water

Article
The Use of GAMLSS Framework for a Non-Stationary
Frequency Analysis of Annual Runoff Data over a
Mediterranean Area
Pietro Scala * , Giuseppe Cipolla, Dario Treppiedi and Leonardo Valerio Noto

Dipartimento di Ingegneria, Università degli Studi di Palermo, 90128 Palermo, Italy

* Correspondence: pietro.scala@unipa.it

Abstract: Climate change affects all the components of the hydrological cycle. Starting from precipi-
tation distribution, climate alterations have direct effects on both surface water and groundwater in
terms of their quantity and quality. These effects lead to modifications in water availability for agricul-
ture, ecology and other social uses. Change in rainfall patterns also affects the runoff of natural rivers.
For this reason, studying runoff data according to classical hydrological approaches, i.e., statistical
inference methods that exploit stationary probability distributions, might result in missing important
information relevant to climate change. From this point of view, a new approach has to be found
in the study of this type of data that allows for non-stationary analysis. In this study, the statistical
framework known as Generalized Additive Models for Location, Scale and Shape (GAMLSS), which
can be used to carry out non-stationary statistical analyses, was applied in a non-stationary frequency
analysis of runoff data collected by four gauges widely distributed across Sicily (Italy) in the period
1916–1998. A classical stationary frequency analysis of these runoff data was followed by a different
non-stationary frequency analysis; while the first was made using annual rainfall as a covariate, with
the aim of understanding how certain statistical parameters of runoff distribution vary with changes
Citation: Scala, P.; Cipolla, G.; in rainfall, the second derived information about the temporal variability of runoff frequencies by
Treppiedi, D.; Noto, L.V. The Use of considering time as a covariate. A comparison between stationary and non-stationary approaches
GAMLSS Framework for a was carried out using the Akaike information criterion as a performance metric. After analyzing
Non-Stationary Frequency Analysis four different probability distributions, the non-stationary model with annual rainfall as a covariate
of Annual Runoff Data over a was found to be the best among all those examined, and the three-parameter lognormal the most
Mediterranean Area. Water 2022, 14, frequently preferred distribution.
2848. https://doi.org/10.3390/
w14182848
Keywords: non-stationarity; GAMLSS; runoff; frequency analysis; rainfall–runoff model
Academic Editor: Gwo-Fong Lin

Received: 4 August 2022

Accepted: 8 September 2022
1. Introduction
Published: 13 September 2022
Runoff frequency analysis is fundamental to the development of methods for the
Publisher’s Note: MDPI stays neutral
statistical estimation of quantiles based on observed data. Such methods are widely used
with regard to jurisdictional claims in
in the development of specific rules concerning water resource management and in the
published maps and institutional affil-
generation of synthetic runoff time series for ungauged basins [1].
iations.
The scientific literature does not contain many studies on probability distributions for
annual runoff. One of the first approaches has been provided by Markovic [2], who investi-
gated the goodness of fit of probability distributions of annual runoff in Western USA and
Copyright: © 2022 by the authors.
southwestern Canada using the chi-squared statistic. The results of the work showed that
Licensee MDPI, Basel, Switzerland. all five probability functions taken into account (normal, two-parameter lognormal, three-
This article is an open access article parameter lognormal, two-parameter gamma and three-parameter gamma) were applicable
distributed under the terms and and that no distribution could be preferred to the others in fitting an observed single-station
conditions of the Creative Commons sample of annual river flow distributions. In particular, the differences in goodness of fit
Attribution (CC BY) license (https:// in ensemble analyses between the two-parameter lognormal and two-parameter gamma
creativecommons.org/licenses/by/ distributions can be considered, for practical purposes, negligible. Therefore, in the analysis
4.0/). of larger-scale distributions, these two functions are interchangeable.

Water 2022, 14, 2848. https://doi.org/10.3390/w14182848 https://www.mdpi.com/journal/water

Water 2022, 14, 2848 2 of 17

Later, Vogel et al. [3] investigated probability distributions for annual maximum,
mean and minimum streamflows at more than 1455 river basins in the USA, with record
lengths ranging from 6 to 115 years and with an average of 45.5 years of records per site,
using L-moment diagrams to measure the goodness of fit between the sample data and
the selected probability distributions. The authors highlighted that, among all the two-
parameter distributions taken into account (gamma and two-parameter lognormal), the
gamma (GAM) distribution was the one that captured the observed relationships between
the L-moments L-CV and L-Skew of average annual flows in the United States. However,
the results also showed that the three-parameter Pearson (P3), three-parameter lognormal
(LN3) and three-parameter log Pearson (LP3) distributions provided better approximations
to the observed L-moment relationships for average annual flows than any two-parameter
distribution considered. Given the theoretical justification provided for the gamma and P3
distributions, the authors recommended the use of either of these likelihood functions to
model average annual flows in the USA.
In contrast, Cannarozzo et al. [1] performed a frequency analysis of annual runoff data
recorded in Sicily (Italy) by identifying homogeneous regions and fitting, for each region,
a single probability distribution function to the annual runoff data, scaled by the index
runoff (the mean annual runoff). The authors used the chi-square test for goodness-of-fit
testing. The advantage of using this procedure was that both the frequency growth curve
and the runoff index could be estimated using morphological and climatic characteristics
of the watersheds easily identified in a GIS environment, such as average annual rainfall,
average elevation and average slope of the watershed.
Due to changes in climate and basin characteristics, the statistics of annual runoff
series show complex and non-stationary changes. For this reason, the assumption that
the distributions in a frequency analysis of hydrological variables will be in equilibrium
around an underlying mean and that the variance will remain constant over time can
be questioned.
The design of different hydraulic infrastructures and the management of water supply
systems, irrigation systems and hydropower are usually based on conventional frequency
analyses, which estimate the statistics for a time series of a certain hydrological variable by
assuming the stationarity of the recorded series, i.e., they are “devoid of trends, shifts or
periodicity (cyclicity)” [4]. Historically, in fact, statistical inference in hydrology has relied
heavily on this assumption, such that the distribution of the variable of interest has been
considered invariant with respect to time.
Generally, however, in a changing environment, combinations of multiple factors, such
as precipitation, temperature, evapotranspiration and, for example, reservoir construction,
can lead to variations in flow regimes by altering flow characteristics, i.e., the seasonality of
runoff and the frequency and magnitude of floods [5]. In fact, stream runoff has shown
significant changes globally due to the impact of climate change, mainly because of an-
thropogenic effects on climate and basin characteristics [6–8]. For this reason, methods
that account for non-stationarity have been developed in order to replace long-established
characteristic principles of estimation of distribution parameters and, consequently, water
resource management and to shift to an evolutionary paradigm. Such a paradigm must
recognize the dynamics of physical and socio-economic processes [9]. Several studies have,
therefore, introduced the concept of hydrological non-stationarity in the analysis of various
hydrological variables and, beyond this, have demonstrated that the stationary method is
no longer reliable [10–16].
Although, today, many still debate whether stationarity is immortal [17], alive [18]
or dead [19], it is well known that human activities and climate change have significant
impacts on runoff and other hydrological processes [8,20–24]. The current literature on
frequency analysis of non-stationary hydrological variables focuses mainly on two issues:
(i) the development of the non-stationary method and (ii) the exploration of covariates
that reflect changes in hydrological variables. Many studies [7,25,26] have presented the
Water 2022, 14, 2848 3 of 17

time-varying moment method, which assumes that the hydrological variable of interest
follows a certain type of probability distribution, whose moments change over time [27].
The choice of probability distribution is also of paramount importance. Frequency
analysis using distributions that poorly match the sample under examination can lead
to errors in the evaluation and estimation of hydrological variables. The basic idea is,
therefore, to assume that the type of distribution for the analyzed hydrological variable is
unchanging, while its statistical parameters may change over time or with other covariates.
In Villarini et al. [28], this method was presented using Generalized Additive Models for
Location, Scale and Shape parameters (GAMLSS; [29]) as a flexible framework for evalu-
ating non-stationary time series, using time as a covariate. The time-varying parameter
method can be extended to the analysis of physical covariates, such as precipitation, by
replacing time with any other physical time-dependent covariate [30–33]. The covariate
approach incorporates covariates into the parameters of distributions because the depen-
dence of model parameters on covariates is useful for representing the dependence of
hydrological time series on slowly varying climate forcing.
For example, Li et al. [34] conducted a non-stationary runoff frequency analysis for
future climate changes and studied the relevant uncertainties. The main purpose of this
study was to analyze the non-stationarity of runoff frequencies adjusted for future climate
change in the Luanhe River Basin, China. Non-stationary GAMLSS models were established
for the analysis of the non-stationary frequency of runoff (1961–2010), using observed
rainfall as a covariate, which is closely related to runoff and contributed significantly to
its non-stationarity. The results showed that the sources of uncertainty in the statistical
parameters of the non-stationary model arise mainly from fluctuations in the precipitation
sequence. This result indicates the need to consider the precipitation sequence as a covariate
in runoff frequency analysis in the future.
The objective of our study was, therefore, to investigate the non-stationarity of annual
runoff through a non-stationary frequency analysis for some Sicilian rivers, considering the
dependence of this variable on time and annual rainfall, which were here used as covariates.
The GAMLSS method was applied for the analysis of stationary and non-stationary runoff
frequencies. First, the stationary frequency analysis was performed, followed by the
non-stationary frequency analysis and a comparison of the two methodologies. The non-
stationary analysis was carried out by considering rain as a covariate, which, in turn,
showed non-stationary characteristics. Secondly, through a non-stationary analysis with
time used as a covariate, information on the temporal variability of the runoff distribution
parameters was derived.
Through these approaches, different probability distributions commonly used in
similar cases were taken into account in trying to identify the one that best fitted the
relevant dataset. The annual runoff data studied were provided by four gauges managed
by Autorità di Bacino of the Regione Siciliana (AdB).
The paper is structured as follows: following this introduction is a section on materials
and methods; the third section describes and discusses the achieved results; and, finally,
the last section presents the conclusions of the study.

2. Materials and Methods

This section presents the data used in this study along with the adopted methodology.
The GAMLSS framework is described, and the adopted statistical distributions are also
presented. In addition, the Akaike information criterion (AIC), used for testing the goodness
of fit of the various statistical distributions and models, is described. For the sake of clarity,
Figure 1 displays a flow chart of the adopted methodology.

2.1. Data
The data used in this study were provided by the Autorità di Bacino of the Regione Sicil-
iana. The gauge stations analyzed were those at the outlets of the watersheds “Belice river
at Sparacia” (hereinafter named BE-SPA), “Imera river at Drasi” (IM-DRA), “San Leonardo
Water 2022, 14, 2848 4 of 17

river at Monumentale” (SL-MON) and “Valle dell’Acqua river at Serena” (VA-SER). For
each station, annual rainfall and runoff time series (averaged over the watershed) were
collected. These stations were chosen among different gauges used by Cannarozzo et al. [1]
because they had the largest sample sizes. In addition, these gauge stations were selected
because their runoff time series exhibited different behaviors. The use of various statistical
tests highlighted the presence of a trend in BE-SPA, SL-MON and VA-SER annual runoff
Water 2022, 14, x FOR PEER REVIEWtime series, while heteroscedastic behavior for annual runoff vs. time was found for 4 ofIM-
18
DRA and VA-SER. This suggested the use of a non-stationary statistical approach to deal
with the annual runoff time series of these gauge stations.

Figure 1.
Figure Flow chart
1. Flow chart of
of the
the employed
employed methodology.
methodology.

The working periods of three measurement sites began in the late 1950s or early 1960s
2.1. Data
while that for SL-MON, which had the largest sample size, began in the late 1920s.
The data usedthe
In particular, in this study
sample were
sizes provided
were equal toby33,the34,
Autorità
53 anddi35Bacino
values of for
the BE-SPA,
Regione
Siciliana. The gauge stations analyzed were those at the outlets of the watersheds
IM-DRA, SL-MON and VA-SER, respectively. The geographical locations of the considered “Belice
Water 2022, 14, x FOR PEER REVIEW 5 of 18
river at Sparacia”
stations, located at(hereinafter
the outlets named BE-SPA),
of the related “Imera river
catchments, are at Drasi”in(IM-DRA),
shown “San
Figure 2, while
Leonardo river at
Figure 3 shows Monumentale”
scatterplots (SL-MON)
of runoff and “Valle
(q) vs. rainfall dell’Acqua
(p) and vs. time river
(t) forateach
Serena” (VA-
gauge.
SER). For each station, annual rainfall and runoff time series (averaged over the water-
shed) were collected. These stations were chosen among different gauges used by Canna-
rozzo et al. [1] because they had the largest sample sizes. In addition, these gauge stations
were selected because their runoff time series exhibited different behaviors. The use of
various statistical tests highlighted the presence of a trend in BE-SPA, SL-MON and VA-
SER annual runoff time series, while heteroscedastic behavior for annual runoff vs. time
was found for IM-DRA and VA-SER. This suggested the use of a non-stationary statistical
approach to deal with the annual runoff time series of these gauge stations.
The working periods of three measurement sites began in the late 1950s or early 1960s
while that for SL-MON, which had the largest sample size, began in the late 1920s.
In particular, the sample sizes were equal to 33, 34, 53 and 35 values for BE-SPA, IM-
DRA, SL-MON and VA-SER, respectively. The geographical locations of the considered
stations, located at the outlets of the related catchments, are shown in Figure 2, while Fig-
ure 3 shows scatterplots of runoff (q) vs. rainfall (p) and vs. time (t) for each gauge.

Figure Locations
2. 2.
Figure Locationsof of
thethe
gauges considered
gauges in this
considered study
in this and and
study the related catchments
the related overlaid
catchments on theon
overlaid
the perimeter
perimeter of Sicily.
of Sicily.
Water 2022, 14, 2848 5 of 17
Figure 2. Locations of the gauges considered in this study and the related catchments overlaid on
the perimeter of Sicily.

Figure3.3.Scatterplots
Figure Scatterplotsofof annual
annual runoff
runoff vs. vs. rainfall
rainfall (first
(first row)row) and annual
and annual runoffrunoff vs.(second
vs. time time (second
row)
row) for IM-DRA, BE-SPA, SL-MON
for IM-DRA, BE-SPA, SL-MON and VA-SER. and VA-SER.

2.2.
2.2.The
TheGeneralized
GeneralizedAdditive
AdditiveModels
ModelsininLocation,
Location,Scale
Scaleand
andShape
Shape(GAMLSS)
(GAMLSS)Framework
Framework
In
Inthis
this study, generalclass
study, a general classofofregression
regressionmodels,models, widely
widely known
known as Generalized
as Generalized Ad-
Additive Models in Location, Scale and Shape (GAMLSS), was adopted
ditive Models in Location, Scale and Shape (GAMLSS), was adopted to carry out the sta- to carry out the
stationary
tionary and and non-stationary
non-stationary runoff
runoff frequency
frequency analyses.
analyses.
In
InGAMLSS,
GAMLSS,the theexponential
exponentialfamily
family distribution
distribution assumption
assumption forfor
thethe
response
responsevariable
varia-
yble
is relaxed and replaced
y is relaxed by a general
and replaced distribution
by a general family, family,
distribution including highly skewed
including highly and/or
skewed
kurtotic distributions.
and/or kurtotic The systematic
distributions. part of the
The systematic part model
of theismodel
extended to allow modeling
is extended not
to allow mod-
only of the mean but also of the other parameters of the distribution. This
eling not only of the mean but also of the other parameters of the distribution. This can be can be carried
out by means
carried of linear
out by meansparametric and/or additive
of linear parametric and/or non-parametric functions of explanatory
additive non-parametric functions of
variables and/or random effects. Maximum likelihood estimation is used to fit the models.
GAMLSS can be defined as semi-parametric regression models. These models are
“parametric”, since they require parametric distribution assumptions for the response vari-
ables, and “semi-” in the sense that modeling of the distribution parameters, as functions
of the explanatory variables, may involve non-parametric smoothing functions.
A GAMLSS model assumes that a certain number of independent observations, yi ,
for i = 1, 2, . . . , n, are distributed according to a probability density function, f(yi |θi ),
conditional on θi = (θ1i , θ2i , θ3i , θ4i ) = (µi , σi , νi , τi ), which represents the ensemble of four
distribution parameters, each of which can be a function of the explanatory variables.
For this reason, hereafter, we refer to (µi, σi , νi , τi) as the distribution parameters. The
first two of them, µi and σi , are usually mentioned as location and scale parameters, while
the remaining parameter(s), if any, are characterized as shape parameters, e.g., skewness
and kurtosis parameters.
In any case, the regression model may be applied more generally to the parameters
of any population distribution and can be generalized to more than four distribution
parameters. Rigby and Stasinopoulos [29] introduced the original formulation of a GAMLSS
model. One can consider y0 = (y1 , y2 , . . . , yn ) the n-length vector of the response variable
and let gk (·) (for k = 1, 2, 3, 4) be the monotonic functions linking the distribution parameters
to the explanatory variables:
J
g1 (µ) = η1 = X1 β1 + ∑j=
1


h x
1 j1 j1
(1)
Water 2022, 14, 2848 6 of 17

J
g2 (σ) = η2 = X2 β2 + ∑j=
2


h x
1 j2 j2
(2)
J
g3 (ν) = η3 = X3 β3 + ∑j=
3


h x
1 j3 j3
(3)
J
g4 (τ) = η4 = X4 β4 + ∑j=
4


h x
1 j4 j4
(4)

where µ, σ, ν, τ and ηk and xjk , (for j = 1, 2, . . . , Jk ) and k = 1, 2, 3, 4 are vectors of length

n. In many practical situations, four distribution parameters are required at most. The
function hjk is a non-parametric additive function of the explanatory variable Xjk evaluated
at xjk . The explanatory vectors xjk are assumed to be fixed and known. In addition, Xk ,
for k = 1, 2, 3, 4, are fixed design matrices (fixed effects design matrices of explanatory
variables, i.e., covariates) while βk are the vectors of the parameters of the distribution.
Usually, in typical applications, a constant or other simple model is often suitable for each
of the two shape parameters (ν and τ).
The above model is called the semi-parametric GAMLSS model and has been extended
to allow random effect terms to be included in the model for µ, σ, ν and τ [29].
The parametric vectors βk and k = 1, 2, 3, 4 are estimated within the GAMLSS frame-
work by maximizing a penalized likelihood function.
The GAMLSS framework implemented in the R package (used for the presented
analyses) allows for the fast fitting of different models (i.e., linear or non-linear dependence
of means with respect to covariates) to a dataset. The selection of the model that best suits
the data can be performed by checking the significance of the fitting improvement, e.g.,
between stationary and non-stationary models, by means of deviance statistics.

2.3. Statistical Distributions in GAMLSS

According to the properties of the analyzed dataset, the theoretical distribution func-
tions that best fitted the observed annual rainfall and runoff data had to reflect the following
characteristics: (i) the function had to be continuous and defined for positive values of the
variable, (ii) the lower tail had to be bound by zero or a positive value, (iii) the upper tail
had to be unbounded, (iv) the density curve had to be asymptotic to the axis for large values
of the analyzed variable, (v) the basic shape had to be a two-tailed bell curve with a wide
variety of skewness, and, lastly, (vi) the number of parameters describing the theoretical
functions had to be limited to three.
The choice of various probability distributions was based on the analysis of the
types of best-fit probability functions for annual precipitation and runoff distributions
reported by Markovic [2], in which the following distribution types were studied: normal
distribution (NO), two parameter log-normal distribution (LOGNO), three-parameter log-
normal distribution (LNO) and two-parameter gamma distribution (GA) [3]. The LNO
distribution, as reported by Stasinopoulos et al. [35], is a two-parameter fit for µ and σ,
while ν is fixed.
The equations for the chosen distributions with the related moments are shown below,
in Table 1.

2.4. The AIC Criterion

The evaluation of the goodness of fit for the above-mentioned distributions to each
runoff dataset was carried out using the Akaike information criterion [36], here referred
to as AIC, which is a method for evaluating and comparing statistical models. It provides
a measure of the estimation quality of a statistical model by taking into account both the
goodness of fit and the complexity of the model. AIC is calculated from:
1. The number of independent variables used to build the model;
2. The maximum likelihood estimates of the model.
In the general case, the AIC is defined as:

AIC = 2k − 2 ln(L) (5)

Water 2022, 14, 2848 7 of 17

where k is the number of parameters in the model and L is the maximum value of the likeli-
hood function for the model. The default k is 2, so a model with one parameter will have a
k of 2 + 1 = 3. In general, the model that best fits the data is the one with the lowest AIC.
If a model is more than 2 AIC units lower than another, then it is considered significantly
better than that model. In the analysis of time series, it is common to try some kind of
transformation on the variable [37]. The decision about the choice of the transformation
can be simply realized by using the likelihoods of the models. The effect of transforming
the variable is represented by the product of the likelihood and the corresponding Jacobian,
and thus by the addition of minus twice the logarithm of the Jacobian to the AIC. In the
case of log {y(n)}, it is 2 ∑log {y(n)}, where the summation extends over n = 1, 2, . . . , N
and N is the length of the data. The correct AICs are obtained after these corrections for
the Jacobians.

Table 1. Probability density functions with related distribution moments.

Distributions Probability Density Function Distribution Moments

1 x−µ 2 E[x] = µ ;
Normal (NO) f(x) = √1 e− 2 ( σ )
σ 2π Var[x] = σ2 ;
x E[x] = µ ;
Gamma (GA) f(x) = 1
xµ−1 e− σ
σµ Γ ( µ ) Var[x] = σ2 µ2
1

−(ln (x)− µ)2

E[x] = ϕ 2 eµ ;
Log-normal 2 parameters (LOGNO) f(x) = e √ 2σ2 Var[x] = ϕ (ϕ − 1) e2 µ ;
x 2π σ 2
where ϕ = eσ
ν2
σ+
E[x] = µ + e 2 ;
2 2 2
Log-normal 3 parameters (LNO) f(x) = √ 1
e − 12 ( log (x−νµ)−σ ) Var[x] = eν −1 e2 σ + ν
;
2πν(x−µ)  2 p
3 ν 2
E [x] = e + 2 eν − 1;

3. Results and Discussion

This section presents the results obtained by applying GAMLSS to the rainfall–runoff
data collected at the examined stations. Section 3.1 is devoted to the stationary frequency
analysis, while the other two sections discuss the non-stationary frequency analysis consid-
ering rainfall (Section 3.2) and time (Section 3.3) as covariates. The non-stationary analysis
for both covariates was performed following three different combinations: (i) linear de-
pendence of µ with respect to the covariate under consideration and constant σ, (ii) linear
dependence of µ and σ with respect to the covariate under consideration and (iii) fitting of
a non-parametric smoothing model for µ and a linear σ dependence, again with respect to
the same covariates.

3.1. Stationary Analysis

Nelson [38] provided a statistical definition of stationarity, stating that it means that
the statistical parameters of a data series computed from different samples do not change
except due to sampling variations.
Shumway and Stoffer [39] distinguish between two orders of stationarity, making a
distinction between strictly stationary and weakly stationary time series. A time series
is said to be strictly stationary if its statistical properties do not vary over time. In other
words, if a time series has a constant mean and variance, it is a stationary time series. A
less strict type of stationarity, known as weak or second-order stationarity, is one in which
first- and second-order moments depend only on temporal differences [40]. Essentially, the
length that time series can be observed is limited, as in the case of hydrological time series.
Therefore, weakly stationary time series are practically treated as stationary time series. A
stationary time series cannot have any trend or periodic component. Conversely, if a series
does not have a constant mean or variance, it is not stationary.
Water 2022, 14, 2848 8 of 17

In order to study the classical approach to the stationary analysis of runoff data, the
above-mentioned distributions (Table 1) were applied and then compared.
For data from all the gauges considered, the log transformation provided better AIC
values. In particular, for IM-DRA, the best distribution was the LOGNO distribution, while
for three gauge sites the best distribution, in terms of AIC, was LNO. Where the behavior
of the runoff data was visibly far from that of a normal distribution, the optimal AIC value
greatly differed from the AIC of the NO distribution (i.e., IM-DRA). The distributions that
best fitted the samples under investigation are shown in Table 2, with the respective AIC
values highlighted in bold, and in Figure 4.

Table 2. AIC values for all the considered distributions of the stationary analysis. In bold are shown
the lowest AIC values among the analyzed distributions for each station.

AIC Values
Distributions BE-SPA IM-DRA SL-MON VA-SER
Normal (NO) 394.90 410.29 632.50 440.18
Gamma
Water 2022, 14, (GA)REVIEW
x FOR PEER 396.54 386.01 630.31 435.31 9 of 18
Log-normal 2 parameters (LOGNO) 402.17 380.99 635.66 441.47
Log-normal 3 parameters (LNO) 394.64 390.44 629.31 434.29

Figure4.4.Empirical
Figure Empiricaland
andtheoretical
theoreticalcumulative
cumulativedistribution
distributionfunctions
functions(cdfs)
(cdfs)and
andworm
wormplots
plotsfor
forthe
the
distributions with the lowest AIC values for all the stations.
distributions with the lowest AIC values for all the stations.

TableIn 2.
particular, the
AIC values forfirst row
all the in Figuredistributions
considered 4 representsofthe
the empirical (points) In
stationary analysis. and theoretical
bold are shown
(red
the line)
lowest cumulative
AIC valuesdistribution functions
among the analyzed (cdfs) for the
distributions adopted
for each distributions for the four
station.
gauge stations.
AIC Values
The second row shows the relatives worm plots of the best distributions provided
Distributionsby the AIC test. The worm BE-SPA plot [41] is aIM-DRA
diagnostic tool forSL-MON VA-SER
checking the residuals within
Normal (NO)different ranges of the 394.90 410.29 632.50 440.18
explanatory variables, with elliptical curves indicating approximate
Gamma (GA)95% point-wise confidence 396.54bands. Ideally, 386.01
the points in the worm630.31 435.31to the
plot should be close
Log-normal 2 parameters (LOGNO) 402.17 380.99 635.66 441.47
Log-normal 3 parameters (LNO) 394.64 390.44 629.31 434.29

3.2. Non-Stationary Analysis with Rainfall as a Covariate

 Water 2022, 14, 2848 9 of 17

horizontal line in the middle with no systematic shape and 95% or more of the points inside
the elliptical curve [42]. With worm plots, it is thus possible to visualize the differences
between different distributions, conditioned on the values of a covariate. The quadratic
and cubic shapes of the residuals in worm plots highlight that the empirical skewness and
kurtosis have not been appropriately captured by the chosen distribution model, even if the
models are characterized by low AIC values. Particularly for IM-DRA and SL-MON, not all
points fell within the confidence band. A cubic trend for the residuals can be observed for
BE-SPA, IM-DRA and VA-SER. For SL-MON, on the other hand, the trend of the residuals
is characterized by a quadratic shape.

3.2. Non-Stationary Analysis with Rainfall as a Covariate

In this section, the application of a non-stationary analysis is presented. In the first
step, a linear dependence of the location parameter µ with annual rainfall was imposed.
The scale parameter σ was constant (µ~p, σ~c, where “~c” means that the µ/σ parameter
was constant and “~p” means that the µ/σ parameter was modelled as a linear function of
annual rainfall). Hereinafter, this is referred to as the P1 model. In the parameter estimation
for the GAMLSS framework, the algorithm proposed by Rigby and Stasinopoulos [43] and
Rigby and Stasinopoulos [44] was used to calculate the parameters of the optimal model,
with the maximum penalized likelihood function as the objective function. Through this
analysis, it was possible to derive plots which showed trends for µ and σ with respect to
the covariate.
Water 2022, 14, x FOR PEER REVIEW For example, considering the P1 model, the distribution best fitting the IM-DRA data 10 of
was the LOGNO. In this case, imposing the linear variation of the µ parameter with rainfall
and keeping the value of σ constant (so that it did not depend on annual rainfall) resulted
in an increasing pattern of µ and a constant pattern for σ (left panel of Figure 5) with respect
median and the variability of the distribution with rainfall, as in the centiles plot show
to rainfall. These patterns of µ and σ resulted in nonlinear behavior of the median and
in the right panel
the variability of Figure 5.with
of the distribution There, the black
rainfall, linecentiles
as in the represents the 50th
plot shown percentile
in the right curve
(i.e., of
panel median),
Figure 5.the dark-gray
There, regions
the black represent
line represents thethe areas
50th within
percentile the 25–75%
curves centile curve
(i.e., median),
the light-gray regions represent the areas within the 5–95% centile curves
the dark-gray regions represent the areas within the 25–75% centile curves, the light-gray and the red do
in the represent
regions centiles plot represent
the areas withinthe
the observed runoff
5–95% centile data.
curves and the red dots in the centiles
plot represent the observed runoff data.

Figure 5. Variation of µ and σ parameters with annual rainfall for the IM-DRA LOGNO distribution
Figure 5. Variation of µ and σ parameters with annual rainfall for the IM-DRA LOGNO distributio
(left panel) and the centiles plot (right panel) for the same station, along with the plot for annual
(left panel) and the centiles plot (right panel) for the same station, along with the plot for annu
runoff vs. annual rainfall.
runoff vs. annual rainfall.

For the P1 model, the variation of the dependent variable (the runoff) was itself no
automatically linear, since this actually depended on the type of distribution. Looking
the moments’ equations related to the LOGNO distribution in Table 1, the linear variatio
Water 2022, 14, 2848 10 of 17

For the P1 model, the variation of the dependent variable (the runoff) was itself not
automatically linear, since this actually depended on the type of distribution. Looking at
the moments’ equations related to the LOGNO distribution in Table 1, the linear variation
of µ reflects an exponential growth of both the first two moments (mean and variance).
This, in turn, leads to a nonlinear increase in the location and shape of the distribution
and, hence, to a progressively nonlinear widening of the centile curves as annual rainfall
increases. Only in the case of the NO distribution will the trend of the median with rainfall
and the variability of the distribution be linear.
Water 2022, 14, x FOR PEER REVIEW For the P1 model, the best distributions, according to the AIC, were the LNO for 11 the
of 18
BE-SPA and VA-SER sites, the LOGNO for the IM-DRA site and the NO distribution for
the SL-MON gauge site. Goodness of fit for the models was assessed using centile curve
diagnostic plots and worm plots [41].
the AIC still
Figure show the
6 shows the residuals withand
centile curves quadratic and cubic
worm plots of theshapes in the wormfor
best distributions plots.
eachIn this
site.
In the first row of Figure 6, the centiles plot highlights how, year by year, the probability a
case, an improvement in the goodness of results, shown through the worm plots, and
decrease in the
distributions AIC values
adopted changeasascompared
a functiontoofthe
theS-model
rainfall make it clear
covariate. Thethat the modeling
variability of theof
the location parameter with an external covariate has a considerable impact
distribution, which does not always remain constant, changes according to the parameters on this type
ofofthe
non-stationary analysis.
distribution itself; in fact, the location parameter varies linearly with annual rainfall,
while, as mentioned above, σ is kept constant for each value of the covariate.

Figure6.6.Summary
Figure Summaryofofresults
resultsfor
forthe
theP1P1model
modelfor allall
for the stations
the with
stations generalized
with generalizedadditive models
additive models
ininlocation, scale and shape parameters and the corresponding worm plots
location, scale and shape parameters and the corresponding worm plots for runoff series.for runoff series. The
The
legend
legendforforthe
the(first line)
(first is is
line) the same
the same asasin in
Figure 5. 5.
Figure

Moreover, most of the

Since a nonlinear red points
pattern are within
was noticed thevariability
in the 5% and 95% curves
of the for the optimal
probability distribu-
models, indicating that the models describe the variability of the observed data reasonably
tions, which in addition to the µ parameter is certainly related to the σ parameter, a further
well. The percentage of the number of points outside the 5–95% centile band is always
analysis was carried out by introducing, in addition to the linear variation of µ with rain-
less than 10% of each station’s sample size and equal to 6%, 6%, 9% and 9% for BE-SPA,
fall, that of σ with annual rainfall as covariate (µ~p, σ~p—P2).
IM-DRA, SL-MON and VA-SER, respectively.
The introduction of a linear modeling of the σ parameter with rainfall improved the
The panels in the second row of Figure 6 show that the points of the optimal models
description of the runoff probability distributions for all four watersheds. In this case, the
are all within the 95% confidence intervals, indicating that the residuals are approximately
distributions with the lowest AIC values, which therefore minimized the likelihood func-
normally distributed. Just as with the stationary analysis, the distributions chosen through
tions, were the NO distribution for the BE-SPA site, the LOGNO distribution for the IM-
DRA site and the LNO distribution for the SL-MON and VA-SER sites.
Thus, the best distributions in terms of AIC values did not remain unchanged for BE-
SPA and SL-MON compared with the previous model. In Figure 7, it can be seen how the
linear modeling of σ leads to a change in the shape of the centile curves with respect to
Water 2022, 14, 2848 11 of 17

the AIC still show the residuals with quadratic and cubic shapes in the worm plots. In this
case, an improvement in the goodness of results, shown through the worm plots, and a
decrease in the AIC values as compared to the S-model make it clear that the modeling of
the location parameter with an external covariate has a considerable impact on this type of
non-stationary analysis.
Since a nonlinear pattern was noticed in the variability of the probability distributions,
which in addition to the µ parameter is certainly related to the σ parameter, a further
analysis was carried out by introducing, in addition to the linear variation of µ with rainfall,
that of σ with annual rainfall as covariate (µ~p, σ~p—P2).
The introduction of a linear modeling of the σ parameter with rainfall improved the
description of the runoff probability distributions for all four watersheds. In this case,
the distributions with the lowest AIC values, which therefore minimized the likelihood
functions, were the NO distribution for the BE-SPA site, the LOGNO distribution for the
IM-DRA site and the LNO distribution for the SL-MON and VA-SER sites.
Water 2022, 14, x FOR PEER REVIEW Thus, the best distributions in terms of AIC values did not remain unchanged 12 for
of 18
BE-SPA and SL-MON compared with the previous model. In Figure 7, it can be seen how
the linear modeling of σ leads to a change in the shape of the centile curves with respect to
the former analysis. In fact, this new modeling approach resulted in a thinning of the bands
in the In this case,
centiles plots, the P2 model
especially is more
the 75th to 95thsuitable thanbands
percentile the previous
suggestingone
that,forcompared
capturing
changes in the variability of distributions with rainfall for all gauge sites, and
with the previous model, the variability of the distributions for the highest centile values this is also
attested byEven
decreased. a decrease in AIC
in this case, thevalues for all
behavior the centile
of the four sites. Foragainst
curves all the gauge
annualsites, the(not
rainfall trend
of residuals within the 95% confidence interval of worm plots underwent
shown in the plot) was no longer perfectly linear, as reported in Zhang et al. [45]. The NO considerable
flattening (Figure
distribution 7), whichdistribution
(the best-fitting is diagnosticfor
ofthe
an improvement compared
BE-SPA site), for which thewith the previous
second-order
analysis.
moment was governed by the square σ, also showed a nonlinear pattern compared with
the previous model.

Figure7.7. Best
Figure Best suitable distribution centile
suitable distribution centilecurves
curves(first
(firstrow)
row)and andthe
the corresponding
corresponding worm
worm plots
plots (sec-
(second row)
ond row) forfor
thethe
P2P2 model.
model. The
The legend
legend forfor
thethe first
first line
line is is the
the sameasasininFigure
same Figure5.5.

As already reported, a decrease in AIC means an improvement in model perfor-

mance. This was more evident for the P1 model than in the case (not shown in this paper)
in which only σ was linearly modeled as a function of rainfall. For this reason, a more in-
depth modeling of the µ parameter was explored. In particular, a further analysis was
carried out with a higher degree of complexity, in which the location parameter, µ, was
considered as a non-parametric smoothing cubic-spline function of the covariate rainfall
Water 2022, 14, 2848 12 of 17

In this case, the P2 model is more suitable than the previous one for capturing changes
in the variability of distributions with rainfall for all gauge sites, and this is also attested by
a decrease in AIC values for all the four sites. For all the gauge sites, the trend of residuals
within the 95% confidence interval of worm plots underwent considerable flattening
(Figure 7), which is diagnostic of an improvement compared with the previous analysis.
As already reported, a decrease in AIC means an improvement in model performance.
This was more evident for the P1 model than in the case (not shown in this paper) in which
only σ was linearly modeled as a function of rainfall. For this reason, a more in-depth
modeling of the µ parameter was explored. In particular, a further analysis was carried
out with a higher degree of complexity, in which the location parameter, µ, was considered
as a non-parametric smoothing cubic-spline function of the covariate rainfall with three
effective degrees of freedom; the linear modeling of the parameter σ with the rainfall
Water 2022, 14, x FOR PEER REVIEW 13 of 18
covariate was maintained (µ~cs(p), σ~p—P3). “~cs(p)” means that the µ/σ parameter was
modelled as a cubic spline of annual rainfall.
At this stage, the best distributions remained unchanged when compared with the
previous
indeed, analysis,
reflectedthe
in aonly exception
worsening, being the
although VA-SER
not site. Centile curves
very pronounced, of AIC for all stations
values, summa-
and thein
rized related
Tableworm
3, for plots are shown
the different in Figure 8.under
distributions What consideration
is interesting toand
point
forout
eachis that
gauge
there
site. is a thinning of the 5–25% centile curve ranges, a widening of the 75–95% ones and a
lowering It isofimportant
the black to(median)
point out line, particularly
that the type offor the IM-DRA
distribution and
that VA-SERthe
provided sites.
bestTheAIC
25–75% bands, on the other hand, also tend to follow the points characterized
values was always the same under the three different models for the IM-DRA station. For by high
runoff values and
the BE-SPA in greater
SL-MON detail.
sites,Inthe
relation to thisdistribution
best-fitting model, thechanged
worm plotsonlyalways
from P1 show
to P2,
good
thendistributions of residuals,
remained unchanged with
in P3. all pointsthe
Generally, within the 95%
P2 model wasconfidence interval,
the one that, but
considering
also quadratic or cubic trend lines that are more evident than in the previous
both AIC values and worm plots, provided the best results. It is important to highlight analyses.
This
thatis,the
indeed,
use of reflected
observedinannual
a worsening,
rainfall although not very
as a covariate, pronounced,
which of AIC to
is closely related values,
runoff,
summarized in Table 3, for the different distributions under consideration
contributed significantly to the non-stationarity of the runoff distribution. and for each
gauge site.

Figure8.8.Best
Figure Best suitable distribution centile
suitable distribution centilecurves
curves(first
(firstrow)
row)and andthethecorresponding
corresponding worm
worm plots
plots (sec-
(second row)
ond row) forfor
thethe µ~cs(p),
µ~cs(p), σ~p
σ~p model.
model. TheThe legend
legend forfor the
the first
first line
line is is the
the sameasasininFigure
same Figure5.5.

Table 3. Comparison of the best stationary and non-stationary models with rainfall as covariate. In
bold are shown the distributions that provided the lowest AIC values between the analyzed distri-
butions for each station.

AIC Values
Water 2022, 14, 2848 13 of 17

Table 3. Comparison of the best stationary and non-stationary models with rainfall as covariate.
In bold are shown the distributions that provided the lowest AIC values between the analyzed
distributions for each station.

AIC Values
Models BE-SPA IM-DRA SL-MON VA-SER
LNO LOGNO LNO LNO
S—Stationary
394.64 380.79 629.31 434.29
LNO LOGNO NO LNO
P1—µ~p, σ~c
373.41 365.33 581.02 409.10
NO LOGNO LNO
P2—µ~p, σ~p LNO 407.57
373.36 361.12 580.46
NO LOGNO LNO LOGNO
P3—µ~cs(p), σ~p
374.66 362.94 584.36 405.71

It is important to point out that the type of distribution that provided the best AIC
values was always the same under the three different models for the IM-DRA station. For
the BE-SPA and SL-MON sites, the best-fitting distribution changed only from P1 to P2,
then remained unchanged in P3. Generally, the P2 model was the one that, considering
both AIC values and worm plots, provided the best results. It is important to highlight
that the use of observed annual rainfall as a covariate, which is closely related to runoff,
contributed significantly to the non-stationarity of the runoff distribution.

3.3. Non-Stationary Analysis with Time as a Covariate

A further GAMLSS analysis was carried out considering time as a covariate. Again,
three non-stationary analyses (see Table 3) were carried out in which (T1) the location
parameter µ was modeled as a linear function with respect to time and the scale parameter
σ was a constant (µ~t, σ~c), (T2) the location parameter µ and σ were modeled as linear
functions of time (µ~t, σ~t) and (T3) the parameter µ was modelled as a cubic spline of
time and σ as a linear function of time (µ~cs(t), σ~t). “~c” means that the µ/σ parameter
was constant, “~t” means that the µ/σ parameter was modelled as a linear function of time
and “~cs(t) means that the µ/σ parameter was modelled as a cubic spline of time.
Figure 9 shows the centile plots and worm plots for the models that provided the
lowest AIC values. These are the T1 models for the BE-SPA and SL-MON sites, considering
the LNO distribution in the two cases. In contrast, the T2 model was to be preferred for the
IM-DRA station, along with the LOGNO distribution. Regarding VA-SER, the T3 model
was the best, using the GA distribution.
While the variability of the distributions over time remained almost constant for the
BE-SPA and SL-MON sites, for VA-SER, the variability had a remarkable effect. In fact,
the lowest AIC value was provided by the T3 model with cubic spline, which is the one
that best captured the variability of this dataset. In this case, the trends of the curves in the
centile plot are characterized by smooth functions that widen for higher values of the runoff
variables. Just as with the IM-DRA values, centiles below 50% show a mainly decreasing
trend, while the opposite can be recognized for centiles above 50%. Due to the higher
runoff values, especially in the last years of the IM-DRA time series, the variability of the
distribution with reference to the 75–95% range increases.
For the IM-DRA and VA-SER stations, the non-stationary analyses with time as covari-
ate provided better AIC values than the stationary analyses.
For the other gauge sites, however, small differences in AIC values were obtained
compared to the stationary analysis, which showed slightly better performance. This was
unlike the case where rainfall was used as a covariate. In fact, there was no clearly marked
trend in the distribution parameters associated with time, and this led to a deterioration in
the predictive capabilities of the model compared with the cases which used precipitation
as a covariate.
 compared to the stationary analysis, which showed slightly better performance. This was
unlike the case where rainfall was used as a covariate. In fact, there was no clearly marked
trend in the distribution parameters associated with time, and this led to a deterioration
in the predictive capabilities of the model compared with the cases which used precipita-
Water 2022, 14, 2848 14 of 17
tion as a covariate.

Figure 9. Centiles plots (first row) of the best model for the four considered stations. In the (second
Figure 9. Centiles plots (first row) of the best model for the four considered stations. In the (second
row) the corresponding worm plots are displayed. The legend for the first line is the same as
row) the corresponding worm plots are displayed. The legend for the first line is the same as in
in Figure 5.
Figure 5.
Furthermore, in Figure 9, all points can be seen to be within the 95% confidence band in
the worm plots, but the trend lines show cubic/quadratic behavior, except for the SL-MON
and VA-SER sites. In any case, these trend lines are more pronounced than those obtained
using rainfall as covariate, indicating a worsening of the time-based models.
In Table 4, the AIC values for the three different types of non-stationary models are
compared with those for the stationary analysis. As one can see from the AIC values,
considering time as covariate did not improve the performance of these models with
respect to the S model; the non-stationary models had worse performances, except for the
VA-SER site.

Table 4. Comparison of the stationary and non-stationary best models with time as covariate. In bold
are shown those models which provided the lowest AIC values among the non-stationary models.

AIC Values
Models BE-SPA IM-DRA SL-MON VA-SER
LNO LOGNO LNO LNO
S—Stationary
394.64 380.79 629.31 434.29
LNO LOGNO LNO LNO
T1—µ~t, σ~c
395.48 382.98 630.21 434.96
LNO LOGNO LNO GA
T2—µ~t, σ~t
397.42 380.87 630.99 429.06
LNO LOGNO LNO GA
T3—µ~cs(t), σ~t
400.30 383.96 631.10 428.69
Water 2022, 14, 2848 15 of 17

4. Conclusions
One of the basic assumptions made in hydrological studies has been that the parame-
ters of the distributions of hydrological variables of interest are constant over time. The use
of a stationary probability distribution, therefore, may be ineffective when the examined
variable is characterized by change in the mean and/or in variability due to the presence of
hydrological change (change in land use or climatic change). In fact, the classical approach
to stationary distributions does not take into account the natural variability of certain
hydrological processes, such as runoff. It also does not consider changes that may occur in
related variables, such as precipitation and temperature, where changes related to climate
change may occur.
For these reasons, a GAMLSS model was applied to four Sicilian annual runoff time
series to compare classical stationary frequency analysis with different non-stationary
frequency analyses, in which, first, annual rainfall and then time were considered as covari-
ates; the distributions investigated were the normal, gamma, two-parameter lognormal,
and three-parameter lognormal distributions.
Three different models were developed for the non-stationary analyses. While the
first assumed a linear relationship between the location parameter of the distributions and
mean annual rainfall, used as covariate, the second exploited a linear relationship between
location and shape parameter and the covariate, and the third modeled the relationship
between location parameter and the covariate as a cubic spline, maintaining the linear
dependence of the shape parameter. The goodness of fit of the models was evaluated
by the AIC method, and worm plots were derived to find which distributions best fitted
the dependence between rainfall and runoff and between time and runoff. It was also
found that, in moving from parametric (i.e., linear) to non-parametric (i.e., cubic-spline)
models, on average, the goodness of fit of the distributions provided by AIC remained
constant for the different analyzed cases. In general, the best model was characterized by a
linear variation of scale and shape parameters as functions of precipitation, while the most
frequently preferred distribution, as determined by AIC testing, was the three-parameter
lognormal distribution.
While, at present, many studies use temporal variables as covariates in non-stationary
hydrological frequency analyses, in this work the introduction of time as a covariate did
not improve the performance metrics because of the absence of statistically significant
temporal trends in the response variables.
In general, the comparison of different types of models indicates that non-stationary
models with observed annual rainfall series as covariates capture the variability of observed
data better than stationary models and non-stationary models with time as a covariate.
These results confirm that it is necessary but also effective to include physical covariates in
the non-stationary frequency analysis of runoff series.
Ultimately, the versatility of these models lies in being able to update the probability
distribution of the response variable as a function of time (if a marked trend is present or if
there are changes in the source variables, as was the case here). The resulting distributions
then allow one to consider an increase/decrease in variability and changes in the mean and
variance of the distributions used as a function of future changes in precipitation (or even
temperature, for example).
A future development of this study could consist of the derivation of rainfall forecasts
from seasonal models or climate models after opportune downscaling, obtaining forecasts
of runoff distribution as outputs. Another possibility is to improve the various models by
using other covariates, such as temperature, this being one of the main climate variables
used to capture climate-change signals. In this perspective, it is possible to expand the
dataset of watersheds in order to carry out a regionalization of the non-stationary dis-
tribution by introducing basin stationary parameters (e.g., curve numbers, impermeable
percentage areas of watersheds, population data, etc.).
Water 2022, 14, 2848 16 of 17

Author Contributions: Conceptualization, L.V.N.; methodology, P.S.; software, P.S. and D.T.; valida-
tion, L.V.N. and G.C.; formal analysis, P.S.; investigation, P.S., G.C. and L.V.N.; resources, L.V.N.; data
curation, P.S., D.T. and G.C.; writing—original draft preparation, P.S., D.T. and G.C.; writing—review
and editing, P.S., G.C., D.T. and L.V.N.; visualization, P.S., G.C., D.T. and L.V.N.; supervision, L.V.N.;
project administration, L.V.N.; funding acquisition, L.V.N. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: The data used in this study are publicly available at https://www.regi
one.sicilia.it/istituzioni/regione/strutture-regionali/presidenza-regione/autorita-bacino-distretto-
idrografico-sicilia/annali-idrologici (accessed on 1 August 2022).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Cannarozzo, M.; Noto, L.; Viola, F.; la Loggia, G. Annual Runoff Regional Frequency Analysis in Sicily. Phys. Chem. Earth Parts
A/B/C 2009, 34, 679–687. [CrossRef]
2. Markovic, R.D. Probability Functions of the Best Fit to Distributions of Annual Precipitation and Runoff Hydrology. Doctoral
Dissertation, Colorado State University, Fort Collins, CO, USA, 1965. Paper No. 8.
3. Vogel, R.M.; Wilson, I. Probability distribution of annual maximum, mean, and minimum streamflows in the united states.
J. Hydrol. Eng. 1996, 1, 69–76. [CrossRef]
4. Salas, J.D. Analysis and Modeling of Hydrological Time Series. Handb. Hydrol. 1993, 19, 11–72.
5. Liu, L.; Xu, H.; Wang, Y.; Jiang, T. Impacts of 1.5 and 2 ◦ C global warming on water availability and extreme hydrological events
in Yiluo and Beijiang River catchments in China. Clim. Change 2017, 145, 145–158. [CrossRef]
6. Donnelly, C.; Andersson, J.C.M.; Arheimer, B. Using flow signatures and catchment similarities to evaluate the E-HYPE multi-basin
model across Europe. Hydrol. Sci. J. 2016, 61, 255–273. [CrossRef]
7. Li, S.; Qin, Y. Frequency Analysis of the Nonstationary Annual Runoff Series Using the Mechanism-Based Reconstruction Method.
Water 2022, 14, 76. [CrossRef]
8. Sadri, S.; Kam, J.; Sheffield, J. Nonstationarity of low flows and their timing in the eastern United States. Hydrol. Earth Syst. Sci.
2016, 20, 633–649. [CrossRef]
9. Debele, S.E.; Bogdanowicz, E.; Strupczewski, W.G. Around and about an application of the GAMLSS package to non-stationary
flood frequency analysis. Acta Geophys. 2017, 65, 885–892. [CrossRef]
10. Jiang, C.; Xiong, L.; Xu, C.Y.; Guo, S. Bivariate frequency analysis of nonstationary low-flow series based on the time-varying
copula. Hydrol. Processes 2015, 29, 1521–1534. [CrossRef]
11. Kang, L.; Jiang, S.; Hu, X.; Li, C. Evaluation of return period and risk in bivariate non-stationary flood frequency analysis. Water
2019, 11, 79. [CrossRef]
12. Nasri, B.R.; Bouezmarni, T.; St-Hilaire, A.; Ouarda, T. Non-Stationary Hydrologic Frequency Analysis using B-Spline Quantile
Regression. J. Hydrol. 2017, 554, 532–544. [CrossRef]
13. Nogaj, M.; Parey, S.; Dacunha-Castelle, D. Non-stationary extreme models and a climatic application. Nonlinear Processes Geophys.
2007, 14, 305–316. [CrossRef]
14. Villarini, G.; Smith, J.A.; Napolitano, F. Nonstationary modeling of a long record of rainfall and temperature over Rome. Adv.
Water Resour. 2010, 33, 1256–1267. [CrossRef]
15. Xiong, L.; Jiang, C.; Du, T. Statistical attribution analysis of the nonstationarity of the annual runoff series of the Weihe River.
Water Sci. Technol. 2014, 70, 939–946. [CrossRef] [PubMed]
16. Yang, L.; Smith, J.A.; Wright, D.B.; Baeck, M.L.; Villarini, G.; Tian, F.; Hu, H. Urbanization and climate change: An examination of
nonstationarities in urban flooding. J. Hydrometeorol. 2013, 14, 1791–1809. [CrossRef]
17. Koutsoyiannis, D.; Montanari, A. Risks from dismissing stationarity. In Proceedings of the AGU Fall Meeting Abstracts,
San Francisco, CA, USA, 15–19 December 2014; p. H54F-01.
18. Matalas, N.C. Comment on the announced death of stationarity. J. Water Resour. Plan. Manag. 2012, 138, 311–312. [CrossRef]
19. Milly, P.C.; Betancourt, J.; Falkenmark, M.; Hirsch, R.M.; Kundzewicz, Z.W.; Lettenmaier, D.P.; Stouffer, R.J. Stationarity is dead:
Whither water management? Science 2008, 319, 573–574. [CrossRef]
20. Caracciolo, D.; Noto, L.V.; Istanbulluoglu, E.; Fatichi, S.; Zhou, X. Climate change and Ecotone boundaries: Insights from a
cellular automata ecohydrology model in a Mediterranean catchment with topography controlled vegetation patterns. Adv. Water
Resour. 2014, 73, 159–175. [CrossRef]
21. Francipane, A.; Fatichi, S.; Ivanov, V.Y.; Noto, L.V. Stochastic assessment of climate impacts on hydrology and geomorphology of
semiarid headwater basins using a physically based model. J. Geophys.Res. Earth Surf. 2015, 120, 507–533. [CrossRef]
22. Giuntoli, I.; Renard, B.; Vidal, J.-P.; Bard, A. Low flows in France and their relationship to large-scale climate indices. J. Hydrol.
2013, 482, 105–118. [CrossRef]
Water 2022, 14, 2848 17 of 17

23. Giuntoli, I.; Villarini, G.; Prudhomme, C.; Hannah, D.M. Uncertainties in projected runoff over the conterminous United States.
Clim. Change 2018, 150, 149–162. [CrossRef]
24. Kormos, P.R.; Luce, C.H.; Wenger, S.J.; Berghuijs, W.R. Trends and sensitivities of low streamflow extremes to discharge timing
and magnitude in Pacific Northwest mountain streams. Water Resour. Res. 2016, 52, 4990–5007. [CrossRef]
25. Jiang, C.; Xiong, L.; Yan, L.; Dong, J.; Xu, C.-Y. Multivariate hydrologic design methods under nonstationary conditions and
application to engineering practice. Hydrol. Earth Syst. Sci. 2019, 23, 1683–1704. [CrossRef]
26. Li, Y.; Chang, J.; Luo, L.; Wang, Y.; Guo, A.; Ma, F.; Fan, J. Spatiotemporal impacts of land use land cover changes on hydrology
from the mechanism perspective using SWAT model with time-varying parameters. Hydrol. Res. 2019, 50, 244–261. [CrossRef]
27. Katz, R.W.; Parlange, M.B.; Naveau, P. Statistics of extremes in hydrology. Adv. Water Resour. 2002, 25, 1287–1304. [CrossRef]
28. Villarini, G.; Smith, J.A.; Serinaldi, F.; Bales, J.; Bates, P.D.; Krajewski, W.F. Flood frequency analysis for nonstationary annual peak
records in an urban drainage basin. Adv. Water Resour. 2009, 32, 1255–1266. [CrossRef]
29. Rigby, R.A.; Stasinopoulos, D.M. Generalized additive models for location, scale and shape. J. R. Stat.Soc. Ser. C (Appl.Stat.) 2005,
54, 507–554. [CrossRef]
30. Jiang, C.; Xiong, L.; Wang, D.; Liu, P.; Guo, S.; Xu, C.-Y. Separating the impacts of climate change and human activities on runoff
using the Budyko-type equations with time-varying parameters. J. Hydrol. 2015, 522, 326–338. [CrossRef]
31. Li, J.; Tan, S. Nonstationary flood frequency analysis for annual flood peak series, adopting climate indices and check dam index
as covariates. Water Resour. Manag. 2015, 29, 5533–5550. [CrossRef]
32. López, J.; Francés, F. Non-stationary flood frequency analysis in continental Spanish rivers, using climate and reservoir indices as
external covariates. Hydrol. Earth Syst. Sci. 2013, 17, 3189–3203. [CrossRef]
33. Villarini, G.; Strong, A. Roles of climate and agricultural practices in discharge changes in an agricultural watershed in Iowa.
Agric. Ecosyst. Environ. 2014, 188, 204–211. [CrossRef]
34. Li, J.; Gao, Z.; Guo, Y.; Zhang, T.; Ren, P.; Feng, P. Water supply risk analysis of Panjiakou reservoir in Luanhe River basin of
China and drought impacts under environmental change. Theor. Appl. Climatol. 2019, 137, 2393–2408. [CrossRef]
35. Stasinopoulos, M.; Rigby, B.; Akantziliotou, C. Instructions on how to use the gamlss package in R Second Edition. 2008. Available
online: http://gamlss.com/wp-content/uploads/2013/01/gamlss-manual.pdf (accessed on 1 August 2022).
36. Akaike, H. A new look at the statistical model identification. IEEE Trans. Autom. Control. 1974, 19, 716–723. [CrossRef]
37. Akaike, H. On the likelihood of a time series model. J. R. Stat.Soc. Ser. D (Stat.) 1978, 27, 217–235. [CrossRef]
38. Nelson, D.B. Stationarity and persistence in the GARCH (1, 1) model. Econom. Theory 1990, 6, 318–334. [CrossRef]
39. Shumway, R.; Stoffer, D. Time Series Analysis and Its Applications with R Examples; Springer: New York, NY, USA, 2011; Volume 9.
40. Chen, H.-L.; Rao, A.R. Testing hydrologic time series for stationarity. J. Hydrol. Eng. 2002, 7, 129–136. [CrossRef]
41. Buuren, S.v.; Fredriks, M. Worm plot: A simple diagnostic device for modelling growth reference curves. Stat. Med. 2001,
20, 1259–1277. [CrossRef]
42. Stasinopoulos, M.D.; Rigby, R.A.; Bastiani, F.D. GAMLSS: A distributional regression approach. Stat. Model. 2018, 18, 248–273.
[CrossRef]
43. Rigby, R.A.; Stasinopoulos, D.M. A semi-parametric additive model for variance heterogeneity. Stat. Comput. 1996, 6, 57–65.
[CrossRef]
44. Rigby, R.A.; Stasinopoulos, M.D. Mean and Dispersion Additive Models. In Statistical Theory and Computational Aspects of
Smoothing; Physica-Verlag HD: Heidelberg, Germany, 1996; pp. 215–230.
45. Zhang, T.; Wang, Y.; Wang, B.; Tan, S.; Feng, P. Nonstationary Flood Frequency Analysis Using Univariate and Bivariate
Time-Varying Models Based on GAMLSS. Water 2018, 10, 819. [CrossRef]