Impact of rainfall variability and land-use changes on river discharge in Sanaga catchment (forest–savannah transition zone in Central Africa)

Changes in river discharges generally result from interactions between climate change and/or land-use changes (Oudin et al. 2018; Ebodé et al. 2021; Li et al. 2021; Wenng et al. 2021), although it is also admitted that the physical characteristics of the watershed (size, slope system, type of soil, etc.) increase their sensitivity to these factors (Gibson et al. 2005). Apart from a few attempts (Dzana et al. 2011; Bussi et al. 2022; Ebodé et al. 2022; Wu et al. 2022), research work focusing on the effects of these two factors are most often dissociated and evaluated separately. Most of the studies undertaken so far around the world have generally focused exclusively on one of these factors, as it is the case with recent studies in Africa and South America (Yira et al. 2017; Namugize et al. 2018; Nonki et al. 2019; Getahun et al. 2020; Gorgoglione et al. 2020).

In sub-Saharan Africa (SSA), the work devoted to investigate the impact of rainfall on discharges has been ongoing since the 1980s. Studies correlating rainfall and discharges are based on the detection of ruptures in the hydrometeorological series. Results from those studies confirm in the case of West Africa that the 1970s appear to be the main period of discontinuity marking the onset of hydroclimatic drought in the region (Mahé et al. 2001; Cisse et al. 2014; Nka et al. 2015; Bodian et al. 2020). In Central Africa, fluctuations in river discharges and rainfall have been observed at seasonal time scale (Ebodé et al. 2022). Liénou et al. (2008) demonstrated in the case of three equatorial rivers (Ntem, Nyong and Kienke) that the most significant climatic variations leading to changes in river discharges result from variations in rainfall during the dry season. The authors explain the sensitivity of the studied watersheds to rainfall variability by the fact that their reduction induces a significant deficit in soil moisture and groundwater storage resulting in a decrease in river discharges. Conversely, an increase in soil moisture during the dry season, therefore, enhances discharges at the start of the rainy season. It appears that the variability of discharge regimes of equatorial rivers can be better appreciated when rainfall of these seasons is considered. Concerning the impact of land-use changes on river discharges, previous studies have used supervised classifications of satellite images using at least two dates, to assess the dynamics of land-use and its impact on river discharges (Ebodé et al. 2020). Their results generally confirm an increase in discharges following an increase in impervious surfaces (buildings, roads and cultivated areas).

We anticipate that the series of nested watersheds in our study areas are likely to undergo similar changes. These watersheds are partly forested, but since the early 1970s, climatic fluctuations have been observed there as in the rest of SSA. In addition, population in these watersheds is increasing rapidly (BUCREP 2011). This demographic growth may likely lead to an increase in impervious areas, with serious implications on the hydrological dynamics of these hydrosystems. However, unlike the West African watersheds, the Sanaga watershed and its sub-watersheds has received substantially less attention from the research community due to the absence of observational data, while in some cases, the existing data in the region is riddled with gaps (Nkiaka et al. 2016). Furthermore, acquiring satellite images of sufficient quality over large areas (no haze and cloud) remains a big challenge. Previous studies focusing on the impact of rainfall variability and land-use changes in the SSA have focused on West, East and Southern Africa subregions while only few studies exist in the Central Africa subregion. This study focuses on the Sanaga river watershed, and its sub-watersheds located in middle part of Cameroon because it is the main source of water supply in Cameroon. In addition, the Sanaga river is also rich in halieutic resources and as such, it serves as a source of employment for many fishermen who supply their fish catch to markets. The Sanaga watershed also has several dams (Bamendjing, Mbakaou, Lom Pangar, etc.) aimed at supplying South Cameroon with electricity. The Sanaga river watershed, therefore, plays three important roles including water and food security, and electricity production in Cameroon.

This study focused on the analysis of rainfall/flows relationships over the recent period in the Sanaga watershed and its sub-watersheds, using updated hydrometeorological time series. It is crucial considering that this watershed is vulnerable to climate change and water scarcity is becoming a source of conflict and famine. Furthermore, insufficient hydroelectric production in the Sanaga watershed has been attributed to recent hydroclimatic changes mainly in the dry season, leading to frequent power cuts and load shedding. The main objective of this article is, therefore, to document the type and extent of regime changes in the Sanaga watershed due to climate change and anthropization. Considering that only few studies have attempted to investigate the impact of rainfall variability and land-use changes on river regimes in Central Africa, this paper contributes to this debate by focusing our analysis on the Sanaga River watershed in Cameroon.

River discharges of the Sanaga watershed were obtained from two sources. The Sanaga at Nachtigal series has been obtained from the Hydrological Research Center (HRC). It covers the period 1950–2015. This center manages a hydrometric database, mostly on a daily time step, which contains almost all the measurements carried out on Cameroonian territory since the beginning of the 1950s, for most of the stations. These data are riddled with gaps during the 1980 and 1990s. Indeed, during these decades, due to budgetary constraints, the hydrological service could no longer sustain the continuity of observations. This led to the abandonment of several hydrometric stations, including those of Sanaga catchment. The data used for the other watersheds (Song Mbengue, Djerem, Lom Pangar, Mape, Noun) comes from the Southern Interconnected Network (RIS) Cameroon database. These are naturalized flows developed jointly by Electricity of Cameroon (EDC), Electricity of France (EDF) and The Energy of Cameroon (ENEO). All the hydrological data used in this work was collected on a daily time step. The monthly, seasonal and annual modules were calculated subsequently.

The rainfall data used in this work comes from the Climate Research Unit (CRU) of the University of East Anglia in the United Kingdom. These data have been available since 1901, via the site https://climexp.knmi.nl/selectfield_obs2.cgi?id=2833fad3fef1bedc6761d5cba64775f0/ in NetCDF format, on a monthly time step and at a spatial resolution of 0.25° ×0.25°.

The spatial data used to analyze changes in land-use in the Sanaga watershed are mainly the Landsat 8 satellite images from January 2020 and Landsat TM from March 1984. All the images are made available to the general public free by the National Aeronautics and Space Administration (NASA), via the US Geological Survey site (https://earthexplorer.usgs.gov/), in GeoTIFF format. Uploaded images taken during the dry season (December to mid-March) were preferred over the rainy season because they are less affected by cloud disturbances.

The analysis of rainfall and average river discharges (at annual and seasonal time steps) was carried out using Pettitt (Pettitt 1979) and Mann-Kendall (Yue et al. 2002) tests, at the 95% significance level.

The concern to match the length of the rainfall series of each watershed to that of the discharges led us to analyze the rainfall of the watersheds studied over the intervals 1950–51 to 2015–16 (Nachtigal) and 1973–74 to 2019–20 (Song Mbengue, Djerem, Lom Pangar, Mape and Noun).

Annual and seasonal rainfall from the Sanaga watershed at Song Mbengue decreased from East to West (Figure 4). The limit of the lowest rainfall class (<1,700 mm for annual rainfall and <1,750 mm for the wet season rainfall) progresses considerably toward the West of the watershed, as do the isohyets 1,600 mm for annual rainfall and 1,620 mm for the wet season rainfall (Figure 4). In addition, the largest class (>2,200 mm for annual rainfall and >2,150 mm for the wet season rainfall), which occupied a considerable portion of the watershed until the 1970s, has almost disappeared in the decades after (Figure 4).

As for the rainfall of the dry season, they decreased from North to South of the watershed. The lowest rainfall class (<35 mm) and the 40 mm isohyet, for example, have progressed considerably toward the South (Figure 4). The fact that the largest rainfall classes (71–105 mm and >105 mm) present in the South of the watershed until the 1970s almost or completely disappeared afterwards also attests to this decrease in precipitation toward the South of the watershed.

The annual and seasonal flows of the Sanaga at Nachtigal decreased over the period 1950–51 to 2015–16. According to the tests used, these decreases are significant only for the average annual and wet season flows. Pettitt's test highlights a rupture in their series in 1970–71. The deficits observed following this rupture are respectively −17 and −22.7% (Figure 2). Other studies have reached a similar result in West (Nka et al. 2015) and Central Africa (Ebodé et al. 2020). Although having decreased overall during the 1970s, the annual and seasonal flows of the Sanaga at Nachtigal have experienced an increase since the 2000s (Table 3).

Between 1973–74 and 2019–20, the annual and seasonal flows of the Sanaga at Song Mbengue and those of its studied sub-watersheds generally increased (Figure 3). However, these increases are not significant according to the Pettitt and Mann-Kendall tests. The average dry season flows of the Lom Pangar watershed are the only ones for which the observed increase is significant according to the Mann-Kendall test. If we stick to the calculated p-values, the largest increases in average annual and wet season flows have been noted for the Mape and the lowest for the Noun (Figure 3). For the dry season, the lowest increases are those of the Noun and Song Mbengue (Figure 3).

The Sanaga watershed at Nachtigal is the only one for which the impact of rainfall is perceptible in the evolution of flows. For this watershed, these two variables experienced a decrease during the 1970s at the different time steps studied. The rupture observed in mean annual runoff and the wet season occurs before that of rainfall during these same periods, yet logically it is the opposite that should be observed. However, the fact that these ruptures occur during the same decade and that, they mark the beginning of an evolution in the same direction proves that there is an important link between these two variables. Other authors have already demonstrated a similar impact of rainfall on flows in SSA (Mahé et al. 2001).

Except for the Sanaga watershed at Nachtigal, flows increase throughout the study period for the rest of the watersheds, yet the precipitation that generates them decreases. It should also be noted that, even in the case of the Sanaga watershed at Nachtigal, despite this general downward trend, we noted a flows increase during the last two decades (2000 and 2010), while their rainfall decreased in recent years. These different situations led us to question the evolution of the land-use modes, which could explain such a phase shift between these two variables supposed to evolve in the same direction.

Such changes can only induce hydrological alterations like those observed in the Sanaga watershed at Song Mbengue and its sub-watersheds (increase in average flows). In a context where the precipitation that generates the flows is decreasing, the most logical thing would have been to see a flows decrease, which is not the case. The current rate of urbanization of these watersheds seems to be the most relevant factor to justify this trend. In this case, the decrease in precipitation seems to have been compensated by the increase in runoff. Thresholds of impervious surfaces beyond which urbanization is supposed to have, at the watershed scale, an influence on the flows of rivers are proposed in the literature, although the figures put forward by the various authors are somewhat little different. The threshold proposed by Yang et al. (2010) is the lowest (3–5%). Some authors place this threshold at 10% of impervious surfaces (Booth & Jackson 1997), while others place it at 20% (Brun & Band 2000). In all cases, the rates observed in the cases studied are higher than those proposed in the literature. Under these conditions, it is logical that hydrological alterations such as those highlighted occur in these watersheds. The impact of urbanization on flows has been already reported in studies carried out in SSA (Leblanc et al. 2007; Ebodé et al. 2020) and elsewhere (Dias et al. 2015; Lee et al. 2018).

The goal of this study was both to characterize, over the last 5 or 7 decades (depending on data availability), the hydroclimatic variability in the Sanaga watershed (at Song Mbengue and Nachtigal gauging stations) and on four of its sub-watersheds (Mbakaou, Lom Pangar, Magba and Bamendjing), and also to look for the key factors explaining the observed hydrological fluctuations. We used strong analytical methods in our study, namely the Pettitt and Mann-Kendall tests, as well as supervised classifications. Our results indicated that annual rainfall decreased throughout the Sanaga watershed. This decrease is statistically significant only in the case of the Sanaga watershed at Nachtigal (−5%), for which the study focused on a relatively long series, from 1950–51 to 2015–16. This seems quite logical insofar as this series incorporates the variability of the 1950 and 1960s, which are well known to be very wet in the region investigated. However, although the rainfall decreased in this watershed, the flows increased in the different cases studied, although these increases are not statistically significant. The 2010s seems to be particularly affected by this increase, including in the Sanaga watershed at Nachtigal, where the general trend of discharges is downward. The flows increase in the Sanaga watershed could be linked to the augmentation of impervious areas in the latter (between +181.3 and +1,300% for built and roads, and between +4.1 and +11.9% for bare soils), which could compensate for the drop in precipitation by increasing runoff.

Although this study provides useful information on the general behavior of flows in the Sanaga watershed, it nevertheless leaves some uncertainties related to the gaps in the observed flow series. Other information that could have been obtained in the absence of these gaps has probably remained latent. Regular measurements of the flows of the rivers in the Sanaga watershed are therefore essential to solve this problem.

The author warmly thanks the direction of the LMI DYCOFAC, at Yaoundé, for their administrative support.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.