Theoretical and Applied Climatology

https://doi.org/10.1007/s00704-023-04579-x

RESEARCH

Long memory and trend in time series of precipitation in Mozambique

Jone Lucas Medja Ussalu1,2 · Amin Bassrei3

Received: 13 February 2023 / Accepted: 25 July 2023

© The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2023

Abstract
Many climate studies in Mozambique have clearly identified signals of climate change, especially changes in the extreme
temperatures. Regarding precipitation, there is still a gap on the knowledge of how it is behaving due to both internal and
external factors in the climate system. In this study, we have investigated the existence of long-term correlations and trend in
time series of precipitation. Two databases were used for this purpose: in situ observations along the period of 1960–2020 and
the Climate Hazards group InfraRed Precipitation with Stations (CHIRPS) dataset, along the period from 1981 to 2021. We
have applied the rescaled-range analysis and the detrended fluctuation analysis for long memory investigation, and the linear
regression and Mann-Kendall methods for trend analysis. Results have shown the existence of long memory in precipitation
in most parts of Mozambique, being stronger in the southern and central regions and weakening toward the north of the
country. On the other hand, significant trend signals of precipitation were detected in some isolated areas of Mozambique,
presenting an increase in some regions such as the southern part of Manica and eastern of Inhambane provinces and a decrease
in other regions such as the coastal areas of Zambezia and Nampula. These findings indicate that the probability of a random
occurrence of precipitation is minimal, and the observed trends are likely to continue for a long period in future. Dry land
agriculture should be prepared to adapt to new precipitation regime in the regions mentioned hereof.

1 Introduction is associated with the slowly responding subsystems such as

oceans and large forestry areas (Yuan et al. 2014).
Climate variability usually exhibits long memory or a long- Mandelbrot (1967) related this scaling behavior to frac-
range dependence, meaning that the present climate observa- tals. A fractal is an object whose geometry presents infinite
tions may have long-term influence on future climate states self-similarities at different scales. In the context of climate
(Beran 1994; Yuan et al. 2014; Qiu et al. 2020). Long memory variables, self-similarity is defined in terms of the distribution
in time series can be understood as the existence of autocor- of the data series. The concept of fractal was introduced after
relation in observations at long lags, that is, observations are Hurst (1951) had formulated the rescaled-range (R/S) anal-
not independent; each observation is affected by the prece- ysis, while studying dams dimensioning over the Nile River.
dent events. This long-range dependence in climate variables Hurst’s work had culminated in an empirical relation, whose
parameter, known as Hurst exponent (H ), carries informa-
tion about the long-range dependence (long memory) of the
B Jone Lucas Medja Ussalu time series.
[email protected] Besides the R/S analysis, many other methods have been
Amin Bassrei proposed for detecting the existence of long memory in time
[email protected] series, such as the detrended fluctuation analysis (DFA), peri-
1
odogram method, aggregated variance method, and structure
Faculty of Agronomy and Forestry Engineering, Eduardo
Mondlane University, Julius Nyerere Avenue, n. 3453,
function method. With these methods, fractal and multifractal
Maputo 257, Maputo, Mozambique properties in different climatic variables have been investi-
2 Centre of Excellence in Agri-Food Systems and Nutrition,
gated by many researchers (Caballero et al. 2002; Vyushin
Eduardo Mondlane University, Praça 25 de Junho, Andar No. and Kushner 2008; Yuan et al. 2014; Qiu et al. 2020; Vera-
5, Edifício da Reitoria, Maputo, Mozambique Valdés 2021).
3 Institute of Geosciences, Federal University of Bahia, Rua Each proposed method has its own potential. When using
Barão de Jeremoabo, s/n, Salvador 40170-115, Bahia, Brazil only one method, in some cases, a false detection of long-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

range dependence is possible, due to the existence of noise The knowledge of local precipitation regime and dynam-
and strong trend signals in the data. While long memory ics is crucial, specially for water resources management and
is intrinsic in the data, noise and trend signals are caused for the agricultural sector in Mozambique, where in most
by external effects, and in particular, trend is usually sup- rural areas (90%), it is mainly practiced under dry land condi-
posed to have a smooth and monotonous or slowly oscillating tions, with limited usage of irrigation technologies, resulting
behavior (Kantelhardt et al. 2001). One example is the global in lower production levels (Salite and Poskitt 2019; Manuel
warming that is being driven by the intensification of green- et al. 2020). It is important to understand the current state
house gases concentration in the atmosphere, leading to and the future climate dynamics in the three major regions
an increasing heat-storage capacity. This particular climate of Mozambique (northern, central, and southern) for capac-
driver is increasingly being associated to human activities. ity building in this sector. For the southeast Africa, which
According to Arias et al. (2021), from the IPCC-AR5 to includes Mozambique, a general picture is provided through
IPCC-AR6 (fifth to sixth assessment reports of the Inter- the IPCC regional projections, indicating that precipitation is
governmental Panel on Climate Change), new techniques generally decreasing (IPCC 2014; Seneviratne et al. 2021).
and analyses have provided greater confidence in attributing This trend signal had been partially supported by Ussalu and
changes in regional weather and climate extremes to human Bassrei (2021) in their study about the climate dynamics of
influence. the southern region of Mozambique, and Machaieie et al.
Climate observations and projections for Mozambique (2020) have also found a decreasing trend in inter-annual
are reported by various researchers (Queiroz et al. 2007; variation of precipitation and increase of drought frequency
Mcsweeney et al. 2010; Machaieie et al. 2020; Ussalu and and severity over Quelimane city, located in the central region
Bassrei 2021; Mavume et al. 2021), and all have found of Mozambique.
climate change signals in Mozambique. For instance, sig- In this work, we investigate the existence of long memory
nificant changes in maximum and minimum temperatures and trend signals in the time series of precipitation over all
have been observed throughout all regions of the country. regions of Mozambique using two databases. One consisting
On the other hand, precipitation behavior throughout the of in situ observations from a gauging network belonging
country is poorly known, principally at local scale. This is to the National Institute of Meteorology (INAM) and the
partly due to the fact that, unlike temperature, precipitation other is a high resolution (0.05o ) gridded remote sensing
is by nature less representative, generally presenting great data combined with in situ stations from the Climate Haz-
variability, both spatially and temporarily. It is significantly ards group InfraRed Precipitation with Stations (CHIRPS)
influenced by local factors, so that globally, it does not present dataset. While the R/S analysis and DFA have been applied
clear trend (IPCC 2014, 2021; Arias et al. 2021), even at for long memory detection, the linear regression and Mann-
regional scale (Lim Kam Sian et al. 2021). Kendall techniques were applied for trend analysis.
The occurrence of precipitation over the southeastern
Africa region is influenced by the following climate drivers:
(i) The Inter-tropical Convergence Zone (ITCZ), (ii) cold
fronts from the south associated with the right arm of the 2 Materials and methods
South Atlantic anticyclone, (iii) continental depressions of
thermal origin, (iv) Indian Ocean sea surface temperature 2.1 Study area and data description
(SST) patterns, (v) tropical cyclone invasions, (vi) El-Niño
South-Oscillation (ENSO), and (vii) the Mozambique Chan- Mozambique is located between the parallels 10◦ 27’S and
nel trough (MCT) (Reason and Jagadheesha 2005; Pomposi 26◦ 52’S and the meridians 30◦ 12’E and 40◦ 51’E, on the
et al. 2018; Ambrosino et al. 2011; Barimalala et al. 2020). southeastern coast of Africa. Figure 1 shows the study area,
While the ITCZ has influence only over the northern region Mozambique territory, and its ten provinces (major admin-
of Mozambique, cold fronts affect the southern region only. istrative division): three in the southern region (Maputo,
The last two factors present a dipole-like impact on precipi- Gaza, and Inhambane), four in the central region (Man-
tation throughout Mozambique. For instance, with El-Niño, ica, Sofala, Tete, and Zambezia), and three in the northern
the southern and central regions experience a prolonged dry region (Nampula, Niassa, and Cabo-Delgado). The general
period, while excessive rainfall is observed in the northern. climate in Mozambique is tropical, with annual precipitation
The reverse happens with La-Niña (Reason and Jagadhee- amount increasing toward the northern region of the coun-
sha 2005). On the other hand, Barimalala et al. (2020) had try. From the plotting of annual precipitation in Fig. 1, it is
observed that strong MCT summers cause an increase in rain- observed that, while Gaza is the driest province with minimal
fall over northern Mozambique, while a rainfall deficit is seen annual precipitation close to 300 mm, Zambezia is the wettest
in southern Mozambique, and the opposite is observed with province of the country, with maximum annual precipitation
weak MCT summers. reaching 1900 mm.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

autoregressive moving average model with an exoge-

nous variable (SARMAX1 ( p, q)(P, Q, m)) to the
time series. Ponkina et al. (2021) have proved that
ARMA2 ( p, q) models provide better gap filling
in hydro-climate time series than classical meth-
ods, such as linear interpolation and multiple linear
regression. CHIRPS precipitation estimates on each
station along the period from 1981 to 2021 were
used as exogenous variable to the SARMAX model,
while gaps existing from 1960 to 1980 were filled
with exogenous set to “none” that is, by fitting a
SARMA model. One example of a gap filled time
series is illustrated in Fig. 2 for the station located in
Chimoio city. In this work, we have considered for
analyses only stations containing no more than 20%
of missing data along the entire period and having no
less than 30 years of observations. Table 1 shows the
names and geographical locations of gauging sta-
tions used in the study, including their respective
percentage of missing data.

At each station, in situ and CHIRPS datasets from the

Fig. 1 Map of the study area, Mozambique with province boundaries same period are highly correlated, as can be seen in Table 2
drawn, and the annual precipitation plotted (based on CHIRPS dataset and in the Taylor diagrams presented in Fig. 3. Although in
from 1981 to 2021). Red stars are the gauging stations used in the study
situ dataset is given importance as ground data at specific
point, yet, due to its poor spatial coverage, the use of CHIRPS
For this study, two databases were used and are described dataset was necessary, for assessing the spatial variability of
in the following items: long memory and trend signals and for guarantying consistent
results.
(i) Climate Hazards group InfraRed Precipitation with
Stations (CHIRPS): CHIRPS dataset is a new quasi- 2.2 Long memory investigation
global (50◦ S - 50◦ N), high resolution (0.05◦ ),
daily, pentadal, and monthly precipitation dataset. 2.2.1 Rescaled-range analysis
CHIRPS uses the Tropical Rainfall Measuring Mis-
sion Multi-satellite Precipitation Analysis version 7 The rescaled-range R/S analysis was formulated by Hurst
(TMPA 3B42 v7) to calibrate global Cold Cloud (1951) while studying dams dimensioning over the Nile
Duration (CCD) rainfall estimates (Funk et al. River. Hurst’s initial idea was to determine the maximum
2015). It also incorporates in situ observations from and minimum volumes in reservoirs (ideal capacity) taking
public data streams, private archives, and national into account the annual flows associated with the river dur-
meteorological agencies using a smart interpola- ing a certain period of some decades. For this purpose, Hurst
tion approach. CHIRPS data are available online analyzed a statistical variable called adjusted range (R) from
at https://data.chc.ucsb.edu/products/CHIRPS-2.0/ the cumulative river flows over time. Then, Hurst normalized
for the period from 1981 to present. this value of R by the standard deviation (S) of the sequence
(ii) In situ data: Collected from the National Institute
of Meteorology (INAM) gauging station network. 1 SARMAX( p, q)(P, Q, m) is actually an ARMA( p, q) model, except
Data consisted of monthly precipitation time series that SARMAX takes into account the seasonal component of order
during the period from 1960 to 2020. A total of (P, Q, m) in the time series, and an eXogenous variable.
27 gauging stations were evaluated. Existing gaps 2ARMA model is the combination of Auto-Regressive AR( p) and
in the time series were filled by fitting a seasonal Moving Average MA(q) models.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

Fig. 2 An illustrative example of gap filling in monthly precipitation nous variable. The SARMAX linear order was ( p = 3, q = 3), and the
series from Chimoio using SARMAX predictions. Shaded areas are the seasonal order was (P = 3, Q = 3, m = 12)
existing gaps in the time series. CHIRPS dataset was used as exoge-

Table 1 List of meteorological

No Latitude (o S) Longitude (o E) Station Name Missing Data (%) Region
stations and their respective
geographical positions and the 1 11.35 40.37 Mocimboa 4.10
percentage of missing data
2 12.98 40.53 Pemba 4.50
3 13.13 39.03 Montepuez 11.50
4 13.30 35.23 Lichinga 1.10 Northern
5 14.82 36.53 Cuamba 2.10
6 15.10 39.28 Nampula 0.50
7 16.22 39.90 Angoche 17.10

8 16.17 33.47 Tete 8.70

9 17.88 36.88 Quelimane 1.00
10 19.12 33.47 Chimoio 4.10 Central
11 19.33 33.23 Sussundenga 9.80
12 19.83 34.85 Beira/Aeroporto 2.80

13 22,00 35.32 Vilânculos 3.60

14 23.87 35.38 Inhambane 3.00
15 23.92 32.16 Massingir 10.60
16 24.05 34.72 Panda 5.90
17 24.48 35.02 Inharrime 14.40
18 24.52 33.00 Chókwe 2.50
19 24.72 33.88 Manjacaze 3.60
20 24.73 33.53 Maniquenique 3.50 Southern
21 25.03 33.10 Macie 11.10
22 25.05 33.63 Xai-Xai 3.00
23 25.37 32.80 Manhiça 1.50
24 25.92 32.57 Map/Mavalane 1.20
25 25.97 32.70 Map/Observatório 1.40
26 26.05 32.38 Umbeluzi 4.70
27 26.30 32.18 Changalane 0.60

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

Table 2 Values of Person’s

Station Correlation Station Correlation Station Correlation
correlation coefficient (r )
between in situ and CHIRPS Mocimboa 0.86 Montepuez 0.84 Pemba 0.90
data at each station, over the
same period Nampula 0.92 Angoche 0.90 Cuamba 0.92
Lichinga 0.94 Chimoio 0.92 Sussundenga 0.89
Beira/Aeroporto 0.87 Tete 0.90 Quelimane 0.91
Map/Mavalane 0.82 Map/Observatório 0.71 Umbeluzi 0.82
Changalane 0.84 Manhiça 0.81 Manjacaze 0.86
Macie 0.85 Xai-Xai 0.84 Maniquenique 0.88
Chókwe 0.88 Massingir 0.79 Inhambane 0.84
Inharrime 0.80 Panda 0.84 Vilânculos 0.85

to obtain what he called rescaled adjusted range, which is the in the past is likely to be followed by a decrease in future
R/S statistics, a dimensionless quantity. and vice-versa; (c) If H = 0.5, the time series has a purely
Considering a certain sequence X t (t = 1, 2, ..., N ) random behavior also known as white noise.
of random numbers (with N observations), not necessar- Here, the null hypothesis (H0 ) is that the Hurst exponent
ily independent, and defining the k-th partial sum as Yk = H = 0.5 (white noise) and the alternative hypothesis (H1 ) is
X 1 + X 2 + ... + X k (k = 1, 2, ..., τ ), τ being the number H = 0.5. A standard normal distribution of H is considered,
of elements of the partial sequence. Then, we define Rτ and a Z score is computed by the expression: Z = (H −
as 0.5)/σ H , where σ H is the standard deviation of H values
    obtained by repeating the process over different lags in the
k k range of τ . Given that we expect both anti-persistence and
Rτ = max Yk − Yτ − min Yk − Yτ , (1)
(1<k<τ ) τ (1<k<τ ) τ persistence in the time series, a two-sided Z test is used so that
at a selected level of significance (1 − α), the null hypothesis
and the R/S(τ ) is given by of white noise is rejected if the absolute value of Z is equal or
greater than Z α/2 (Z  ≥ Z α/2 ). In this work, we consider a
Rτ
R/S(τ ) =  confidence level of 90% (α = 0.1), and in some cases, 95%
τ  2 . (2)
1
τ k=1 X k − τ Yτ
1 (α = 0.05).

In analyzing this statistics for observations involving dif-

2.2.2 Detrended fluctuation analysis
ferent natural phenomena, Hurst found that there was a
function relating the values of R/S to the number of obser-
The detrended fluctuation analysis (DFA) was introduced
vations that entered in the calculations (τ ), and it is given by
by Peng et al. (1994) when studying Mosaic organization
the following expression:
of DNA nucleotide, and it is a method for determining the
τ H statistical self-affinity of a signal. It is useful for analyzing
R/S(τ ) = , (3) time series that appear to be long-memory processes. The
2
obtained exponent is similar to the Hurst exponent (Kantel-
where H is a constant known as the Hurst exponent. It is hardt et al. 2001), except that DFA may also be applied to
estimated as the slope of a straight line fit in a log-log graph signals whose underlying statistics such as mean and vari-
of R/S against τ . ance are non-stationary (changing over time).
For stationary processes, H ranges from 0 to 1 (Mandel- Given a time series xi (i = 1, 2, ..., N ), the DFA proce-
brot and Wallis 1969), having the following interpretation: dure consists of the following steps (Kantelhardt et al. 2001).
(a) If 0.5 < H < 1, the time series is said to be per- In the first step, we determine the profile X t :
sistent; it indicates positive long-term correlations, which
means that an increase in the past will tend to be followed by t
another increase in future; the opposite situation is also true, Xt = (xi − x̄) , (4)
a decrease in the past will probably be followed by another i=1
decrease in future; (b) If 0 < H < 0.5, the time series is
said to be anti-persistent, increasing and decreasing values where x̄ denotes the mean value of the time series, and X t is
occur alternately in adjacent pairs, meaning that an increase called the cumulative sum or profile.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

Fig. 3 Taylor diagrams, comparing in situ and CHIRPS datasets over the same period at each used station

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

Next, X t is divided into time windows of length n sam- decrease of the values in the time series, X̄ is the average
ples each, and a local least squares polynomial fit (the local value of the time series, and t¯ is the average value of t.
trend) is calculated by minimizing the squared errors within The estimated a is a random variable and has statistical
each time window. If the local fit applied to the profile is lin- properties such as standard error or bias. Confidence intervals
ear, thus it is called DFA1, it removes constant trends in the (CI) for a were computed assuming a Student’s t-distribution
original sequence (stationary process), and in this particular in the residual’s variability of observations with regard to the
case, DFA is equivalent to R/S analysis. For polynomial fit of straight line (Hartmann et al. 2013; Heumann et al. 2016).
order j, we have DFA j, and this may remove trends of higher At a given statistical significance level or confidence level
order (non-stationary process). Let Yt indicate the resulting (CL), the estimated trend slope is given as a ± . While a
piecewise sequence of polyfits. Then, the root-mean-square is the best guess or a point estimate, CI = [a − ; a + ]
deviation from the trend, the fluctuation, is calculated by is an interval estimate that informs how good a guess is. At
a given confidence level 0 ≤ 1 − α ≤ 1, the CI for a is
N CIa,1−α = [a L ; aU ], where a L = a −  is the lower and
1 aU = a +  is the upper endpoint of the CI.
Fn = (X t − Yt )2 . (5)
N In this work, we consider 90% of CL (α = 0.1) for the
t=1
CIs, which is a reasonable choice for climatological prob-
This process is repeated over a range of different window lems (Hartmann et al. 2013). CI has the property coverage
sizes n > j + 2, and a log-log graph of Fn against n is con- defined by the confidence level CL and the interval length
structed, which indicates statistical self-affinity expressed as 2, which should ideally be small. The null hypothesis of no
Fn ∝ n α . Finally, the scaling exponent α is calculated as trend is rejected only if both endpoints of CI have the same
the slope of a straight line fit, using least-squares. This expo- signal as a under the considered CL. Here, this is diagnosed
nent is a generalization of the Hurst exponent, and it gives by computing the ration between the absolute value of a and
the following information about the series self-correlations: the value of , obtaining a confidence index a/, which
α < 0.5: anti-correlated; α = 0.5: uncorrelated or white should be greater than unit (confidence index > 1), to accept
noise; α > 0.5: correlated; α = 1: 1/f-noise, pink noise; the alternative hypothesis of significant trend in the time
α > 1: non-stationary; α = 1.5: Brownian noise. In this series.
work, we have used the DFA1 (which is comparable to the
Hurst’s R/S analysis, α = H ) and the DFA2. Statistical 2.3.2 Mann-Kendall test
significance of estimated α is computed under the same con-
siderations described above. The Mann-Kendall (MK) test is a non-parametric trend statis-
tics that can be applied to time series not conforming to a
2.3 Trend analysis normal distribution. It is robust to time series presenting miss-
ing values, tied values, and values below the limit of detection
2.3.1 Linear regression (Taylor and Loftis 1989; Yu et al. 1993; Meals et al. 2011).
Let X = (x1 , x2 , x3 , ..., x N ) be a deseasonalized time
Linear regression (LR) is a parametric method for trend anal- series of N observations at a given station. The Mann-Kendall
ysis in climate series, and its application is well and deeply statistic (Mann 1945; Kendall 1975) is computed by the fol-
explained in Hartmann et al. (2013) and Heumann et al. lowing expression:
(2016). This method has been adopted for general use in the N −1 N
IPCC (2014) report for trend calculation. The trend parame- s= sgn(x j − xi ) , (7)
ter is determined as a slope of the best fitting straight line in i=1 j=i+1
the dispersion of the data series.
Let X n be the n-th value of a time series observed at a where
time tn (n = 1, 2, ..., N ). The trend slope is obtained by
⎧
the ordinary least squares solution given by the following
⎨ 1, if θ > 0,
⎪
expression: sgn(θ ) = 0, if θ = 0, (8)
⎪
⎩
N −1, if θ < 0.
¯
n=1 (tn − t )(X n − X̄ )
a= N , (6)
¯2
n=1 (tn − t ) A positive value of s indicates an upward trend, and a
negative value of s indicates a downward trend. According
where a is the trend slope (or the change rate) whose to Mann, the null hypothesis (H0 ) of this test states that X is
sign determines the trend direction, either an increase or a an independent and identically distributed random variable,

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

and the alternative hypothesis (H1 ) of a two-sided test is that 3.1 Long memory
the values of X are not distributed identically.
The variance of s is estimated by the expression (Hirsch For long memory detection, three different approaches were
et al. 1982): considered, the R/S, DFA1, and DFA2 methods. Overall,
precipitation had presented strong memory over the south-
1  ern region of Mozambique, weakening toward the northern.
V ar (s) = N (N −1)(2N +5)− t(t −1)(2t +5) , (9)
18 This result is consistent from both in situ and CHIRPS data.
t
Long memory exponents (H and α) and their respective con-
fidence intervals derived from in situ observations are listed
where t is the size of any existing tied data (a sequence of
in Table 3, while those derived from CHIRPS dataset are
repeated values) in the time series. The standard normal score
mapped in Fig. 4.
for statistical significance is computed by
In particular, from Table 3, we observe that at 95% con-
⎧ fidence level (Z  > 1.96), the southern and some central
⎪
⎨ V ar (s) , if s > 0,
⎪ √ s−1
stations (from bottom-most to middle stations in the tables)
Z= 0, if s = 0, (10) had presented the H exponent varying from 0.55 to 0.91,
⎪
⎪
⎩ √ s+1 , if s < 0. the DFA1 whose exponent is compared to the R/S method
V ar (s)
(α = H ) ranged from 0.54 to 0.96 and the DFA2 presented
values ranging from 0.56 to 1.26. These intervals indicate
For a two-sided Z test at a selected level of significance that the time series of precipitation were overall persistent
1−α, the null hypothesis of no trend is rejected if the absolute in the southern and in some central areas of Mozambique,
value of Z is equal or greater than Z α/2 (Z  ≥ Z α/2 ). which means that the correlation between past and future
Confidence levels of 90% (α = 0.1) and in some cases of trends exists and is positive. On the other hand, we can say
95% (α = 0.05) are considered. that the possibility of a random occurrence of the data series
Up to this point, we have only identified whether a trend is minimal. The behavior observed in the precipitation time
exists; still, it is also very important to estimate the magnitude series is more likely to persist for a long period in the future.
of a trend or the change rate. We have used the slope estimator Some northern stations (uppermost stations in the tables)
proposed by Sen (1968) and used by many other researchers such as Mocimboa and Pemba had presented H and α values
(Gan 1998; Xu et al. 2007; Fu et al. 2009), defined as below 0.5 with at least 90% significance level, suggesting
x − x  that time series on these particular stations are likely to be
j k
slope = median , (11) anti-persistent or even white noise.
j −k From the spatial mapping of H and α based on CHIRPS
dataset in Fig. 4, we can observe the full picture of how long
where 1 < k < j < N . The slope estimator is the median memory strength varies throughout Mozambique territory.
over all possible combinations of pairs in the time series. Quite similar results in terms of long memory strength are
For a particular time series of N observations, the number of presented between the annual precipitation (top panels) and
combinations will be N (N − 1)/2. precipitation within the two main seasons, during summer
months (middle panels) and during winter months (lower
panels), although slight differences in the spatial distribution
3 Results and discussion exist. While persistency is generally exhibited in the south-
ern and central regions, the northeastern part of Mozambique
Both long memory and trend analyses were evaluated to had presented anti-persistency tending to white noise (with
deseasonalized time series consisting of (i) total annual pre- H and α values ranging from 0.1 to 0.5), especially in annual
cipitation, (ii) total precipitation during the summer period, and summer precipitation over Cabo-Delgado province (refer
from November to February (NDJF), and (iii) total precipi- to Fig. 4a, b). This result is true from the three approaches.
tation during the winter period, from May to August (MJJA). A synthesis of overall long memory over each province
Mozambique presents two main seasons, the winter, which of Mozambique based on CHIRPS data is presented in
is relatively dry, and the summer, which is relatively wet Table 6.
(Machaieie et al. 2020; Ussalu and Bassrei 2021). Results The results presented above collaborate partially with the
derived from in situ observations are displayed through study from Araneda-Cabrera et al. (2021), which evaluated
tables, as they could not be interpolated and presented the spatio-temporal characteristics of drought in Mozam-
through colored maps, for being very sparse and not uni- bique and their relationship with large-scale climate vari-
formly distributed over the country, while results based on ability, using data from the Climate Research Unit (CRU). In
CHIRPS database are mapped. their study, they have computed the persistence of the Stan-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

Table 3 Long memory

Station name R/S DFA1 DFA2
exponents: Hurst exponent H ,
H Z score α Z score α Z score
DFA1 α, and DFA2 α, with their
respective Z score for statistical Mocimboa 0.45 ± 0.02 −2.39 0.37 ± 0.04 −3.22 0.49 ± 0.05 −0.21
significance
Pemba 0.47 ± 0.03 −0.80 0.25 ± 0.06 −3.93 0.42 ± 0.11 −0.70
Montepuez 0.66 ± 0.01 17.79 0.49 ± 0.04 −0.24 0.62 ± 0.03 4.66
Lichinga 0.69 ± 0.01 16.30 0.61 ± 0.02 5.34 0.74 ± 0.03 7.06
Cuamba 0.66 ± 0.01 13.35 0.63 ± 0.02 6.72 0.68 ± 0.02 10.91
Nampula 0.68 ± 0.02 11.75 0.73 ± 0.02 9.18 0.52 ± 0.08 0.28
Tete 0.62 ± 0.02 6.40 0.47 ± 0.03 −1.02 0.58 ± 0.06 1.23
Angoche 0.56 ± 0.02 3.38 0.51 ± 0.05 0.21 0.66 ± 0.06 2.70
Quelimane 0.55 ± 0.01 3.20 0.55 ± 0.01 3.56 0.56 ± 0.01 6.82
Chimoio 0.67 ± 0.01 24.95 0.61 ± 0.01 8.30 0.62 ± 0.12 0.97
Sussundenga 0.66 ± 0.01 19.47 0.65 ± 0.04 4.12 0.60 ± 0.06 1.55
Beira/Aeroporto 0.60 ± 0.02 4.09 0.64 ± 0.03 4.28 0.82 ± 0.04 8.77
Vilânculos 0.77 ± 0.01 35.73 0.71 ± 0.02 9.10 0.91 ± 0.03 12.08
Inhambane 0.65 ± 0.01 20.40 0.63 ± 0.01 8.90 0.74 ± 0.02 12.10
Massingir 0.75 ± 0.03 9.03 0.81 ± 0.03 11.36 0.72 ± 0.03 8.30
Panda 0.83 ± 0.01 25.33 0.77 ± 0.02 15.61 0.81 ± 0.02 19.26
Inharrime 0.91 ± 0.02 26.31 0.96 ± 0.02 19.36 1.26 ± 0.03 28.53
Chókwe 0.76 ± 0.01 17.15 0.71 ± 0.02 8.97 0.83 ± 0.02 19.55
Manjacaze 0.73 ± 0.03 7.63 0.62 ± 0.02 7.74 0.73 ± 0.02 10.59
Maniquenique 0.68 ± 0.01 23.02 0.69 ± 0.01 18.23 0.83 ± 0.04 8.14
Macie 0.70 ± 0.01 30.30 0.72 ± 0.02 11.10 0.89 ± 0.02 16.14
Xai-Xai 0.59 ± 0.01 7.11 0.52 ± 0.03 0.69 0.67 ± 0.03 4.75
Manhiça 0.85 ± 0.03 12.67 0.80 ± 0.03 10.26 1.05 ± 0.04 13.23
Map/Mavalane 0.60 ± 0.01 8.76 0.59 ± 0.04 2.37 0.66 ± 0.06 2.61
Map/Observatório 0.56 ± 0.02 2.48 0.48 ± 0.04 −0.47 0.68 ± 0.07 2.65
Umbeluzi 0.58 ± 0.02 3.99 0.52 ± 0.04 0.43 0.75 ± 0.04 5.88
Changalane 0.66 ± 0.01 10.61 0.54 ± 0.02 2.37 0.58 ± 0.04 1.94
Evaluated data consists of in situ time series of the period from 1960 to 2020

dardized Precipitation and Evapotranspiration Index (SPEI) Mozambique (refer to Fig. 4c), and in winter precipitation
through R/S method, having found H values greater than 0.5 in the central area of Tete province (refer to Fig. 4g–i), sug-
in all regions, but higher values of H were seen in the south- gesting that time series on these areas are non-stationary. It
ern and central regions, suggesting that the negative trends is very likely that some underlying statistics such as mean,
of SPEI they obtained in all regions will persist with greater standard deviation, or variance are not constant over time.
strength over these regions, while trend persistence strength If it is true, it is possible that over these particular areas,
will be weak in the northern region. Although the variable some local or regional climate controllers are modifying or
evaluated by Araneda-Cabrera et al. (2021) was SPEI and new ones emerging over time. At least, we are certain of
not precipitation directly, which is the case of our work, it the changes in frequency and intensity of tropical cyclones
is clear that an agreement exists on the distribution of the landing the coastal area of Mozambique in connection to
persistence strength throughout Mozambique. the global warming (Mavume et al. 2009). It might also be
According to Mandelbrot and Wallis (1969), when 0 < the effect of changes in other factors, and this could be a
H < 1, the occurrence of the phenomenon presents self- subject of further investigations, just as the study from Bari-
similarities as those of the fractal. A change of scale in malala et al. (2020), where the variability and impacts of the
the sampling of the series would lead to approximately the Mozambique Channel Trough on southeast African rainfall
same result. The DFA2 had indicated α > 1 (mapped with were investigated, or the study from Ambrosino et al. (2011),
blank areas) in annual precipitation along the coastal area where large-scale climate factors controlling the southern
of Inhambane, extending to the southern Sofala, southeast African rainfall were investigated.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

Fig. 4 Spatial distribution of long memory in precipitation from panel), (ii) the total summer (NDJF) precipitation (middle panel), and
CHIRPS database. The H exponent (left column), DFA1 α (central (iii) the total winter (MJJA) precipitation (lower panel). Blank areas
column), and DFA2 α (right column) were evaluated for the deseason- indicate non-stationary time series (where α > 1). Evaluation based on
alized time series consisting of (i) the total annual precipitation (top CHIRPS dataset during the period of 1981–2021

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

3.2 Trend Summer precipitation is behaving similarly to the annual,

just as it could be expected, given that this period coincides
For trend analysis, we have employed the parametric LR with the rainy season in Mozambique, thus containing the
and the non-parametric MK tests. Trend in annual precipita- major part of annual precipitation. Significant decrease in
tion throughout Mozambique territory has generally shown annual and summer precipitation is observed over the coastal
a considerable spatial variability. Both methods indicated a areas of Zambezia and Nampula, while the southern Manica
decreasing rainfall in some regions and an increase in others, and eastern Inhambane is experiencing a significant increase
although some trend signs are not statistically significant. (refer to Figs. 5a, c and 6a, d). Besides, winter precipitation

Fig. 5 Trend in precipitation by

linear regression. The trend
slope or the change rate is given
in mm/yr (left column) and is
evaluated for the deseasonalized
time series consisting of (i) the
total annual precipitation (top
panel), (ii) the total summer
(NDJF) precipitation (middle
panel), and (iii) the total winter
(MJJA) precipitation (lower
panel). Statistical significance is
shown in the right column.
Confidence intervals were
computed assuming a
t-distribution of the estimated
slope. Red plus signs in the right
column indicate locals where
the trend in precipitation is
significant with at least 90%
confidence level. Evaluation
based on CHIRPS dataset
during the period of 1981–2021

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

is decreasing over almost all the country, except at some iso- increase of 135.3 mm (about 15%) during the period of
lated points where a positive trend is true (refer to Figs. 5e 1981–2021. Overall trend in precipitation in each province
and 6g). of Mozambique based on CHIRPS data is also synthesized in
Based on in situ observations, trend results by LR and Table 6.
MK methods are presented in Tables 4 and 5, respectively. Most in situ stations are in line with the CHIRPS based
Most stations have exhibited negative trend signals, that is, a results in terms of trend signals. For instance, stations located
decreasing trend over time, with differences in significance in the vicinity of coastal areas of Zambezia (Quelimane
levels. Although most trend signals are not statistically signif- city) and Nampula (Angoche district) have also indicated
icant, some stations had presented a trend with at least 90% a decreasing rainfall, where the MK determined an average
confidence. Examples of these trends include the decreas- change rate of −4.51 and −3.49 mm/yr, equivalent to an
ing rainfall detected over Massingir and Inharrime and the average total decrease of about 181.6 mm (-13%) and 143.1
increasing trend over Sussundenga and Manhiça. mm (-10%), respectively, in annual precipitation during the
Using CHIRPS data, trend in precipitation by the LR period of 1981–2021. It should be noted that these trend mag-
method is shown in Fig. 5, while trend by the MK method nitudes cannot be compared to those referred previously from
is shown in Fig. 6. It can be observed that some iso- CHIRPS dataset, due to the fact that calculations were based
lated areas exhibited an exceptionally significant trend on different periods of data availability. On this particular
with 95% confidence. While an exceptionally significant area, similar trend results were obtained by Machaieie et al.
decrease in precipitation has been observed during both sea- (2020), while studying the variability and trend of precipi-
sons in the northeast Zambezia and southeast Nampula, an tation in Quelimane, having pointed as a possible cause of
increase is also observed in the southern Manica, eastern the decreasing rainfall, the strengthening of ENSO after the
Inhambane, and isolated parts of Tete province, being this 1970s climate shift.
last trend more pronounced in the summer period (90% These findings converge to the macro scale IPCC AR5
confidence). and AR6 projections over southeast Africa (IPCC 2014;
Some regions have presented different trend signals in Seneviratne et al. 2021). For instance, the CMIP53 multi-
between summer and winter seasons, just as it could be model ensemble mean of projected changes in precipitation
expected. This is certainly due to seasonal or dynamic climate for 2016–2035 relative to 1986–2005 under RCP4.54 sce-
factors, such as the cold fronts which present seasonal vari- nario, indicated, on average, negative anomalies (-10%) in
ations in position and strength, the tropical cyclones which the northern Mozambique in all seasons, while a positive
affect during the summer season only, SSTs with seasonal anomaly (10%) was also expected for summer months (DJF)
patterns, among others (Mavume et al. 2009, 2021; Bari- in the southern Mozambique (IPCC 2014). On evaluating
malala et al. 2020). As an example, in the eastern part of these regional projections for Mozambique, Mavume et al.
Inhambane province, precipitation has presented an increase (2021) have found similar spatial variability in precipitation
during the summer and a decrease during the winter (90% trend.
confidence). This result is consistent from the two methods The Working Group I of the IPCC-AR6 have presented
(refer to Figs. 5c, e and 6d, g), which converged to quite sim- precipitation projections separated into the average precip-
ilar trend signals in both databases, differing slightly in the itation trend and the trend on extreme precipitation at a
trend magnitudes. This agreement, somehow, provides more daily scale, having found that, while average precipitation
confidence in the results obtained hereof. The non-parametric is overall decreasing over the southeast Africa, the extreme
MK test provides higher statistical power in the case of non- precipitation at a daily scale is lightly increasing on the region
normality and is robust against outliers, and similar to the LR (Seneviratne et al. 2021). This last trend signal is certainly
method, it is also robust against large data gaps. Therefore, connected to the intensification of extreme events, such as
even if data gaps had not been filled, we would still expect the tropical cyclones and ENSO.
overall trustable trend results.
Considering CHIRPS data, the decreasing rainfall at the
coastal areas of Zambezia and Nampula provinces has pre- 4 Conclusions
sented a change rate of at least −4.1 mm/yr in annual
precipitation, meaning that, during the period from 1981 to Overall, precipitation exhibited strong long memory in south-
2021, the total annual precipitation has dropped on average ern and central Mozambique, weakening toward the northern
about 168.1 mm, which represents about -12% relative to the region. The southern and central regions are generally per-
average annual precipitation of this area. On the other hand,
the southern Manica and the eastern Inhambane had pre- 3 CMIP5 - Coupled Model Intercomparison Project Phase 5.
sented a significant increase in summer precipitation with 4 RCP4.5 is an acronym designated to represent the intermediate sce-
a change rate of at least 3.3 mm/yr, equivalent to a total nario of greenhouse gas emission policy.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

Fig. 6 Trend in precipitation by the Mann-Kendall test. The parameter (middle panel), and (iii) a set of total winter precipitation (MJJA) (lower
s (left column), the trend slope or change rate given in mm/yr (central panel). Red plus signs in the right column indicate locals with at least
column), and the Z score for statistical significance (right column) were 90% significance level of the given trend. Evaluation based on CHIRPS
evaluated for deseasonalized time series consisting of (i) a set of annual dataset during the period of 1981–2021
precipitation (top panel), (ii) a set of total summer precipitation (NDJF)

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

Table 4 Trend in precipitation

Station name Change rate a in precipitation (mm/yr)
by linear regression
Annual precipitation Summer precipitation Winter precipitation

Mocimboa −0.22 ± 3.84 0.16 ± 2.89 −0.06 ± 1.13

Pemba 1.76 ± 3.16 3.73 ± 2.78 −0.26 ± 0.86
Montepuez 3.51 ± 5.97 4.90 ± 5.60 0.27 ± 0.67
Lichinga 1.06 ± 3.33 1.78 ± 2.33 −0.41 ± 0.53
Cuamba −1.97 ± 4.14 1.82 ± 3.04 −0.86 ± 0.51
Nampula 5.80 ± 3.37 5.13 ± 3.60 −0.15 ± 0.70
Tete 3.25 ± 3.29 4.32 ± 3.19 −0.04 ± 0.17
Angoche −1.69 ± 4.81 −1.51 ± 3.92 −0.90 ± 0.77
Quelimane −4.48 ± 5.11 −0.87 ± 3.69 −1.58 ± 1.25
Chimoio −2.95 ± 4.44 −3.21 ± 3.98 −0.93 ± 0.75
Sussundenga 13.73 ± 11.93 9.90 ± 10.23 0.32 ± 0.99
Beira/Aeroporto 0.57 ± 5.67 2.18 ± 4.86 −0.90 ± 1.50
Vilânculos 0.69 ± 4.30 −0.40 ± 3.77 −0.25 ± 0.84
Inhambane −0.12 ± 3.03 0.35 ± 2.67 −0.92 ± 0.97
Massingir −7.88 ± 4.98 −3.79 ± 3.57 −1.69 ± 0.94
Panda −0.15 ± 2.98 0.30 ± 2.41 −0.34 ± 0.78
Inharrime −5.98 ± 6.36 −3.83 ± 3.81 −2.04 ± 2.36
Chókwe −0.63 ± 3.77 −1.29 ± 3.00 0.15 ± 0.75
Manjacaze 3.81 ± 4.87 2.24 ± 3.00 −0.01 ± 1.37
Maniquenique −2.15 ± 3.48 −0.70 ± 2.39 −0.75 ± 0.81
Macie −1.11 ± 3.84 −0.56 ± 3.12 −0.98 ± 1.21
Xai-Xai −0.15 ± 3.89 1.36 ± 2.72 −1.37 ± 1.30
Manhiça 15.86 ± 8.46 9.38 ± 5.65 1.71 ± 1.94
Map/Mavalane 2.50 ± 4.33 2.50 ± 3.24 −0.35 ± 0.69
Map/Observatório −2.17 ± 4.01 0.19 ± 3.27 −0.19 ± 0.78
Umbeluzi 0.59 ± 3.49 1.62 ± 2.72 −0.73 ± 0.59
Changalane −2.01 ± 3.24 −1.74 ± 2.67 −0.50 ± 0.57
The trend slope or the change rate a is given in millimeter per year (mm/yr), evaluated for the deseasonalized
time series: (i) the total annual precipitation, the total summer (NDJF) precipitation, and the total winter (MJJA)
precipitation. Confidence intervals were computed under 90% confidence level, assuming the t-distribution
of the estimated slope. Evaluated data consists of in situ observations during the period of 1960–2020

sistent with the H exponent and DFA α ranging from 0.54 southern Manica, where a total increase of about 15% is
to 1.26, what implies that the probability of a random estimated in annual precipitation, during the same period.
occurrence of precipitation is minimal. Besides, the north- These findings are in line with the IPCC-AR5 and AR6
eastern part of Mozambique had presented some isolated projections over the southeast Africa region. In locals where
anti-persistent areas tending to white noise, especially over precipitation exhibits persistency, it is very likely that the
Nampula and Cabo-Delgado provinces, and over these spe- observed trends prevail for a long period in future. This
cific areas, we can say that precipitation occurs randomly; insight is useful for planning and decision-making in the
past precipitation states have no influence on the future agricultural sector, as well as for setting climate change
events. adaptation plans of contingency in Mozambique and for sup-
On the other hand, trend in precipitation had presented porting seasonal weather forecasting services.
great spatial variability. It is decreasing in some regions, with The present work did not investigate contributing local
more significance over Zambezia and Nampula provinces, factors to the observed memory and trend at different areas;
where annual precipitation is estimated to have dropped thus, further studies aiming to explore the physical processes
about -12% from 1981 to 2021, and increasing in other or possible local drivers that may have conditioned such
regions, with more significance over Inhambane and results are necessary.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

Table 5 Trend in precipitation by the Mann-Kendall test

Station name Annual precipitation Summer precipitation Winter precipitation
s Slope (mm/yr) Z score s Slope (mm/yr) Z score s Slope (mm/yr) Z score

Mocimboa -62 −1.03 −0.54 40 0.40 0.35 -30 −0.16 −0.25

Pemba 118 1.84 0.92 273 4.09 2.21 -95 −0.31 −0.74
Montepuez -58 −1.59 −0.51 -16 −0.32 −0.13 37 0.05 0.32
Lichinga 44 0.79 0.34 157 2.17 1.27 -107 −0.14 −0.84
Cuamba -161 −2.32 −1.47 107 1.91 0.97 -223 −0.14 −1.97
Nampula 340 4.79 2.75 277 5.22 2.24 -17 −0.03 −0.13
Tete 150 2.28 1.18 183 2.13 1.48 31 0.02 0.24
Angoche -146 −3.49 −1.25 -122 −2.78 −1.04 -242 −1.01 −2.08
Quelimane -160 −4.51 −1.25 -19 −0.33 −0.15 -334 −1.63 −2.63
Chimoio -161 −3.59 −1.30 -111 −2.88 −0.89 -303 −0.71 −2.45
Sussundenga 80 11.24 1.08 74 7.43 0.99 28 0.16 0.37
Beira/Aeroporto -25 −0.80 −0.20 66 1.88 0.56 -125 −1.03 −1.04
Vilânculos -25 −0.33 −0.16 -121 −1.67 −0.80 -54 −0.14 −0.36
Inhambane 8 0.09 0.04 -36 −0.21 −0.22 -328 −1.00 −2.03
Massingir -251 −5.98 −2.62 -108 −1.71 −1.08 -335 −0.84 −3.50
Panda 23 0.34 0.15 42 0.51 0.28 -131 −0.40 −0.87
Inharrime -179 −8.14 −1.63 -177 −3.62 −1.61 -269 −2.24 −2.38
Chókwe -55 −1.44 −0.51 -98 −1.82 −0.95 -9 −0.04 −0.08
Manjacaze 90 3.31 0.90 126 2.50 1.26 -8 −0.07 −0.07
Maniquenique -126 −1.94 −0.88 -70 −0.79 −0.49 -272 −0.86 −1.87
Macie -68 −0.88 −0.47 42 0.75 0.29 -232 −0.86 −1.63
Xai-Xai -6 −0.08 −0.03 92 1.00 0.58 -292 −1.20 −1.81
Manhiça 223 11.21 2.32 193 7.34 2.01 96 0.84 0.96
Map/Mavalane 82 1.41 0.56 208 3.23 1.42 -47 −0.14 −0.32
Map/Observatório -118 −2.34 −0.92 49 0.77 0.39 -7 −0.02 −0.05
Umbeluzi -18 −0.37 −0.12 144 1.51 1.01 -246 −0.54 −1.69
Changalane -177 −2.30 −1.28 -149 −1.67 −1.07 -243 −0.46 −1.76
The parameter s, the Z score for statistical significance, and the trend slope or change rate given in mm/yr were evaluated for time series consisting
of (i) total annual precipitation, (ii) total precipitation during summer months (NDJF), and (iii) total precipitation during winter months (MJJA).
Evaluated data consists of in situ observations during the period of 1960–2020

Table 6 Overall long memory

Province Annual precipitation Summer precipitation Winter precipitation
and trend in precipitation for
Long memory Trend Long memory Trend Long memory Trend
each province of Mozambique
Maputo Pc ➙ Pc ➙ Pa to WN ➘b
Gaza Pc ➙ Pb ➙ Pb to WN ➘c
Inhambane Pc ➚b Pc ➚b Pb ➘b
Sofala Pc ➙ Pb to WN ➚a Pb to WN ➘c
Manica Pc ➚b Pa to WN ➚b Pa to WN ➘a
Tete WN ➙ WN ➚a Pc ➘b
Zambezia Pb ➘b Pc ➘b Pa to WN ➘c
Nampula Pa to WN ➘b WN ➘b WN ➘c
Niassa Pa to WN ➘a Pa to WN ➘a Pa to WN ➙
Cabo-Delgado APb ➙ APa to WN ➙ Pa to WN ➘a
Results are derived from CHIRPS data. Meaning of abbreviations and symbols: P, persistent; AP, anti-
persistent; WN, white noise; ➘, decrease; ➚, increase; ➙, no trend
a confidence level (30–60%), b confidence level (60–90%), c confidence level (>90%)

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 J.L.M. Ussalu and A. Bassrei

Acknowledgements The authors would like to thank the National Insti- Funk C, Peterson P, Landsfeld M, Pedreros D, Verdin J, Shukla S,
tute of Meteorology (INAM) and the Climate Hazards group InfraRed Husak G, Rowland J, Harrison L, Hoell A, Michaelsen J (2015) The
Precipitation with Stations (CHIRPS) for providing the climate data climate hazards infrared precipitation with stations - a new environ-
used in this study. A special thanks to Mr. Guelso Mauro Manjate for mental record for monitoring extremes. Scientific Data 2:150066.
his kind facilitation on the acquisition process of INAM’s data. https://doi.org/10.1038/sdata.2015.66
Gan TY (1998) Hydroclimatic trends and possible climatic warming
Author Contributions JLMU proposed the objectives and most of the in the Canadian Prairies. Water Resour Res 34(11):3009–3015.
methodology, compiled the data, performed the data analysis, gener- https://doi.org/10.1029/98wr01265
ated the figures, and wrote and reviewed the manuscript. AB proposed Hartmann DL, Klein Tank AMG, Rusticucci M et al (2013) Observa-
additional methodology, commented the results, and reviewed the tions: atmosphere and surface supplementary material. In: Stocker
manuscript. TF, Qin D, Plattner G-K et al (eds) Climate Change 2013: The
Physical Science Basis. Contribution of Working Group I to
Availability of data and material In situ datasets analyzed during this the Fifth Assessment Report of the Intergovernmental Panel on
study can be obtained from the National Institute of Meteorology under Climate Change. Cambridge University Press, cap. 2SM. www.
specific request, and CHIRPS datasets are publicly available and can climatechange2013.org. Accessed 26 June 2021
freely be downloaded from the link https://data.chc.ucsb.edu/products/ Heumann C, Schomaker M, Shalabh (2016) Introduction to statistics
CHIRPS-2.0/. Code scripts and data generated from the study (gap filled and data analysis. Springer International Publishing, Control Num-
station data, figures, etc.) are available from the corresponding author ber: 2016955516, Switzerland, pp. 451. ISBN 978-3-319-46162-5.
on reasonable request. https://doi.org/10.1007/978-3-319-46162-5
Hirsch RM, Slack JR, Smith RA (1982) Techniques of trend analysis
for monthly water quality data. Water Resour Res 18(1):107–121.
Declarations
https://doi.org/10.1029/wr018i001p00107
Hurst H (1951) Long term storage capacity of reservoirs. Trans Am
Soc Civ Eng 116(1):770–799. https://doi.org/10.1061/TACEAT.
Ethics approval and consent to participate Not applicable. 0006518
IPCC (2014) Climate change 2014: synthesis report. In: Pachauri RK,
Consent for publication Both authors gave explicit consent for the pub- Meyer LA (eds) Contribution of working groups I, II and III to the
lication of this article. fifth assessment report of the intergovernmental panel on climate
change. IPCC, Geneva, Switzerland, pp 151. https://doi.org/10.
Conflict of interest The authors declare no competing interests. 1017/CBO9781107415324.008
IPCC (2021) Climate change 2021: the physical science basis. In:
[Masson-Delmotte V, Zhai P, Pirani A et al (eds) Contribution
of working group I to the sixth assessment report of the inter-
governmental panel on climate change. Cambridge University
References Press, Cambridge, UK, NY, USA, pp 2391. https://doi.org/10.
1017/9781009157896
Ambrosino C, Chandler RE, Todd MC (2011) Southern African Kantelhardt JW, Koscielny-Bunde E, Rego HHA, Havlin S, Bunde A
monthly rainfall variability: an analysis based on generalized (2001) Detecting long-range correlations with detrended fluctu-
linear models. J Clim 24:4600–4617. https://doi.org/10.1175/ ation analysis. Physica A 295:441–454. https://doi.org/10.1016/
2010JCLI3924.1 s0378-4371(01)00144-3
Araneda-Cabrera RJ, Bermudez M, Puertas J (2021) Revealing the Kendall MG (1975) Rank correlation methods. 4th edn. Charles Griffin,
spatio-temporal characteristics of drought in Mozambique and pp 212. ISBN: 0852641990 9780852641996
their relationship with large-scale climate variability. J Hydrol: Lim Kam Sian KTC, Wang J, Ayugi BO, Nooni IK, Ongoma V
Regional Studies 38:100938. https://doi.org/10.1016/j.ejrh.2021. (2021) Multi-decadal variability and future changes in precipi-
100938 tation over southern Africa. Atmosphere 12:742. https://doi.org/
Arias PA, Bellouin N, Coppola E et al (2021) Technical summary. In: 10.3390/atmos12060742
Masson-Delmotte V, Zhai P, Pirani A et al (eds) Climate change Machaieie HA, Silva CG, de Oliveira EN, Júnior HIT, de Almeida HA
2021: the physical science basis. Contribution of working group I (2020) Variability and trends of precipitation in Quelimane, Cen-
to the Sixth Assessment Report of the Intergovernmental Panel on tral Mozambique, and their relation to El Niño southern oscillation.
Climate Change. Cambridge University Press, pp 33–144. https:// Journal of Geoscience and Environment Protection 8:1–16. https://
doi.org/10.1017/9781009157896.002 doi.org/10.4236/gep.2020.87001
Barimalala R, Blamey RC, Desbiolles F, Reason CJC (2020) Variabil- Mandelbrot B (1967) How long is the coast of Britain? Statistical self-
ity in the Mozambique Channel trough and impacts on southeast similarity and fractional dimension. Science 156(3775):636–638.
African rainfall. J Climate 33:749–765. https://doi.org/10.1175/ https://www.jstor.org/stable/1721427. Accessed 01 May 2021
JCLI-D-19-0267.1 Mandelbrot B, Wallis J (1969) Computer experiments with fractional
Beran J (1994) Statistics for long-memory processes. 1st edn. Rout- Gaussian noises. Part 2, Water Resour Res 5(1):242–259. https://
ledge, New York, pp 315. eBook ISBN: 9780203738481. https:// doi.org/10.1029/WR005i001p00242
doi.org/10.1201/9780203738481 Mann HB (1945) Nonparametric tests against trend. Econometrica
Caballero R, Jewson S, Brix A (2002) Long memory in surface 13(3):245–259. https://doi.org/10.2307/1907187
air temperature: detection, modeling, and application to weather Manuel L, Tostão E, Vilanculos O, Mandlhate G, Hartley F (2020)
derivative valuation. Clim Res 21(2):127–140. https://doi.org/10. Economic implications of climate change in Mozambique. SA-
3354/cr021127 TIED Working Paper 136, p 21. https://ceppag.uem.mz/images/
Fu G, Charles SP, Yu J, Liu C (2009) Decadal climatic variability, trends, pdf/SA_TIED_WP_136.pdf. Accessed 13 April 2023
and future scenarios for the North China Plain. J Climate 22:2111– Mavume A, Rydberg L, Rouault M, Lutjeharms J (2009) Climatology
2123. https://doi.org/10.1175/2008JCLI2605.1 and landfall of tropical cyclones in the south-west Indian Ocean.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Long memory and trend in time series of precipitation in Mozambique

Western Indian Ocean J. Mar. Sci. 8(1):15–36. https://doi.org/10. Sen PK (1968) Estimates of the regression coefficient based on
4314/wiojms.v8i1.56672 Kendall’s tau. J Am Stat Assoc 63(324):1379–1389. https://doi.
Mavume AF, Banze BE, Macie OA, Queface AJ (2021) Analysis of org/10.1080/01621459.1968.10480934
climate change projections for Mozambique under the represen- Seneviratne SI, Zhang X, Adnan M, Badi W, Dereczynski C, Di Luca
tative concentration pathways. Atmosphere 12(5):588. https://doi. A et al (2021) Weather and climate extreme events in a chang-
org/10.3390/atmos12050588 ing climate. In: Masson-Delmotte V, Zhai P, Pirani A et al (eds)
McSweeney C, New M, Lizcano G, Lu X (2010) The UNDP climate Climate Change 2021: The Physical Science Basis. Contribution
change country profiles improving the accessibility of observed of Working Group I to the Sixth Assessment Report of the Inter-
and projected climate information for studies of climate change governmental Panel on Climate Change. Cambridge University
in developing countries. Bull Am Meteorol Soc 91(2):157–166. Press, Cambridge, United Kingdom and New York, NY, USA, pp
https://doi.org/10.1175/2009BAMS2826.1 1513–1766. https://doi.org/10.1017/9781009157896.013
Meals DW, Spooner J, Dressing SA, Harcum JB (2011) Statisti- Taylor CH, Loftis JC (1989) Testing for trend in lake and ground water
cal analysis for monotonic trends. Tech Notes 6. U.S. Envi- quality time series. Water Resour Bull 25(4):715–726. https://doi.
ronmental Protection Agency by Tetra Tech, Inc., Fairfax, org/10.1111/j.1752-1688.1989.tb05385.x
VA, p 23. https://www.epa.gov/polluted-runoff-nonpoint-source- Ussalu JLM, Bassrei A (2021) Climate dynamics of southern region of
pollution/nonpoint-source-monitoringtechnical-notes. Accessed Mozambique: statistics and Fourier analysis. Braz J Climatology
20 March 2022 29:134–156. https://doi.org/10.5380/rbclima.v29i0.75088
Peng CK, Buldyrev SV, Havlin S, Simons M, Stanley HE, Goldberger Vera-Valdés JE (2021) Temperature anomalies, long memory,
AL (1994) Mosaic organization of DNA nucleotides. Phys Rev E and aggregation. Econometrics 9:9. https://doi.org/10.3390/
49(2):1685–1689. https://doi.org/10.1103/physreve.49.1685 econometrics9010009
Pomposi C, Funk C, Shukla S, Harrison L, Magadzire T (2018) Distin- Vyushin DI, Kushner PJ (2008) Power-law and long-memory char-
guishing southern Africa precipitation response by strength of El acteristics of the atmospheric general circulation. J Climate
Niño events and implications for decision-making. Environ. Res. 22:2890–2904. https://doi.org/10.1175/2008JCLI2528.1
Lett. 13:074015. https://doi.org/10.1088/1748-9326/aacc4c Xu ZX, Li JY, Liu CM (2007) Long-term trend analysis for major
Ponkina E, Illiger P, Krotova O, Bondarovich A (2021) Do ARMA climate variables in the Yellow River basin. Hydrol Process
models provide better gap filling in time series of soil temperature 21(14):1935–1948. https://doi.org/10.1002/hyp.6405
and soil moisture? The case of arable land in the Kulunda Steppe. Yu Y, Zou S, Whittemore D (1993) Non-parametric trend analysis of
Russia. Land 10(6):579. https://doi.org/10.3390/land10060579 water quality data of rivers in Kansas. J Hydrol 150(1):61–80.
Qiu M, Yuan N, Yuan S (2020) Understanding long-term memory in https://doi.org/10.1016/0022-1694(93)90156-4
global mean temperature: an attribution study based on model sim- Yuan N, Fu Z, Liu S (2014) Extracting climate memory using fractional
ulations. Atmos Ocean Sci Letters 13(5):485–492. https://doi.org/ integrated statistical model: a new perspective on climate pre-
10.1080/16742834.2020.1778418 diction. Scientific Reports 4(6577):1–10. https://doi.org/10.1038/
Queiroz A, Manjate T, Maure G, Mosquito DP (2007) Avaliação da srep06577
vulnerabilidade dos parâmetros climáticos e projeção de cenários
climúticos. Maputo - Moçambique, Technical report, p 39
Reason C, Jagadheesha D (2005) A model investigation of recent ENSO Publisher’s Note Springer Nature remains neutral with regard to juris-
impacts over Southern Africa. Meteorog Atmos Phys 89:181–205. dictional claims in published maps and institutional affiliations.
https://doi.org/10.1007/s00703-005-0128-9
Salite D, Poskitt S (2019) Managing the impacts of drought: the role Springer Nature or its licensor (e.g. a society or other partner) holds
of cultural beliefs in small-scale farmers’ responses to drought exclusive rights to this article under a publishing agreement with the
in Gaza Province, southern Mozambique. International Journal of author(s) or other rightsholder(s); author self-archiving of the accepted
Disaster Risk Reduction 41. https://doi.org/10.1016/j.ijdrr.2019. manuscript version of this article is solely governed by the terms of such
101298 publishing agreement and applicable law.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

[email protected]