admin
October 20, 2024
Strengthening of the hydrological cycle in the Lake Chad Basin under current climate change
Central Sahel is affected by a multidecadal wet and dry alternating period 1. During the twentieth century, this region experienced at least two wet phases, the last and most important of which occurred from the 1950s to the 1960s 2. This wet period was followed from the 1970s to the 1990s by severe and dramatic droughts 3, which are recognized as the largest climatic signals ever recorded over such large areas since the observation period 4,5. However, after the 1990s, partial rainfall recovery occurred, resulting in the regreening of Sahel 6. Although this trend is affected by high interannual variability, it has been confirmed since the mid-2010s 7. This period is particularly characterized by extreme rain events 8. It is estimated that these extreme events represent 90% of the rain causing floods of unprecedented magnitude 9. For example, the city of Ouagadougou in Burkina Faso experienced heavy flooding in 2009 following a 10-h rain event that totalled 263 mm. This event resulted in the displacement of 150,000 inhabitants and destroyed many infrastructures 10. Similarly, in 2010 and 2012, Niamey in Niger experienced the largest floods ever recorded since the observations began in 1929 10. These events are generally reported in the media and in reports from humanitarian nongovernmental organizations. However, few scientific studies have been conducted on flood characterization and its consequences for the population (loss of life, destruction of houses and infrastructure) and on the rainfed agriculture that characterizes this region 11. In addition, with recent climate modelling experiments predicting an increase in precipitation by + 30% by 2050 12, documenting what is currently happening in the central Sahel and what could be expected in the context of ongoing global warming is crucial.
In 2022, Sahel from Senegal to Chad was particularly affected by both fluvial and pluvial (rainfall-related) flood events 13. This year was characterized by an earlier rainy season onset and an amount of precipitation above normal in many regions. This situation was anticipated by the Regional Climate Outlook Forum for West Africa, which issued an alert in early May on the risk of a longer and more intense rainy season than normal 14. With at least 612 and 195 fatalities, respectively, the floods over Nigeria and Niger are among the deadliest in the country’s history 15,16,17. The devastation in Nigeria is worse than that in the 2012 flood disaster 18, in which 34 out of 36 states and more than 3.2 million people were affected—including 1.5 million displaced and 2.776 injured 19,20. Several hundred thousand hectares of land have been inundated, causing damage to more than 300 thousand homes and more than half a million hectares of farmland.
The Lake Chad basin is one of the areas that was strongly impacted by this event. Occupying an area of 2,500,000 km2, it is now hydrologically active only in its southern area, which is characterized by a Sudanese climate and drains the rainfall received over an area of 610,000 km2. In 2022, 19 out of 23 provinces located in the southern part of the basin were flooded, with more than 1,468,847 people affected and 350,000 ha of land devastated 21. Due to its location at the confluence of the Chari and Logone Rivers, the capital, N’Djamena, is an extremely exposed city. In 2022, 254,483 people were affected and supported by the humanitarian system 21. On October 2023, 19th, Chad declared a state of emergency 22.
In this report, we focused on fluvial floods. We used ground observations based on gauge stations from the hydrological network of the Chari and Logone Rivers complemented with satellite observations, especially for Lake Chad, which is still poorly monitored (see “Methods”). Using optical imagery from the Moderate Resolution Spectroradiometer (MODIS) and time series of lake water levels derived from multimission altimetry satellites (i.e., TOPEX/Poseidon, ENVISAT, Jason-2 and -3, SARAL, Sentinel-3A, and Sentinel-6A), we reconstructed the surface water extent and level of the lake until December 2022. These methods have already been tested in a previous study demonstrating their validity for estimating open water surfaces compared to surfaces covered by vegetation 23; lake Chad, which is in a flat landscape, is a shallow lake partly covered by aquatic vegetation that could partly hamper good estimations of the water surface by satellite survey.
The Lake Chad Basin: a hydrological giant in central Sahel
The Lake Chad drainage basin covers 2.5 million km2, representing 8% of the African continent (Fig. 1). It is a hydrologically closed drainage system located at the southern fringe of the Sahara in the Central Sahel region of northern Africa. The basin features a Sudanese climate in the southern region and a semiarid climate in the northern region. The climate is tropical with two seasons. The wet season begins in May and ends in October, and most of the precipitation falls in July and August. Evaporation rates show a strong seasonal cycle that follows changes in the air temperature 24.
Location of the Chari-Logone basin in central Africa, south of the Lake Chad Basin, corresponding to the present hydrologically active basin (SRTM Topo 30 QGIS 3.4.5). The main gauge stations are indicated by squares. The main cities are indicated by circles.
Today, water intakes come from the southern portion of the drainage basin and are transported by river discharge from the Chari-Logone system, the Komadougou Yobe River located in Niger and by precipitation on the lake surface (Fig. 1). Even if the Chari-Logone River system covers only 25% of the lake Chad basin, it accounts for up to 82% of the lake water supply 25. The Chari and Logone Rivers drain a watershed with an area of 610,000 km2 located between 5° N and 12° N (Fig. 1). The Logone River, with a length of 1000 km, originates on the Adamawa Plateau in Cameroon and has an altitude ranging from 305 to 835 m 26. The total area of the Logone catchment is 90,000 km2. The Chari River starts in the Central African Republic at an altitude ranging from 500 to 600 m, and it flows nearly 1200 km. The Chari Basin covers an area of approximately 523,000 km2. Both rivers converge at N’Djamena before reaching Lake Chad, 110 km downstream of the city (Fig. 1). Based on river discharge time series, eight subbasins were delineated, and their contributions to the debit flow were estimated 24.
Lake Chad is the terminal lake of this drainage basin system. Today, it is a shallow lake (< 3 m) subdivided into two main areas most of the time, the northern and southern pools, which are separated by an east‒west vegetation-covered sand barrier known as the Great Barrier. The northern pool has an altitude of 275.3 m, whereas the southern pool is located at 278.2 m and the great barrier is located at 279.3 m 26. The inputs from the rivers in the lake are balanced by evaporation, estimated at more than 2000 mm/year, and by seepage to groundwater 27.
The lake Chad basin is considered as a good sensor of climate 25,27. Studies have investigated the role of anthropic activities on the runoff and flow of the main tributaries, demonstrating that there is no clear evidence of an anthropogenic impacts, e.g. withdrawal for domestic and cropping uses, responsible for a decrease in flows or a modification of the hydrological regime in the Chari-Logone basin 24. Moreover, the basin has been poorly managed, any hydraulic infrastructures have been built and the few managements existing do not imply a modification of the river flow.
2022: The greatest flood recorded in the Lake Chad basin since the 1960s
The flood hydrograms from the Chari and the Logone are derived from data recorded by gauge stations located in the Lake Chad Basin (Fig. 1). These gauge stations are read daily at the same time at 8 a.m. GMT + 0. The level of the rivers was measured in meters above the zero reference point when the limnometric scale was used (see “Methods”). Here, we focused on the gauge station from N’Djamena (named N’Djamena TP), which is located 2.4 km downstream of the confluence of the Chari and the Logone at 288.58 m. We also used the gauge stations located at Sarh and Manda located upstream in the southern part of the basin at 355.18 m and 354.11 m, respectively (Fig. 1). Both stations represent 58% of the flow rate from the Chari River 24. From Logone, we used the Logone Gana station, which is located at 295.21 m and reflects 34% of the water input for this river.
The 2022 hydrological season in the southern basin began in early June (Supplementary Fig. S1). At Sahr, the level of the Chari increased slowly until the end of May, starting at 1.50 m before accelerating after this month. It quickly reached 2.32 m on July 23, 2021, and then increased steeply until October 16, 2021, when it reached its maximum at 7.14 m. Then, it decreased until the end of the dry season at the end of April 2023. In Manda, we observe the same dynamics, with an acceleration of the river flow occurring slightly earlier between July 10 and 12, 2022 (Supplementary Fig. S1). We also observe a period of high variability between August 9 and 28, 2022, at the Manda gauge station. Then, the level increased steeply, reaching its highest level on October 4, 2022, at 7.27 m, slightly earlier than in Sarh.
From the Logone River at Logone Gana station (Supplementary Fig. S1), the level began to rise slowly at the end of May and then rose very quickly until the end of August, when it reached an average of 2 cm per day. The increase in river level then slowed to less than 1 cm per day. The peak flood occurred between October 23 and November 12, 2022, oscillating between 6.19 m and 6.23 m.
At N’Djamena, as in the southern part of the basin, the river level started increasing slowly in June (Fig. 2). A first increase in the level was recorded early in August; it slowed at the beginning of September before a rapid increase until October 22, 2022. At this elevation, it lasted until October 30, 2022. Then, it decreased at 8.02 m on November 6, 2022, before continuing to increase until it reached its highest level on November 13, 2022 at 8.14 m. After this date, the level decreased rapidly.
Hydrogram of N’Djamena (gauge Station N’Djamena TP) for 2022 compared with 1961, 2012 and 2020. The plain lines indicate the mean maximum level measured during the decades 1950–1969, 1970–1989, 1990–1999, 2000–2009 and 2010–2022.
In N’Djamena, we have a record of the river level since 1933 (Supplementary Fig. S2), suggesting a historical perspective. When comparing the level reached in 2022 with that in this time series, we observed that the year 2022 was the year with the greatest difference recorded over the last 60 years. In 1961, the elevation of the Chari River reached 9.10 m, the wettest year recorded during the last ninety years (Fig. 2). This corresponded to a flow rate recorded at 5160 m3/s (Supplementary Fig. S3). The years when the Chari level exceeded 7.50 m at N’Djamena were considered to be very wet years because significant impacts were observed. During recent decades, the two wetter years recorded before 2022 were 2012 and 2020, with heights of 7.66 m and 7.47 m, respectively (Fig. 2). The flow rates recorded in 2012 and 2020 were 3548 m3/s and 3409 m3/s, respectively. The years 1988 and 1998, with 7.58 m (3472 m3/s) and 7.33 m (3311 m3/s), respectively, were also characterized by a strong rainy season, especially 1988, which occurred during the dry phase (Supplementary Fig. S2). However, for that year, we have no records about the societal impacts.
Lake Chad, the outlet of the Chari-Logone rivers, expands according to water inputs. In 2022, the water level of the lake started to increase beginning in the middle of August, reaching its maximum at 3.3 m recorded at the Bol gauge station on December 16, 2022 (Supplementary Fig. S4). After this date, it decreased slowly until April 2023. The delay observed between the maximum level of Chari at N’Djamena (Fig. 2) and the maximum level of Lake Chad was approximately one month (Supplementary Fig. S4). This delay usually corresponds to the timing of water transfer from Chari to the lake 7,25,26. We completed these ground observations with MODIS imagery to estimate its maximum extent, including the southern and northern pools. In 2022, the maximum surface water extent, without vegetation, was 5370 km2 (Fig. 3). This corresponds to a 24.76% increase compared to the average annual maximum lake extension for the 2012–2022 period, which was approximately 4310 km2 (Pham et al., 2020). However, for obtaining a total estimation of the water extent of the lake, it is necessary to consider the water covered by vegetation. Then, the total surface water area reached 18,800 km2, corresponding to an extension of approximately 7.5% compared to the mean annual maximum of the lake during the 2012–2022 period, which was approximately 17,500 km223. During the wet 1950–1960 observation decades, it was estimated that the lake area reached 25,000 km225. Using satellite altimetry data (see “Methods”), the lake reached a maximum level of 281.36 m in December 2022. In 1963, the maximum level was 282.48 m, and in 1973, it was 281.47 m 26.
Time series (top) and anomalies (bottom) of the surface water extent in the northern pool (blue), southern pool (red) and Lake Chad (green) for the 2001–2022 period. The linear trends are also plotted.
The impact of global climatic change
Precipitation in tropical Africa results from a combination of factors, including the monsoonal on-land flow of moist air, low-level convergence of air at the intertropical convergence zone and, of particular importance, the deep vertical motion of air, which is modulated by the interaction of the Tropical Easterly Jet (TEJ) and the African Easterly Jet (AEJ) over northern Africa 28. The monsoon circulation is driven by the thermodynamic contrast between the cooler ocean (the equatorial Atlantic) and an intense heat low (the Saharan heat low), which develops over the western Sahara during the boreal summer. Southwesterly winds are established between the equatorial Atlantic and the Saharan heat low, bringing moisture to the continent 28. The African Easterly Jet (AEJ) is also driven by the low-temperature gradient from the Gulf of Guinea to the Sahara Desert. Associated with easterly waves, these synoptic-scale disturbances propagating westwards on the AEJ, from central and eastern Africa across West Africa and into the tropical Atlantic, are characterized by a peak in convection and precipitation 29. Moreover, since the beginning of the century, Central Sahel has experienced a new era of climate extremes characterized by a transient state combining a rarefaction of rainfall occurrence and a reinforcement of rainfall intensities 30. The average intensity of rainy events has increased, especially for the largest rainy events at daily and subdaily timescales 31. The 2022 rainy season was characterized by mixed conditions. Flooding occurred because of above-average rainfall reinforced by extreme events. These events are attributed to mesoscale convective systems (MCSs), which tend to increase their vertical development, favouring convergent humidity and producing exceptional cumulative rainfall 9. A warmer Sahara intensifies convection within Sahelian storms through increased wind shear and changes to the Saharan air layer, leading to more intense, heavy precipitation.
For 2022, climate change is estimated to emphasize the rainfall event by approximately 80 times, leading to an increase in intensity of approximately 20% 32. The seasonal rainfall increases significantly due to rising global temperatures, as confirmed by observations over the Lake Chad catchment 32. The return time of the 2022 event was estimated to be approximately 1 event in 10 years, with rainfall increases ranging between 30 and 40%. Moreover, the most recent experiments further noted that tropical sea surface temperatures and associated dynamic changes could increase rainfall in central Africa, even if rainfall projections are uncertain because of contrasting signals from models with especially large discrepancies during the main rainy season 33. From ground observations, the mean annual observed rainfall in N’Djamena has increased since 2018 (except for 2021), whereas in Sarh, which is south of Chad, this trend is pronounced (Supplementary Fig. S5).
The challenges of water management
In this context, it is crucial to obtain the best estimation of these recurrent flooding events for the next few decades. We first estimate the historical flood return period based on the maximum flow rates observed at N’Djamena between 1960 and 2022 (see “Methods”). We estimated that the historical floods of 2012 (3548 m3/s measured in November), 2020 (3409 m3/s measured in November) and 2022 (3986 m3/s measured in November) had return periods between 2 and 10 years (Fig. 4). A statistical analysis of the flow data revealed that for a return period of 2 years, the maximum observed flow reached 2753 m3, which was comparable to the values recorded in 1958 (2802 m3/s), 1968 (2781 m3/s) and more recently in 2018 (2760 m3/s). For a 20-year return period, the maximum flood is 4814 m3, comparable to the flow rate observed during the moister decades in 1950 (4400 m3/s) and 1960 (4008 m3/s). In 1961, the maximum recorded flow rate was 5156 m3/s (Supplementary Fig. S3; Supplementary Table S1).
Return period for historical floods based on the distribution function of Gumbel.
We completed these estimations by predicting the future flow rate of the Chari–Logone River system. A hydrological model developed at the scale of the Chari-Logone basin using the ‘Modèle Grand Bassin’ (MGB) was first developed (see “Methods”). Qualitative analysis based on four performance criteria highlights the confidence in the accurate representation of hydrological processes across the whole basin (Supplementary Table S2). For the N’Djamena station, the model adequately represented the general dynamics of discharge with good representations of low and high flows (Supplementary Fig. S6). The uncertainties in the simulations are estimated to be approximately 17% and 12% compared to the observed values when the 5th and 95th percentiles are considered for the observations and simulations, respectively. Then, we selected five CMIP6 models and two shared socioeconomic pathway scenarios (e.g., the SSP2-4-5 medium forcing scenario and the SSP3-7-0 high forcing scenario) to predict the Chari-Logone flow rates for the prospective 2040–2060 decade (Supplementary Table S3; see “Methods”). Figure 5 shows that historical data tend to underestimate low flow rates, especially moderate flow rates (< 3000 m3/s), and conversely overestimate high flow rates (> 3000 m3/s). However, compared to the observations, the climate projections from the ‘ipsl,’ miroc,’ and ‘nesm’ models consistently yield significantly greater flow rates across all flow rate ranges. Regarding the results from the ‘access’ model, we observe that the estimations align more closely with the current observations. Additionally, there is a noticeable decrease in flow rate compared to that of the observations in the case of the ‘cmcc’ model. Figure 6 shows the results of the corrected flow quantiles \({Q}_{F}^{c}\) for the 10- and 5-year return periods. Note the crossover of the curves for the recent period by construction. Models struggle to reproduce the wet/dry cycle observed in the 1950s and 1990s. However, despite these uncertainties, the simulation outputs under the constraints of climate change from two selected scenarios revealed a clear increase in average flow rates over the prospective period of 2040–2060, particularly a significant increase in peak flow values, surpassing the annual mean value almost every year for most models. For the 2050 horizon, the return flows are approximately the same as the level of return flows observed in the 1950s, the most recently observed wettest period. It is worth noting that the climatic signal of wetter central Sahel in the future from CMIP6 models is characterised as ‘low confidence’ by the IPCC report due to the variability between CMIP6 models and between regional (CORDEX) and global experiments (CMIP6) (6th report section 12.4.1.2). From IPCC 6th report, increasing precipitation for a 1.5 °C and 2 °C global warming levels are found in central and eastern Sahel with low confidence and the wet signal is getting stronger and more extended for a 3 °C and 4 °C warmer world. Although rainfall is not a robust variable in climate projections, we believe that the impacts of such an increase should be taken into account.
Outputs from projection models (2040–2060), historical CMIP6 data and observations (1950–2014) at N’Djamena based on MGB simulation.
Evolution of peak flows at 10- and 20-year return periods between 1933 and 2100. The pink curve shows the evolution of the Q_0 observations. The brown curve shows the Q_R reconstructed flows (model + observations). The colour curves show the flow rates from the forced MGB with CMIP6 models. Before 2015, these are the historical Q_H flows, and after 2015, the prospective Q_F flows are for the SSP245 scenario.
As observed, the limitations in accurately representing certain years in the current climate may be attributed to the choice of precipitation input data for the model. The use of long-term data is often associated with reduced interannual variability. Therefore, it would be worthwhile to work with improved rainfall datasets at finer spatial and/or temporal scales to better capture the interannual variability in climate events and, consequently, hydrological phenomena. Regarding the current period, it would be practical to verify whether the rating curves used to estimate flow rates require recalibration, especially for a better understanding of low-flow conditions. Concerning the negligible impact of the new land use map for 2050 on flow predictions (Supplementary Fig. S7), it would be interesting to conduct a similar analysis while increasing anthropogenic areas near water bodies to assess sensitivity, regardless of the precipitation dataset used.
To ensure that projection models align with observational reality for past periods and are capable of forecasting future river floods in N’Djamena, it would be useful to implement dichotomous prediction methods using a threshold, for instance, probability of detection (POD) indices or false alarm rate (FAR) indices. Preliminary tests were conducted using these two methods to compare observed data and historical data from various CMIP6 models for the period of 1950–2013, with a defined threshold of 3000 m3/s (Supplementary Table S4). False alarms are prevalent across all the models, indicating a trend towards overestimating flood peaks, even in the past, and detecting an excessive number of flood events compared to reality.
Despite the uncertainties in climate modelling experiments, our results show that the trend of increasing flood events has been confirmed in response to the intensification of the hydrological cycle in the context of ongoing global warming. For the Lake Chad basin, we observed the important and dramatic consequences that these events had on the Sahelian economy and on populations. The Chari and Logone Rivers are poorly regulated, which makes the Lake Chad Basin highly vulnerable. The southern part of Chad, which is the most populated area of the country and the most productive in terms of livelihoods, is strongly impacted. Additionally, the city of N’Djamena located on the shore of Chari after the confluence with the Logone was truly exposed. Historically, the city first developed in the downtown centre in the western part of the current city. However, during the dry decades of the 1970s and 1980s, the city was urbanized in the flood-prone areas of the Chari and Logone confluence, making neighbourhoods extremely vulnerable. These areas are currently home to approximately 300,000 inhabitants and are among the poorest 34.
Due to inevitable climate change and variability, long-term policies appear to be the only way to reduce the exposure of infrastructure and people to flood-prone areas. Based on our study and the models we developed, we recommend stepping up the monitoring of the basin. Our models can be used to improve river flood forecasting and to build early warning systems, helping for preventing floods for instance at N’Djamena city. We know that it takes three weeks for water to be transferred from the south of the basin, e.g. Sarh, to N’Djamena 24. By applying the MGB model, this forecast could be reinforced. Moreover, combining with the spatialisation of land use, this could provide an interesting tool for considering the siting of hydraulic infrastructures in view of the water management. This could help for considering their best location and sizing according to the areas that could be flooded regularly and the areas that could be preserved for instance for cropping.
Otherwise, from a more global perspective, unlike central Sahel, it is expected that the continent’s Mediterranean margin could be affected by severe droughts in the next 30 years 35. The estimate of water stress for 2040 shows that central Sahel will become a region with excess water. Thus, water management in this basin is challenging because it could become an attractive area for climate migrants.
