendall results show
a decrease in precipitation during 1996–2000 and then an
increase in precipitation for the following period. For the
whole study period, Mann–Kendall suggests no change in
precipitation, while Spearman rho’s reveals a negative trend
with a rate of change of −4.93mm/year.
Region 3 The time series shows that from 1981 until the
mid-1990s, an increase in precipitation levels is experienced.
However, the LOWESS curve suggests that a sharp dip in
precipitation levels is observed between the mid-1990s and
1995, followed by a decrease of a lesser magnitude until
2010. The trend statistics reveal a different picture. An
increase in precipitation is observed until 1996. During
1996–2000, a decrease in precipitation is observed, after
which an increase is experienced until 2010. Overall, both
statistics show a decrease, of 1.51mm/year on average, in
precipitation since 1981.
Region 4 According to the graph, during 1981–1990,
precipitation levels remain stable. From 1990 to 1995, a
decrease is seen, after which an increase is observed up to
2010. On the other hand, other than a decrease in precipi-
tation during 1985–1990 suggested by the Mann–Kendall
statistics, both trend statistics show increases or no change
during the different intervals. Overall, both statistics agree
that an increase is observed in this region for the whole study
period with a rate of change of 4.36mm/year.
Region 5 An increase in precipitation levels is observed
from 1981 to 1990 according to the time series. From 1990
to 1995, a slight decrease is experienced. During the forth-
coming years until 2010, a sharp increase is observed. The
trend statistics relatively agree with the time series, showing
Table 5 Summary of precipitation characteristics and trend statistics of subregions in Mauritius at 5-year intervals for the period 1981–1990
*Significant trend at 5% significance level of two-tailed test
Rate of
change (mm/
year)
Statistics 1981–1985 1986–1990 1991–1995 1996–2000 2001–2005 2005–2010 Whole study period
Region 1 0.99 Mean 1007.8 955.5 939.6 758.6 1050.7 1037.6 956.2
Zmk
−0.20* −0.20* 0.60 0.20* 0.20* 0.40* 0.10
Zsp
−0.20* −0.10* −0.10* −0.10* −0.10* 0.20* 0.10*
Region 2 −4.93 Mean 3181.1 3165.1 3041.0 2533.2 3054.5 3033.3 2989.8
Zmk
0.00* −0.20* 0.40* −0.20* 1.00 0.00* 0.00*
Zsp
−0.30* −0.30* −0.10* −0.50* −0.30* 0.10* −0.03*
Region 3 −1.51 Mean 3546.8 3753.1 3802.6 3223.8 3661.0 3501.4 3554.1
Zmk
0.00* 0.00* 0.40* 0.00* 0.80 0.20* −0.03
Zsp
0.70 0.40* 0.60 0.40* 0.70 0.70* −0.07*
Region 4 4.36 Mean 1900.7 1987.3 1805.8 1518.6 2016.5 2031.6 1870.1
Zmk
0.00* −0.40* 0.40* 0.20* 1.00 0.40* 0.14
Zsp
0.30* 0.00* 0.10* 0.30* 0.10 0.00 0.15
Region 5 2.59 Mean 1445.5 1473.8 1293.9 1078.8 1329.9 1523.3 1356.5
Zmk
0.00* −0.20* 0.60 0.00* 1.00 −0.20* 0.08*
Zsp
0.30* 1.00 0.90 1.00 1.00 0.50* 0.08*
Region 6 0.33 Mean 1686.4 1727.3 1709.2 1479.1 1779.9 1696.2 1674.9
Zmk
0.20 0.00* 0.60 −0.20* 0.40* −0.20* 0.09*
Zsp
0.40 −0.10* 0.00* 0.60 −0.20* −0.30* 0.08*
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
a decrease in precipitation during 1996–2000 instead of
1991–1995. Both statistics show an increase, of 2.59mm/
year, in mean precipitation during this period.
Region 6 From 1981 until the mid-1990s, the time series
shows that precipitation levels remain stable. From the
mid-1990s to 1995, a slight decrease is seen, after which an
increase is observed up to 2010. Mann–Kendall statistics
agree with the time series. Spearman’s rho, on the other
hand, shows that precipitation trends kept changing during
the different intervals. However, an overall increase is shown
by both trend statistics, with a rate of change of 0.33mm/
year.
While Mann–Kendall and Spearman’s rho test results
show mixed results for the 5-year intervals, they agree with
each other when applied for the whole study period. It is
noted that Spearman’s rho was generally more sensitive to
the changes in precipitation as compared to the Mann–Ken-
dall test. Figure8 shows the regions of Mauritius with
increasing and decreasing trends, according to Spearman’s
rho statistical tests. It is observed that while most of the
regions in Mauritius experienced an increase in mean total
annual precipitation, two of the highland regions, namely,
Regions 2 and 3, experienced a decline in precipitation lev-
els during the study period.
5 Discussion andconclusions
This study investigated variability in annual precipitation in
Mauritius over a 30-year study period (1981–2010). Precipi-
tation trends were analysed for each subregion of the island
and also the island as a whole.
Overall, the trend analysis showed an increase in precipi-
tation levels in Mauritius. This increase was also represented
in the coastal subregions of the island. It can be argued that
this increase is the result of the increase of non-cyclonic
extreme precipitation events that have seen an increase in
the recent climate of Mauritius (Sumner etal. 2016). This
is due to the intensification of the hydrological cycle of the
island as the globe warms due to non-uniform warming
of the oceans, leading to enhanced precipitation. Models
also show that climate change may increase the frequency
of ENSO warm phases by increasing the warm pool in the
tropical western Pacific or by reducing the efficiency of heat
Fig. 8 Regions of Mauritius
with increasing and decreasing
trends, according to Spearman’s
rho statistical tests and location
of water bodies
N.B.Raja, O.Aydin
1 3
loss (Dore 2005), thus causing a wetter climate in Mauritius.
However, it is seen that the two subregions located in the
highlands exhibited a significant decrease in precipitation
during the period 1981–2010. Seven out of the ten reservoirs
in Mauritius are located in the two inland subregions, where
a drying trend has been observed. This has negative implica-
tions for water resources and availability in Mauritius, which
relies mainly on reservoirs for its water supply.
This spatial trend can be attributed to the increase of
extreme precipitation events, affecting the whole island,
whose effect is negated in the inland subregions by a sig-
nificant decrease in “normal” precipitation. The increase
in extreme precipitation is associated with the increase of
the frequency of extreme positive IOD events as a result of
greenhouse warming (Taylor etal. 2013; Cai etal. 2014).
Two distinct increasing intervals were also observed, which
coincide with the transition of the AMO from negative to
positive. However, a positive AMO phase would normally be
associated with decreased precipitation in Mauritius. Hence,
it can be assumed that the resulting effect of the positive
AMO phase has likely strengthened or weakened other pro-
cesses which would have likely increased or decreased pre-
cipitation on the island.
The results of this study agree with that of Dore and
Singh (2013) who computed future precipitation scenarios
for Mauritius. They used the central point of the island,
namely, the city of Vacoas-Phoenix as a frame of reference
for measuring precipitation of Mauritius and observed a
decline in net annual precipitation. Their results coincide
with that for Region 2, where their study area is located.
A baseline study conducted in Region 5 also shows signs
of decreased precipitation amounts during the period
1960–2006 (Fairhurst etal. 2011). This complies with the
current study which observed a decrease in precipitation in
most regions of the island, mainly during 1996–2000. This
specific period coincides with the most severe drought expe-
rienced on the island, lasting from late 1998 to the beginning
of 2000.
Droughts in the eastern African regions usually occur as
a result of a cold phase of ENSO (Masih etal. 2014). The
period 1996–2000, which observed a significant decrease
in precipitation, corresponds to a “protracted” La Niña epi-
sode. Another drought period, not considered in this study,
occurred in 2010–2011, and is recognised to be one of
the worst water crises of Mauritius, after the 1998–2000
drought (Selvon 2012). Drought conditions in the Indian
Ocean Region, along with cold ENSO phases, are also exac-
erbated by the warming of the Ocean accelerated by green-
house gas and aerosol emissions after the latter half of the
20th century. The latter has been correlated with the decline
in precipitation over the East African region (Funk etal.
2008; Williams and Funk 2011). These studies suggest that
the warming of the central Indian Ocean drives changes in
the local Walker circulation, causing a reduction in seasonal
precipitation and inducing drought conditions in East Africa.
As such, there is also a need to study precipitation trends on
the seasonal scale.
Nowbuth and Saiboo (2009) who studied the response of
two catchments in Mauritius, located in highlands (Regions
3) and East (Region 4), respectively, showed that the catch-
ment located east of the island was more resistant than its
highland counterpart. Due to its varying topography and
microclimate, different regions of the island will respond dif-
ferently to increased variability in precipitation. It is, there-
fore, imperative to understand the catchment characteristics
of these regions to understand how they will respond to dif-
ferent scenarios. Available evidence from the past shows that
the African region is likely to face a decline in precipitation
in the future, including extreme and widespread droughts
(Masih etal. 2014).
In addition, Shongwe etal. (2011) who investigated pro-
jected changes in mean and extreme precipitation in the
Eastern African region state that the change in the local
Walker circulation is consistent with an increase in East
Africa precipitation, as in the case of Mauritius, as compared
to other regions within the same latitudinal belt. However,
this increase in precipitation is associated with an increase in
anomalously intense short precipitation period. The results
of the study point to a wetter climate with more intense wet
seasons. This is true, in the case of Mauritius, where the wet
summer season is becoming wetter, but the dry winter sea-
son is also getting drier (Dore and Singh 2013). According
to the Mauritius Meteorological Services (2009), despite a
decrease in the number of rainy days, heavy rainfall events
leading to numerous flash floods during the summer months
of February and March has increased. Furthermore, the fre-
quency and intensity of extreme weather events, heavy rains,
and tropical cyclones have substantially increased over the
last two decades.
Increased precipitation variability will influence water
resources, both surface and groundwater. Kundzewicz and
Doll (2009) state that many sub-humid and humid areas will
suffer from freshwater stress due to a decrease in groundwa-
ter levels. An increase in extreme precipitation may result in
the infiltration capacity being exceeded and hence prevent
infiltration, thus hindering groundwater recharge. “Recharge
zones” are located around the shoreline in coastal areas in
Mauritius (Nowbuth and Saiboo 2009) and hence are situ-
ated in areas, where precipitation is naturally low. In addi-
tion, an increase in extreme precipitation and thus in runoff
may inhibit recharge, leading to increased water stress. In
addition, the highlands, where most of the reservoirs are
located, have observed a decrease in overall precipita-
tion, as seen in this study. In a country, where only 33% of
annual precipitation is stored (Ellayah and Nowbuth 2013),
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
the results reveal additional water stress if existing storage
facilities are kept.
The current precipitation trend in Mauritius suggests an
increase in annual precipitation. However, this positive trend
shows a contrasting water-constrained scenario with lim-
ited and/or depleted groundwater and surface water supply.
Nonetheless, predictions of regional climate change for the
next decades are characterized by high uncertainty (Hawk-
ins and Sutton 2009; Shepherd 2014). To support decision
makers, quantitative climate predictions, more specifically in
terms of regional and local changes in climate, are required.
Efficient water policies are based on the understanding of
the changing water resources of the region. This, therefore,
requires a deep analysis of the changing variability of water
resources. Climate variability has not been greatly investi-
gated in Mauritius. There is, therefore, a need for a more
thorough study of the water cycle, including the changing
precipitation trends. Since a likely change in seasons has
also been observed (Dore and Singh 2013; Mauritius Mete-
orological Services 2009), there is a need to quantify these
changes for further understanding of the past, current and
future hydrological dynamics of the island.
References
Ahmad I, Tang D, Wang T etal (2015) Precipitation trends over time
using Mann-Kendall and Spearman’s rho tests in swat river basin,
Pakistan. Adv Meteorol 431860:1–16
Alexandersson H, Moberg A (1997) Homogenization of Swedish
temperature data. Part I: homogeneity test for linear trends. Int J
Climatol 17(1):25–34
Allan R, Chambers D, Drosdowsky W etal (2001) Is there an Indian
Ocean Dipole, and is it independent of the El Nino—Southern
Oscillation? CLIVAR Exch 6:18–22
Allan RJ, Reason CJC, Lindesay JA etal (2003) Protracted’ ENSO
episodes and their impacts in the Indian Ocean Region. Deep-Sea
Res II: Trop Stud Oceanogr 50(12–13):2331–2347
Annamalai H, Potemra J, Murtugudde R etal (2005) Effect of precon-
ditioning on the extreme climate events in the Tropical Indian
Ocean. J Clim 18(17):3450–3469
Argent RE, Raleight NC, Semazzi F etal. (2016) Influence of the
Atlantic Ocean on the interannual variability of the Eastern Africa
rainfall (96th American Meteorological Society Annual Meeting).
In Peter Lamb Symposium, Oklahoma
Ashok K, Wing-Le C, Motoi T etal (2004) Decadal variability of the
Indian Ocean Dipole. Geophys Res Lett 31(L24207):1–4
Asong ZE, Khaliq MN, Wheater HS (2015) Regionalization of pre-
cipitation characteristics in the Canadian Prairie Provinces using
large-scale atmospheric covariates and geophysical attributes.
Stoch Env Res Risk Assess 29(3):875–892
Birol F, Morrow R (2001) Source of the baroclinic waves in the south-
east Indian Ocean. J Geophys Res: Oceans 106(C5):9145–9160
Brandt P, Stramma L, Schott F etal (2002) Annual Rossby waves in
the Arabian Sea from TOPEX/POSEIDON altimeter and insitu
data. Deep Sea Res Part II 49(7–8):1197–1210
Buishand TA (1982) Some methods for testing the homogeneity of
rainfall records. J Hydrol 58:11–27
Cai W, Santoso A, Wang G etal (2014) Increased frequency of extreme
Indian Ocean Dipole events due to greenhouse warming. Nat Int
J Sci 510:254–258
Chelton DB, Schlax MG, Samelson RM etal (2007) Global observa-
tions of large oceanic eddies. Geophys Res Lett 34(L15606):1–5
Chen H, Guo S, Xu C etal (2007) Historical temporal trends of hydro-
climatic variables and runoff response to climate variability and
their relevance in water resource management in the Hanjiang
basin. J Hydrol 344(3–4):171–184
Cleveland WS (1979) Robust locally weighted regression and smooth-
ing scatterplots. J Am Stat Assoc 74(368):829–836
Cleveland WS (1984) Graphs in scientific publications. Am Stat
38(4):261–269
Collins M, Knutti R, Arblaster JM etal (2013) Long-term climate
change: projections, commitments and irreversibility. In: Stocker
TF, Qin D, Plattner GK, Tignor M, Allen SK, Boschung J, Nau-
els A, Xia Y, Bex V, Midgley PM (eds) Climate change 2013:
the physical science basis, Contribution of Working Group I to
the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cambridge, pp
1029–1136
Dahmen ER, Hall MJ (1990) Screening of hydrological data: tests
for stationarity and relative consistency. Wageningen, The
Netherlands
Dhurmea KR, Boojhawon R, Rughooputh SDDV (2009) Geostatistical
approaches for estimating rainfall over Mauritius. In 3rd Research
Week 2009–2010, Mauritius
Dore MHI (2005) Climate change and changes in global precipitation
patterns: what do we know? Environ Int 31:1167–1181
Dore MHI, Singh RG (2013) Projected future precipitation scenarios
for a small island state: the case of Mauritius. Clim Change Water
Resour 25:47–66
Ellayah R, Nowbuth MD (2013) Adapting to the impacts of climate
change in the water sector. The Journal of the Institution of Engi-
neers Mauritius, 50–56
ESRI (2013) ArcGIS. Retrieved January 20, 2017, from https ://www.
arcgi s.com/featu res/index .html
ESRL (2017) NOAA Earth System Research Laboratory. Retrieved
November 16 2017, from https ://www.esrl.noaa.gov/
Fairhurst L, Rowswell P, Jochlinger I (2011) Sub-Saharan African
cities: a five-city network to pioneer climate adaptation through
participatory research&local action. Port Louis Baseline Study,
Local Governments for Sustainability (ICLEI)
Feng M, Meyers G (2003) Interannual variability in the tropical Indian
Ocean: a two-year time-scale of Indian Ocean Dipole. Deep Sea
Res Part II 50(12–13):2263–2284
Fowdur SC, Rughooputh SDDV, Cheeneebash J etal (2014) Rainfall
analysis over Mauritius using principal component analysis. Envi-
ron Manag Sustain Dev 3(2):94–108
Funk C, Dettinger MD, Michaelsen JC etal (2008) Warming of the
Indian Ocean threatens eastern and southern African food security
but could be mitigated by agricultural development. Proc Natl
Acad Sci USA 105(32):11081–11086
Hawkins E, Sutton R (2009) The potential to narrow uncertainty in
regional climate predictions. Bull Am Meteor Soc 90:1095–1107
Helsel DR, Hirsch RM (1992) Statistical methods in water resources.
Volume 49 1st Edition, Elsevier Science, USA
Kahya E, Kalaycı S (2004) Trend analysis of streamflow in Turkey. J
Hydrol 289(1–4):128–144
Kang HM, Yusof F (2012) Homogeneity tests on daily rainfall series in
Peninsular Malaysia. Int J Contemp Math Sci 7(1):9–22
Kelman I (2014) No change from climate change: vulnerability and
small island developing states. Geogr J 180(2):120–129
Kendall MG (1948) Rank correlation methods. Charles Griffin, London
N.B.Raja, O.Aydin
1 3
Krishnamurthy V, Kirtman BP (2003) Variability of the Indian
Ocean: relation to monsoon and ENSO. Quart J R Meteorol Soc
129(590):1623–1646
Kundzewicz ZW, Doll P (2009) Will groundwater ease freshwater
stress under climate change? Hydrol Sci J 54(4):665–675
Lehmann EL, Erich L, D’Abrera HJM (2006) Nonparametrics : sta-
tistical methods based on ranks. Revised Edition, Springer, USA
Li ZL, Xu ZX, Li JY etal (2008) Shift trend and step changes for runoff
time series in the Shiyang River basin, northwest China. Hydrol
Process 22(23):4639–4646
Longobardi A, Villani P (2009) Trend analysis of annual and seasonal
rainfall time series in the Mediterranean area. Int J Climatol
30(10):1538–1546
Mann HB (1945) Nonparametric tests against trend. Econometrica
13(3):245–259
Masih I, Maskey S, Mussá FEF etal (2014) A review of droughts on
the African continent: a geospatial and long-term perspective.
Hydrol Earth Syst Sci 18:3635–3649
Masumoto Y, Meyers G (1998) Forced rossby waves in the
southern tropical Indian Ocean. J Geophys Res Oceans
103(C12):27589–27602
Mauritius Meteorological Services (2009) Climate Change Impacts
on Mauritius
Mauritius Meteorological Services (2016) Mauritius Meteorological
Services. Retrieved July 16, 2016, from metservice.intnet.mu
Mavrogenis S, Kelman I, Mercer J etal. (2014) Comparing tools and
methodologies for climate change adaptation in small island
developing states [SIDS], 27–28 March. In ADAPTtoCLIMATE.
Nicosia, Cyprus
Mcleod AI (2011) Kendall rank correlation and Mann-Kendall trend
test. Retrieved from http://www.stats .uwo.ca/facul ty/aim
Meehl GA, Julie M, Arblaster JM etal (2003) Coupled ocean–atmos-
phere dynamical processes in the Tropical Indian and Pacific
Oceans and the TBO. J Clim 16(13):2138–2158
Meyers G, McIntosh P, Pigot L etal (2007) The years of El Niño, La
Niña, and interactions with the Tropical Indian Ocean. J Clim
20(13):2872–2880
Nicholson SE (2000) The nature of rainfall variability over Africa
on time scales of decades to millenia. Global Planet Change
26(1–3):137–158
Nigel R, Rughooputh SDDV (2010) Soil erosion risk mapping with
new datasets: an improved identification and prioritisation of high
erosion risk areas. CATENA 82(3):191–205
NOAA (2016) Historical El Nino/La Nina episodes (1950-present).
Retrieved January 24, 2017, from http://www.cpc.noaa.gov/produ
cts/analy sis_monit oring /ensos tuff/ensoy ears.shtml
Nowbuth MD, Saiboo A (2009) Analysis of catchments response to
severe drought event for improved water resources. Univ Mauri-
tius Res J 15(1):513–526
Padya BM (1989) Weather and climate of Mauritius. Mahatma Gandhi
Institute, Mauritius
Pebesma EJ (2004) Multivariable geostatistics in S: the gstat package.
Comput Geosci 30(7):683–691
Pettitt AN (1979) A Non-parametric approach to the change-point
problem. J Roy Stat Soc: Ser C (Appl Stat) 28(2):126–135
Pohlert T (2016) Non-parametric trend tests and change-point detection
Quartly GD, Buck JJH, Srokosz MA etal (2006) Eddies around
Madagascar—the retroflection re-considered. J Mar Syst
63(3–4):115–129
R Core Development Team (2016) The R Project for Statistical Comput-
ing. Retrieved September 11, 2016, from https ://www.r-proje ct.org/
Rao SA, Behera SK, Masumoto Y etal (2002) Interannual subsur-
face variability in the Tropical Indian Ocean with a special
emphasis on the Indian Ocean Dipole. Deep Sea Res Part II
49(7–8):1549–1572
Rodriguez-Puebla C, Encinas AH, Nieto S etal (1998) Spatial and tem-
poral patterns of annual precipitation variability over the iberian
peninsula. Int J Climatol 18(3):299–316
Saji NH, Goswami BN, Vinayachandran PN etal (1999) A dipole mode
in the Tropical Indian Ocean. Nat Int J Sci 401:360–363
Schott FA, Xie SP, McCreary JP Jr (2009) Indian Ocean circulation
and climate variability. Rev Geophys 47(1):1–46
Selvon S (2012) A new comprehensive history of Mauritius, vol 1.
MDSelvon, Mauritius
Senapathi D, Underwood F, Black E etal (2010) Evidence for long-
term regional changes in precipitation on the East Coast Moun-
tains in Mauritius. Int J Climatol 30(8):1164–1177
Shadmani M, Marofi S, Roknian M (2012) Trend analysis in refer-
ence evapotranspiration using Mann-Kendall and Spearman’s Rho
Tests in arid regions of Iran. Water Resour Manage 26(1):211–224
Shepherd TG (2014) Atmospheric circulation as a source of uncertainty
in climate change projections. Nat Geosci 7:703–708
Shongwe ME, van Oldenborgh GJ, van den Hurk B etal (2011) Pro-
jected changes in mean and extreme precipitation in Africa under
global warming. Part II: east Africa. J Clim 24(14):3718–3733
Sneyers R (1990) On the statistical analysis of series of observations.
Secretariat of the World Meteorological Organization
Staub CG, Stevens FR, Waylen PR (2014) The geography of rain-
fall in Mauritius: modelling the relationship between annual and
monthly rainfall and landscape characteristics on a small volcanic
island. Appl Geogr 54:222–234
Sumner P, Werner N, Rughooputh S etal (2016) Rainfall erosivity and
soil erosion risk assessment in tropical island environments: a case
study of Mauritius. Advances in geographical and environmental
sciences. Springer, Tokyo, pp 217–230
Taylor RG, Todd MC, Kongola L etal (2013) Evidence of the depend-
ence of groundwater resources on extreme rainfall in East Africa.
Nat Clim Change 3:374–378
Tierney JE, Smerdon JE, Anchukaitis KJ etal (2013) Multidecadal
variability in East African hydroclimate controlled by the Indian
Ocean. Nat Int J Sci 493:389–392
Ummenhofer CC, England MH, McIntosh PC etal (2009) What
causes southeast Australia’s worst droughts? Geophys Res Lett
36(L04706):1–5
Wang C, Kucharski F, Barimalala R etal (2009) Teleconnections of
the tropical Atlantic to the tropical Indian and Pacific Oceans: a
review of recent findings. Meteorol Z 18(4):445–454
Wang Q, Fan X, Qin Z etal (2012) Change trends of temperature and
precipitation in the Loess Plateau Region of China, 1961–2010.
Global Planet Change 92–93:138–147
Webster PJ, Moore AM, Loschnigg JP etal (1999) Coupled ocean–
atmosphere dynamics in the Indian Ocean during 1997–98. Nat
Int J Sci 401:356–360
White WB (2001) Evidence for coupled rossby waves in the annual
cycle of the Indo-Pacific Ocean. J Phys Oceanogr 31:2944–2957
White WB, Tourre YM (2007) A Delayed action oscillator shared
by the ENSO and QDO in the Indian Ocean. J Oceanogr
63(2):223–241
Williams AP, Funk C (2011) A westward extension of the warm pool
leads to a westward extension of the Walker circulation, drying
eastern Africa. Clim Dyn 37(11):2417–2435
Yamagata T, Behera SK, Luo JJ etal. (2013), Coupled ocean-atmos-
phere variability in the Tropical Indian Ocean. In Earth’s Climate:
the ocean–atmosphere interaction pp 189–211
Yang XL, Xu LR, Liu KK etal (2012a) Trends in temperature and
precipitation in the Zhangweinan river basin during the last 53
Years. Procedia Environ Sci 13:1966–1974
Yang ZF, Yan Y, Liu Q (2012b) The relationship of streamflow–pre-
cipitation–temperature in the yellow river basin of China during
1961–2000. Procedia Environ Sci 13:2336–2345
Trend analysis ofannual precipitation ofMauritius fortheperiod 1981–2010
1 3
Yaning C, Changchun X, Xingming H etal (2009) Fifty-year climate
change and its effect on annual runoff in the Tarim River Basin,
China. Quater Int 208(1–2):53–61
Yanming Z, Jun W, Xinhua W (2012) Study on the change trend of
precipitation and temperature in Kunming City Based on Mann-
Kendall analysis. Future Comput, Commun, Control Automation
AISC 119:505–513
Yu L, Rienecker MM (2000) Indian Ocean warming of 1997–1998. J
Geophys Res 105(C7):923–939
Yue S, Pilon P, Cavadias G (2002) Power of the Mann-Kendall and
Spearman’s rho tests for detecting monotonic trends in hydrologi-
cal series. J Hydrol 259(1–4):254–271
Zhang Q, Liu C, Xu C etal (2006) Observed trends of annual maxi-
mum water level and streamflow during past 130years in the
Yangtze River basin, China. J Hydrol 324(1–4):255–265
A preview of this full-text is provided by Springer Nature.
Content available from Meteorology and Atmospheric Physics
This content is subject to copyright. Terms and conditions apply.