Elsevier

Journal of Hydrology

Volume 566, November 2018, Pages 531-545
Journal of Hydrology

Trends in hydrological extremes in the Senegal and Niger Rivers

https://doi.org/10.1016/j.jhydrol.2018.07.063Get rights and content

Highlights

  • Streamflow extremes in West Africa are non-stationary and currently increasing.

  • Sudano-Guinean and Sahelian streamflow maxima evolution differ significantly.

  • Sudano-Guinean streamflow maxima switched from decreasing to increasing in the 1980s.

  • Sahelian streamflow maxima increased at a greater rate starting in the 1970s.

  • In the Sahel, current non-stationary return levels exceed stationary return levels.

Abstract

In recent years, West Africa has witnessed an increasing number of damaging floods that raise the question of a possible intensification of the hydrological hazards in the region. In this study, the evolution of extreme floods is analyzed over the period 1950–2015 for seven tributaries in the Sudano-Guinean part of the Senegal River basin and four data sets in the Sahelian part of the Niger River basin. Non-stationary Generalized Extreme Value (NS-GEV) distributions including twelve models with time-dependent parameters plus a stationary GEV are applied to annual maxima of daily discharge (AMAX) series. An original methodology is proposed for comparing GEV models and selecting the best for use. The stationary GEV is rejected for all stations, demonstrating the significant non-stationarity of extreme discharge values in West Africa over the past six decades. The model of best fit most commonly selected is a double-linear model for the central tendency parameter (μ), with the dispersion parameter (σ) modeled as either stationary, linear, or a double-linear. Change points in double-linear models are relatively consistent for the Senegal basin, with stations switching from a decreasing streamflow trend to an increasing streamflow trend in the early 1980s. In the Niger basin the trend in μ is generally positive since the 1970s with an increase in slope after the change point, but the change point location is less consistent. The recent increasing trends in extreme discharges are reflected in an especially marked increase in return level magnitudes since the 1980s in the studied Sahelian rivers. The rate of the increase indicated by the study results raises urgent considerations for stakeholders and engineers who are in charge of river basin management and hydraulic works sizing.

Introduction

River floods are one of the deadliest natural hazards in the world. They produce major damages on infrastructure, lead to economic losses, and favor water-borne diseases. In order to better understand such floods, hydrologists have long focused on assessing the rare (large in magnitude) river discharge values, represented by the tails of underlying statistical distributions (Gumbel, 1957). The primary aim, besides theoretical understanding, is to provide practical tools for flood risk management and civil engineering structure design. The main challenge for practical applications is estimating return levels for high return periods (typically 10, 50, or 100 years). The more the return period is larger than the length of the series, the greater the challenge of estimating the tail of the distribution.

Extreme Value Distributions (EVDs) are statistical tools designed for the study of such rare values (see e.g. Coles et al., 2001, Katz et al., 2002). For several decades, the main challenge when applying EVDs was to have a proper estimation of the tail (heavy, light, or bounded). The focus was then on developing robust estimation procedures (Regional Frequency Analysis – Hosking and Wallis, 1997; GRADEX and adaptation – Guillot, 1993, Paquet et al., 2013; Bayesian inference – Coles and Tawn, 1996 among many other developments) and then applying them to the longest hydrological series available (Koutsoyiannis, 2004). The stationarity assumption reigned during this “old hydrological world” (Milly et al., 2008).

However, both increases and decreases of extreme discharges have been reported via the evaluation of historical series around the world (e.g. Kundzewicz et al., 2005, Bower, 2010, Condon et al., 2015). A main challenge of hydrological extremes thus concerns the validity of the stationarity assumption and the implications of its rejection. Ongoing global changes are expected to increase flood hazard mainly through the intensification of the hydrological cycle due to global warming (Hirabayashi et al., 2013, Arnell and Gosling, 2016) and the degradation of land surfaces due to anthropic pressure (Brath et al., 2006, Elmer et al., 2012). Other factors also tend to reduce flood hazards, including negative precipitation trends found in drying regions and flood protection structures such as dams. While some regions witness resulting changes in flood frequency, in other regions, no changes have been detected (Villarini et al., 2009). This could be a result of either the absence of substantial changes in drivers that could trigger/influence flood trends or competing phenomena that act in opposite ways. It could also be due to the use of non-robust methodology to detect concrete changes in the EVD of discharge. The last case necessitates improved methods able to detect trends in series characterized by low signal to noise ratio.

While numerous studies on flood hazard evolution have been undertaken for developed countries, less has been done in the developing world. This is the case in particular over the tropics (Kundzewicz et al., 2005) which contain two thirds of the developing countries, including the poorest. Populations living in the tropics are notoriously vulnerable to climate hazards, including droughts and floods that can occur within the same year at a given place. Global changes are expected to strongly impact flood risks in the tropics with studies already reporting significant increases in the frequency of rainfall extremes (Allan et al., 2010, O’Gorman, 2012, Asadieh and Krakauer, 2015), land use/cover changes (Lambin et al., 2003, Erb et al., 2016), rapid rates of urbanization (Di Baldassarre et al., 2010), and increasing vulnerability of populations due to very high demographic growth; the population of the least developed countries is expected to double from now to 2050 (Population-Reference-Bureau, 2016). The strong internal variability of tropical climates, the lack of long-term hydrological observations, and the large uncertainty of climate projections in the tropics challenge the scientific community to provide reliable and relevant information to stakeholders so they can define suitable flood risk management strategies.

West Africa is one of the most critical tropical regions for examining hydrological non-stationarities as it is a region in which the issues described above are exacerbated. West Africa is known for having strong precipitation variability, especially at the decadal level (Nicholson, 2013). It underwent a devastating and long-lasting drought that abruptly started in the late 1960s and persisted through the 1970s and 1980s (Lamb, 1983, Barbé et al., 1997, Nicholson, 2000, Camberlin et al., 2002, Barbé et al., 2002, L’hote et al., 2002, Dai et al., 2004, Panthou et al., 2014, Bodian et al., 2011, Bodian et al., 2016). At the regional scale, this led to a decline in the flow of large rivers that was proportionally greater than the decrease in rainfall (Lebel et al., 2003, Andersen and Golitzen, 2005, Mahé and Paturel, 2009).

At the subregional scale however, two diametrically opposed hydrological behaviors were observed (Descroix et al., 2018). In the Sudano-Guinean subregion of West Africa (south of 12°N), a decrease in river flow was observed for small to regional scale catchments (Mahé et al., 2005, Descroix et al., 2009) until the 1970s and 1980s (Diop et al., 2017). The decrease in flow was attributed to a gradual drying up of the groundwater and thus a gradual decrease in the base flow of the rivers (Mahé et al., 2000, Mahé and Paturel, 2009). In the Sahelian region (belt between roughly 12°N and 16°N), runoff coefficients and runoff volumes increased despite the drought. This phenomenon – the so-called “Sahelian paradox” – was understood to have been caused by a change in surface conditions (Albergel, 1987, Descroix et al., 2009, Aich et al., 2015, Cassé et al., 2016). Droughts played a role in increasing surface crust and decreasing vegetation (Gal et al., 2017), which consequently increased runoff coefficients and counterbalanced the effects of drought (Boulain et al., 2009). Anthropogenic changes (including land use change) appear to be a major factor in some basins (Seguis et al., 2004, Li et al., 2007, Leblanc et al., 2008, Gal et al., 2017). Other factors such as an increase in the density of the drainage network may have played a role in the increase of flow (e.g. Favreau et al., 2009, Gal et al., 2017).

Since the early 1990s, both total rainfall and streamflow amounts have increased compared to the drought decades of the 1970s and 1980s, though they remain lower than in previous pre-drought decades (Lebel and Ali, 2009, Mahé and Paturel, 2009, Panthou et al., 2014, Tarhule et al., 2015, Diop et al., 2017). In the Sahel, the increase was accompanied by higher interannual variability (Ali and Lebel, 2009, Panthou et al., 2014) and overall persistence of drought conditions under certain indices (L’hote et al., 2002, Ozer et al., 2009). Of note is the increase in the intensity of rainfall during recent years (Ly et al., 2013, Panthou et al., 2014, Sanogo et al., 2015, Taylor et al., 2017). During the same period, an increase in the number and magnitude of extensive floods has been reported (Tarhule, 2005, Tschakert et al., 2010, Samimi et al., 2012, Sighomnou et al., 2013, Cassé and Gosset, 2015), causing extensive fatalities, damages, and population displacement. From the mean hydrographs of the Niger River at Niamey plotted for six decades from 1951 to 2010, Descroix et al. (2012) and Sighomnou et al. (2013) illustrated a strong increase in the intensity of the summer flood peak of the Sahelian tributaries during the 2000s, while the flood peaks coming from the remote Guinean tributaries and arriving at Niamey later in the year at Niamey were as low as in the 1970s. They also noted successive discharge records produced by Sahelian floods in 2010 and 2012, exceeding the Guinean flood.

The strong current and projected demographic growth in West Africa (Population-Reference-Bureau, 2016) is likely to increase the exposure of populations to floods, both from intensive and unplanned human settlements in flood-prone areas (Di Baldassarre et al., 2010), and from human-induced changes in land cover which affect runoff. Changes in hydrological extremes consequentially are particularly pressing for decision makers in West Africa, as the statistical tools used for infrastructure design have not been updated since the 1970s (Amani and Paturel, 2017). An improved quantitative understanding of how extreme flows are changing over time in the region has generated an urgent demand to design and manage structures such as dams and dikes and, as a result, aid in risk mitigation, as well as the development of hydroelectric energy and irrigation systems.

However there is still very little literature on quantifying extreme flow changes in West Africa. For Sudano-Guinean regions, Nka et al. (2015) found an overall decreasing trend that closely followed rainfall indices (although decreasing at a higher rate); this was the case for the Falémé branch of the Senegal River at Fadougou. When only a more recent time period is considered (since 1970), no significant trends were found in the Sudano-Guinean catchments, including the Falémé. Bodian et al. (2013) explored trends in annual maximum daily discharge (AMAX) values on the Bafing tributary of the Senegal River (Bafing Makana and Daka Saidou stations). They found that high points in the series occurred during the pre-drought period (1967 and 1955), whereas the minima of the AMAX occurred in 1984. Diop et al. (2017) found that extreme highs in the Bafing Makana series decreased by 18% over the series and especially since 1971, while extreme lows stayed stable. Aich et al. (2016) analyzed time series of AMAX values at several stations along the Niger River. They found that changes in AMAX series followed the decadal variability of mean annual precipitation over Guinean and Benue-area catchments (a wet period during 1950s and 1960s, followed by a dry period during 1970s and 1980s, and values close to the long-term mean after), while the floods produced by Sahelian tributaries have recorded a monotonic increase since the beginning of the 1970s. Nka et al. (2015) found positive trends in the extreme values of three Sahelian catchments studied (Dargol River at Kakassi, the Gorouol River at Koriziena, and the Goudebo River at Falagontou). They also found significant (Mann–Kendall test) increases in extreme values in both AMAX series and peak-over-threshold (POT) series for the Dargol River at Kakassi. Breaks in AMAX were detected in 1987, and for POT in 1993. Mean extreme values were found to be greater (twice as high) during the later subperiods.

The aim of this paper is to detect and quantify trends in extreme hydrological values in West Africa. Discharge series are analyzed in tributaries of the Niger and the Senegal rivers, two catchments that reflect two differing hydrological and climatic processes of the Sahelian and the Sudano-Guinean West Africa. The temporal evolution of the AMAX series is assessed by exploring different Generalized Extreme Value (GEV) models that range from a stationary GEV (S-GEV) to more complex non-stationary GEV (NS-GEV) models. The following study proposes an original methodology for identifying the most significant model to represent the evolution of extremes for a particular data series. Notably, the retained model is accompanied by significance levels and estimates of uncertainty. The retained model is used to compute time-varying frequencies of extreme flows. These changes are represented by the evolution of flow return levels over the last fifty to sixty years. Our results thus have implications for operational applications, as the design and operation of hydraulic structures depend on the magnitude of a flood event at a given return period.

Section snippets

West African hydro-climatic features

The climate of West Africa is controlled by the West African Monsoon. The rainfall belt roughly follows the seasonal migration of the position of the Intertropical Convergence Zone (ITCZ). During the boreal winter (December-February), the rainfall belt is located over the Gulf of Guinea at around 4°N. It moves northward during the spring and reaches its northernmost position during boreal summer (June-September). At the regional scale, it implies that the hydro-eco-climatic features vary along

Theoretical framework and methodology

Extreme value distributions (EVDs) applied in a non-stationary context comprise some of the recent robust methods proposed for the detection of non-stationarity (Olsen et al., 1998, Cunderlik and Burn, 2003, Re and Barros, 2009, Marty and Blanchet, 2012, Park et al., 2011, Beguería et al., 2011, Panthou et al., 2013, Blanchet et al., 2016). Besides performing detection, EVDs also permit the quantification of trends and the evaluation of uncertainty. The principle of these methods is based on

Selected GEV model

Table 2 shows the model of best fit that was selected for each station with a p-value (α) of 0.10 used in the LRT. One can note that for all stations, an NS-GEV model always fit the series of AMAX better than the stationary GEV model.

Fig. 6 visualizes an example of these models, with moving window parameter values (circles) and selected NS-GEV model μ values (line). Confidence intervals for μ are shown with dashed lines. One can see that the selected models follow the general trends seen in the

Comparison of results with literature

The above results confirm the rising trend in extreme streamflow values since the 1970s and 1980s, previously found in other studies (Cassé and Gosset, 2015, Nka et al., 2015, Aich et al., 2016).

The trends found in the Guinean stations in the Senegal River basin in this study followed the trends found in Aich et al. (2016) for the Guinean stations in the Niger River basin. In both studies a decreasing trend was found until approximately the mid-1980s, followed by an increasing trend. Results

Conclusions and implications

The preceding analysis proposed a selection of NS-GEV models for hydrological annual maxima in West Africa. In all cases, the NS-GEV model was significantly more representative of the data series than the stationary GEV model. The trend is positive since the 1970s for the Sahelian stations and the mid-1980s in the Sudano-Guinean stations, with both regions demonstrating an intensification of the hydrological signal. Certain parameters and return levels, notably the μ parameter and the 2 and

Acknowledgements

The research leading to these results has received funding from the

NERC/DFID Future Climate For Africa program under the AMMA-2050 project, grant NE/M020428/1. This work was also supported by the French national program EC2CO-LEFE “Recent evolution of hydro-climatic hazards in the Sahel: detection and attribution.” Ansoumana Bodian benefited from a grant received from the French Embassy in Senegal which financed exchanges with other researchers. We warmly acknowledge the Niger Basin Authority

References (112)

  • V. Aich et al.

    Time series analysis of floods across the Niger River basin

    Water

    (2016)
  • V. Aich et al.

    Climate or land use? Attribution of changes in river flooding in the Sahel zone

    Water

    (2015)
  • H. Akaike

    A new look at the statistical model identification

    IEEE Trans. Autom. Control

    (1974)
  • J. Albergel

    Sécheresse, désertification et ressources en eau de surface: application aux petits bassins du Burkina Faso

    The influence of climate change and climatic variability on the hydrologic regime and water resources

    (1987)
  • A. Ali et al.

    The Sahelian standardized rainfall index revisited

    Int. J. Climatol.

    (2009)
  • R.P. Allan et al.

    Current changes in tropical precipitation

    Environ. Res. Lett.

    (2010)
  • A. Amani et al.

    Le projet de révision des normes hydrologiques en Afrique de l’Ouest et Afrique Centrale

    La Météorologie

    (2017)
  • I. Andersen et al.

    The Niger River Basin: A Vision for Sustainable Management

    (2005)
  • A. Anyamba et al.

    Analysis of Sahelian vegetation dynamics using NOAA-AVHRR NDVI data from 1981–2003

    J. Arid Environ.

    (2005)
  • N.W. Arnell et al.

    The impacts of climate change on river flood risk at the global scale

    Clim. Change

    (2016)
  • B. Asadieh et al.

    Global trends in extreme precipitation: climate models versus observations

    Hydrol. Earth Syst. Sci.

    (2015)
  • S. Beguería et al.

    Assessing trends in extreme precipitation events intensity and magnitude using non-stationary peaks-over-threshold analysis: a case study in northeast Spain from 1930 to 2006

    Int. J. Climatol.

    (2011)
  • C.J. Bélisle

    Convergence theorems for a class of simulated annealing algorithms on Rd

    J. Appl. Probab.

    (1992)
  • J. Blanchet et al.

    Trend in the co-occurrence of extreme daily rainfall in West Africa since 1950

    J. Geophys. Res.: Atmos.

    (2018)
  • J. Blanchet et al.

    Spatial analysis of trend in extreme daily rainfall in southern France

    Clim. Dyn.

    (2016)
  • G. Blöschl et al.

    At what scales do climate variability and land cover change impact on flooding and low flows?

    Hydrol. Process.

    (2007)
  • A. Bodian et al.

    Caractérisation spatio-temporelle du régime pluviométrique du haut bassin du fleuve Sénégal dans un contexte de variabilité climatique

    Physio-Géo. Géographie physique et environnement

    (2011)
  • A. Bodian et al.

    Analyse des débits de crues et d’étiages dans le bassin versant du fleuve Sénégal en amont du barrage de Manantali

    Revue du LACEEDE. Climat et développement

    (2013)
  • A. Bodian et al.

    Hydrological evaluation of TRMM rainfall over the upper Senegal River basin

    Hydrology

    (2016)
  • A. Bodian et al.

    Evolution des caractéristiques des pluies journalières dans le bassin versant du fleuve Sénégal: Avant et après rupture

    Hydrol. Sci. J.

    (2016)
  • Bonneau, M., 2001. Programme d’optimisation de la gestion des réservoirs. Besoins en eau de l’agriculture irriguée et...
  • N. Boulain et al.

    Water balance and vegetation change in the Sahel: a case study at the watershed scale with an eco-hydrological model

    J. Arid Environ.

    (2009)
  • S.S. Bower

    Natural and unnatural complexities: flood control along Manitoba’s Assiniboine River

    J. Historical Geography

    (2010)
  • J.-F. Boyer et al.

    SIEREM: an environmental information system for water resources

    IAHS publication

    (2006)
  • A. Brath et al.

    Assessing the effect on flood frequency of land use change via hydrological simulation (with uncertainty)

    J. Hydrol.

    (2006)
  • R.H. Byrd et al.

    A limited memory algorithm for bound constrained optimization

    SIAM J. Sci. Comput.

    (1995)
  • P. Camberlin et al.

    Pluviométrie et crises climatiques en Afrique tropicale: changements durables ou fluctuations interannuelles?

    Historiens et Géographes

    (2002)
  • B. Cappelaere et al.

    The AMMA-CATCH experiment in the cultivated Sahelian area of south-west Niger-investigating water cycle response to a fluctuating climate and changing environment

    J. Hydrol.

    (2009)
  • P. Cappus

    Etude des lois de l’écoulement-Application au calcul et à la prévision des débits

    La houille blanche

    (1960)
  • C. Cassé et al.

    Analysis of hydrological changes and flood increase in Niamey based on the PERSIANN-CDR satellite rainfall estimate and hydrological simulations over the 1983–2013 period

    Proc. Int. Assoc Hydrol. Sci.

    (2015)
  • C. Cassé et al.

    Model-based study of the role of rainfall and land use–land cover in the changes in the occurrence and intensity of Niger red floods in Niamey between 1953 and 2012

    Hydrol. Earth Syst. Sci.

    (2016)
  • S. Coles et al.
    (2001)
  • S.G. Coles et al.

    A Bayesian analysis of extreme rainfall data

    Appl. Stat.

    (1996)
  • L. Condon et al.

    Climate change and non-stationary flood risk for the upper Truckee River basin

    Hydrol. Earth Syst. Sci.

    (2015)
  • J.M. Cunderlik et al.

    Non-stationary pooled flood frequency analysis

    J. Hydrol.

    (2003)
  • A. Dai et al.

    A global dataset of Palmer Drought Severity Index for 1870–2002: relationship with soil moisture and effects of surface warming

    J. Hydrometeorol.

    (2004)
  • A.C. Davison et al.
    (1997)
  • L. Descroix et al.

    Change in Sahelian rivers hydrograph: The case of recent red floods of the Niger River in the Niamey region

    Global Planet. Change

    (2012)
  • L. Descroix et al.

    Evolution of surface hydrology in the sahelo-sudanian strip: an updated review

    Water

    (2018)
  • L. Descroix et al.

    Spatio-temporal variability of hydrological regimes around the boundaries between Sahelian and Sudanian areas of West Africa: a synthesis

    J. Hydrol.

    (2009)
  • G. Di Baldassarre et al.

    Flood fatalities in Africa: from diagnosis to mitigation

    Geophys. Res. Lett.

    (2010)
  • L. Diop et al.

    Trend analysis of streamflow with different time scales: a case study of the upper Senegal River

    ISH J. Hydraul. Eng.

    (2017)
  • T. d’Orgeval et al.

    Sensitivity of the West African hydrological cycle in ORCHIDEE to infiltration processes

    Hydrol. Earth Syst. Sci.

    (2008)
  • B. Efron

    Computers and the theory of statistics: thinking the unthinkable

    SIAM Rev.

    (1979)
  • B. Efron et al.

    An Introduction to the Bootstrap

    (1994)
  • F. Elmer et al.

    Drivers of flood risk change in residential areas

    Nat. Hazards Earth Syst. Sci.

    (2012)
  • K.-H. Erb et al.

    Exploring the biophysical option space for feeding the world without deforestation

    Nat. Commun.

    (2016)
  • G. Favreau et al.

    Land clearing, climate variability, and water resources increase in semiarid southwest Niger: a review

    Water Resour. Res.

    (2009)
  • Faye, C., 2015. Impact du changement climatique et du barrage de Manantali sur la dynamique du régime hydrologique du...
  • R. Fletcher et al.

    Function minimization by conjugate gradients

    Comput. J.

    (1964)
  • Cited by (56)

    • Analysis of the nonstationarity characteristics and future trends of flood extremes in the Dongting Lake Basin

      2022, Journal of Hydrology: Regional Studies
      Citation Excerpt :

      Frequent global occurrence of flood disasters is closely correlated to the critical changes in the variability and trends of extreme hydrometeorological events (Lu et al., 2020; Tegegne et al., 2020; Zhang et al., 2021; Zhou et al., 2021). According to literature, in many watersheds worldwide, the time series of flood extremes exhibits nonstationarity in changing contexts (Ishak et al., 2013; López and Francés, 2013; Solander et al., 2017; Wilcox et al., 2018; Sun et al., 2018b, 2018a; Grandry et al., 2020; Engeland et al., 2020; Chen et al., 2021). Milly et al. (2015) reported that "stationarity is dead."

    • Long-term variability in hydrological droughts and floods in sub-Saharan Africa: New perspectives from a 65-year daily streamflow dataset

      2022, Journal of Hydrology
      Citation Excerpt :

      These results are consistent with trend patterns in the duration, intensity, and frequency of hydrological extremes, with changing trends over the recent period (post-1990s), as compared to a historical baseline (1950–2014). These results also corroborate previous findings, highlighting increasing frequency and intensity (based on annual maximum) in flood events or disasters after the 1980s–90s over western Africa (Nka et al., 2015; Aich et al., 2016; Li et al., 2016; Fiorillo et al., 2018; Wilcox et al., 2018; Tazen et al., 2019; Tramblay et al., 2020a), the Blue Nile (Conway 2009; Degefu et al., 2019; Tramblay et al., 2020), eastern Africa (Bernard et al., 2013; Tramblay, et al., 2020a), and southern Africa (Do et al. 2017; Tramblay et al. 2020a), and confirm significant variations in flood and drought hazards over the region. More importantly, using multi-temporal trend analysis and MODWT, we found that analysis of long-term changes and trends in flood and drought characteristics are significantly influenced by aperiodic decadal variability across SSA.

    • High-impact weather and urban flooding in the West African Sahel – A multidisciplinary case study of the 2009 event in Ouagadougou

      2022, Weather and Climate Extremes
      Citation Excerpt :

      In Burkina Faso, floods typically occur between August and September as a result of intense rainfall from the WAM (Tazen et al., 2018). Evidence suggests an increasing trend in flood magnitude and frequency in Sahelian catchments mirrors trends shown in extreme rainfall (Nka et al., 2016; Wilcox et al., 2018) but has also been linked to change in land use and land management (Mahe et al., 2005; Descroix et al., 2012). For Burkina Faso the frequency of floods has increased substantially from around 1 event per year during the period 1986–2005 to 5 events during 2006–2010 and highlights a recent period of increasing flood frequency (Tazen et al., 2018).

    View all citing articles on Scopus
    View full text