Article Highlights

  • Agricultural, meteorological, and hydrological droughts are related phenomena, causing socioeconomic and environmental problems.

  • There is a complex array of interplaying factors: both El Niño Southern Oscillation and global Sea Surface Temperatures are inducing factors.

  • Meteorological drought is inferred from observed or reanalysis precipitation; some indicators use observed temperature or soil moisture.

  • Remote sensing techniques were developed in the last decades, often combined with in situ data, to identify, classify, and monitor drought.

1 Introduction

West African rainfall fluctuations have worldwide implications and, consequently, there is a growing interest in research on hydrometeorological extreme events such as heavy precipitations, flooding, heat, droughts, and the significant socio-economic and environmental problems they cause (Masih et al. 2014; Quenum et al. 2019; Puczko and Jekatierynczuk-Rudczyk 2020). These authors recognized that the first cause of drought is natural climatic variability and rainfall scarcity, and they addressed various drought-related issues in Africa. Their study is specific and focuses on drought in West Africa because available records indicate that this hydrometeorological extreme event has become more frequent, intense, and widespread over the past 50 years (Masih et al. 2014). These papers present a comprehensive account of droughts that occurred during 1900–2013, including evidence available from past centuries. Indeed, several indications, as well as predictions from climatic models, have shown aggravation of droughts in West Africa caused by rising global temperature affecting the spatial pattern of rainfall and leading to extensive land use changes, causing a switch from forests, savannas, and small-holder agriculture to large-scale farming that has intensified in response to the rise in world food prices (Grippa et al. 2011; Jung et al. 2019; Rulli et al. 2013; Quenum et al. 2019). Additionally, between 1970 and 2009, 23 of the 25 most expensive hydrometeorological disasters for insurance companies worldwide were climate-related, especially droughts (Décamps 2010). Trends in aridity and drought are projected to increase during the twenty-first century.

Given that drought is a complex and recurrent hydrometeorological phenomenon across Africa, it is paramount to investigate it as it varies in its onset, intensity, duration, and geographical coverage. Moreover, Quenum et al. (2019) studied how rising global temperature will affect the spatial pattern of rainfall and consequently drought in West Africa; employing indicators for four global warming levels. Five main types of droughts closely related to natural and temporary phenomena are generally identified, namely agricultural drought, meteorological drought, hydrological drought, geotechnical drought, and socio-economic drought. What is common is that all of these results from a decrease in the amounts of rainfall below the usual average causing water shortage (Décamps 2010; Masih et al. 2014; N’Da 2016; République du Togo 2018; Puczko and Jekatierynczuk-Rudczyk 2020).

Based on the propensity of droughts to create environmental and economic disasters its analysis merits the understanding of the following questions: What is a drought? What are its causes and consequences? To what degree could remote sensing data be used to analyze this climatic event in West Africa? What is the impact of drought on a hydro-system? To identify potential regions at risk, this research will review the causes of recurring droughts and their impacts (seasonal climate forecasts and short-term weather forecasts) in West Africa. To achieve this, the study will not only evaluate the impact of drought at a regional level using remote sensing, but it will also, as an example, examine water resource scarcity in the hydro-system of the Bandama River in Côte d’Ivoire.

2 Overview of Drought

In Sub-Saharan West Africa, drought frequently occurs and causes large economic, social, and environmental problems. This area is highly vulnerable to single and multi-year droughts that create disruptive consequences including water shortage for agriculture, social conflicts over resources, blackouts, and energy rationing (Rulli et al. 2013; Quesada-Montano et al. 2019). Drought affects the ecosystem, populations, soil moisture, agriculture, groundwater aquifers, rivers, lakes, and surface water reservoirs (Grippa et al. 2011; République du Togo 2018; Jung et al. 2019; Quenum et al. 2019; Alsumaiei 2020; Puczko and Jekatierynczuk-Rudczyk 2020). It causes severe socio-economic and environmental disasters in many regions, resulting in massive-scale migration, famine, and environmental degradation.

2.1 Definition and Typology of Drought

Drought has several meanings that are closely related to the subject matter. Indeed, the definitions are adjusted from water resource managers to farmers or hydropower plant operators as well as biologists specializing in wild fauna and flora. As such, several types ranging from meteorological to agricultural droughts, hydrological to geotechnical droughts, and even socio-economic droughts (Décamps 2010; République du Togo 2018; Puczko and Jekatierynczuk-Rudczyk 2020). Meteorological drought is a natural phenomenon that has multiple causes varying from one region to another. It is generally described by an annual or seasonal deficit of precipitation totals that are lower than those generally recorded during the same period. It can also refer to a lack of precipitation over time. Meteorological drought is defined by a threshold corresponding to a certain deficit of precipitation over a determined period. While Agricultural drought relates to a deficit in the quantity of water in the soil such that it can alter the growth of plants, hydrological drought occurs when lakes, rivers, or groundwater show abnormally low levels. It also refers to the deviation of the supply of surface and groundwaters from certain average conditions at different times. Geotechnical drought is a period where rainfall deficits cause a decrease in the water content of the subsoil. Finally, the socio-economic drought reflects the relationship between supply and demand for commodities or economic goods such as water, livestock fodder, and hydroelectric power.

2.2 Causes of Droughts

Droughts are a part of natural climatic variability, as discussed, e.g., in Masih et al. (2014) and Quenum et al. (2019). These works review several potential natural and anthropogenic causes of droughts, to reveal that there is no unique set of factors responsible for the geospatial and temporal variation of droughts across Africa. There is rather a complex array of interplaying factors. Indeed, the physical mechanisms that are causing droughts are different from one region to another, even though both El Niño Southern Oscillation (ENSO), global Sea Surface Temperatures (SSTs), and warming of tropical Oceans are perceived as major inducing factors, as was concluded in for the Sahelian zone. For instance, droughts in the southern part of Africa have mainly occurred during the warm phase (El Niño) of ENSO in addition to several local and global factors influencing the drought phenomenon. These authors further argued that on contrary, East African regions face droughts during the cold phase (La Niña) of ENSO. They also noticed that a strong La Niña event was the main cause of the 2010–2011 drought in the Horn of Africa, while an array of complex processes was responsible for drought events in the West African Sahel.

It should also be noted that, in West Africa, low precipitation during the summer is the result of cool waters (Atlantic Ocean) north of the equator and warm waters south of the equator. El Niño events boost vertical thermal stability from a warming troposphere combined with the deterioration of vegetation cover, which increases albedo and decreases evapotranspiration (Masih et al. 2014; Quenum et al. 2019). Other factors that contribute to the West African droughts include the land–atmosphere feedback through natural vegetation and land cover change, the anthropogenic contribution in land use change altering the land surface feedback mechanisms, and at some point, the human-induced greenhouse gas emission. Nonetheless, there is a general understanding that additional research is required to better understand drought mechanisms in Northwest Africa because neither SST nor ENSO shows a clear relationship with drought patterns (Masih et al. 2014).

The African Centre of Meteorological Applications for Development (ACMAD 1998) and République du Togo (2018) revealed that in recent years there has been a general decrease in rainfall as well as in the number of rainy days in Togo. The index of aridity, which is the ratio of rainfall to potential evapotranspiration (P/ETP), has also been decreasing. This situation reflects a trend toward aridification of the climate. Moreover, climate data showed that between 1961 and 2012, major climatic risks are yielding to extreme drought or shockingly flooding. Thus, these opposing extreme events following one another have led to overall confusion for farming communities (UNDP 2011; République du Togo 2018).

2.3 Impacts of Droughts

This part of the study focuses on identifying the direct and immediate consequences of drought. These impacts include socio-economic and environmental effects as well as tracking of secondary and longer-term consequences.

2.3.1 Socio-economic Impacts of Drought

Drought is a natural hazard that affects directly and indirectly many people. Unfortunately, Africa remains the continent with the most negative impacts of this phenomenon. Between 1970 and 2012, this severe hydrometeorological extreme event caused nearly 680,000 deaths in Africa (WMO 2014). Drought is for instance responsible for food shortage (famine) because of its damaging impacts on food production. Furthermore, it affects public health with issues such as heat stroke, anxiety, depression, and death. Moreover, it especially worsens epidemics like malaria. Additionally, the bearings on societies are seen in malnourished children, loss of life, and economic loss. For instance, in Kenya, the drought that occurred from 1999 to 2001, cost around 2.5 billion US dollars to the economy (Vicente-Serrano et al. 2010). The effects on human beings can be observed through the massive displacement of the population, known as climatic migration.

It is important to note that successive hydrometeorological extreme events often have dramatic effects on the balance of the agricultural economy, with significant impacts on farming activities, including lower crop production and yields, loss of livestock, reduced incomes, increased rural exodus, food insecurity, and even famine. For instance, major droughts experienced by Togo have led to declines in agricultural production and famines during the periods 1942–1943, 1976–1977, and 1982–1983. The associated damages affected more than 40,000 hectares, which corresponds to an economic loss estimated at more than 8 billion CFA Franc (Gross Domestic Product (GDP) in Togo was worth 5151 billion CFA Franc in 2021 (World Bank 2020), with a predominant decline in yields of corn and sorghum (FAO 2018). The yield drops an average of 60 to 80% for cereals and 20 to 80% for legumes, according to assessments from the Togolese government mission (République du Togo 2018). Furthermore, shrinkage by drying clayey soils can happen during a pronounced geotechnical drought and/or lasting deformations of the soil surface (differential settlements). This shrinkage can be followed by swelling phenomena as the initial hydrogeological conditions are re-established or more rarely by creep phenomena with softening. The phenomenon of shrinkage-swelling of clays regularly causes considerable damage to buildings in France. The loss was estimated to be 60 million euros per department between 1989 and 1998. Indeed, because of their superficial foundations, individual houses are particularly vulnerable to this phenomenon. More than 4 million houses are potentially exposed (Decamps et al. 2010). Additionally, the indirect effects of drought on welfare can be assessed through impacts on crop and livestock prices that may be higher than the direct production effects of drought. This hydrometeorological extreme event increases structural problems with direct or indirect costs in Africa. In the past decades, drought has caused decreases in crop yields, unemployment, and impoverishment. Moreover, it has increased famine and social conflicts in Africa.

2.3.2 Environmental Impacts of Drought

There are several consequences of drought on the environment. This hydrometeorological extreme event negatively impacts ecosystem functions and increases vulnerability to other natural disasters such as heat waves. Drought exacerbates savanna and forest fires by drying up biomass more likely to burn. However, the extent of deforestation in West Africa is still under debate, and regional greening has been observed. Using high-resolution satellite imagery and deep learning, a surprisingly high density of trees has been discovered in the Sahara and the Sahel; but it is not classified as forests (Brandt et al. 2020). Both Sahara and monsoonal shifts have been suggested as policy-relevant tipping elements in the climate system. There is a strong coupling between soil moisture, vegetation, and rainfall (Taylor et al. 2011a). Trends are mainly driven by changes in the ocean, they are enhanced by land–atmosphere feedback. Several authors hypothesize that prevailing droughts were partly caused by land cover change, and satellite data reveal a correlation between droughts, vegetation cover, and land clearing in the Sahel. However, when simulating the impact of land cover change, climate model uncertainty plays a role (Wang et al. 2015). In addition, the effects on the ecosystems are drought-related environmental damage and irreversible degradation (Vicente-Serrano et al. 2010).

The consequences on water resources are the drying out of rivers and lakes and the depletion of soil moisture levels. The edaphic drought caused by the meteorological drought affects vegetation by causing a reduction in soil moisture and oxygen (Grippa et al. 2011; République du Togo 2018). Indeed, hydrological drought affects rivers and continental water bodies (ponds, lakes, etc.) whose level or flow shows a lower value than the one usually measured at the same period. The rivers and water bodies may dry up temporarily, partially, or totally, depending on the intensity and duration of the drought. To the latter, it can be added phreatic drought, which affects the level of aquifers and generally results in a serious water shortage forcing populations to travel long distances to obtain water (République du Togo 2018).

At larger spatial scales, Forootan et al. (2014) analyzed relations between observed water storage changes and rainfall and sea surface temperature data over the Atlantic, Pacific, and Indian Oceans. They found that major teleconnections realize a strong link that would allow predicting terrestrial water better as compared to hydrological modeling. Moreover, Grippa et al. (2011), as well as Hassan and Jin (2016), compared water storage data to soil moisture estimations, with a particular focus on the Sahelian area. They found generally good agreement in storage variability (over a yet relatively short-time frame); in particular, their results point to processes whose representation needs improvements in the models (e.g., dry season evapotranspiration). Furthermore, land degradation is another important impact of drought. A study conducted on the state of land degradation in northern Togo revealed different forms of land erosion. These include surface material erosion, which has completely stripped the topsoil. Thus, quartz blocks and armored plateaus are exposed. These sites with indurated tropical ferruginous soils with concretions have become uncultivated, which is unsuitable for agriculture (Grippa et al. 2011; Republic of Togo 2018).

3 West African Drought

This part presents and discusses drought in West Africa as observed from space (e.g., Forootan et al. 2014). Then it assesses rainfall and evapotranspiration between 1970 and 2013 as indicators of the risk of water resources scarcity in the hydro-system of the Bandama river in Côte d'Ivoire (Hassan and Jin 2016; N’Da 2016). This aims to evaluate drought at a hydro-system scale to better understand the causes of water storage increase and decline for major African river basins.

3.1 Drought as Observed from Space

Meteorological drought is typically inferred from observed or reanalysis precipitation, though some indicators use observed temperature or soil moisture (Grippa et al. 2011; Forootan et al. 2014; Masih et al. 2014; Gerdener et al. 2022). To derive these variables, several remote sensing techniques were developed in the last decades, often combined with in situ data (Zaitchik et al. 2008; Eicker et al. 2014; Li et al. 2019). Satellite radar and microwave sensors provide daily and/or monthly datasets with a spatial resolution of down to a few 10 km. Meteorological reanalysis data are more complete than those of remote sensing data; but often display biases, while in situ observation networks are relatively sparse in most parts of Africa. Nevertheless, to identify, classify, and monitor meteorological drought with observed variables, several indicators were developed to describe the intensity, severity, and duration of a drought event. Among other indices, the Standardized Precipitation Index (SPI, McKee et al. 1993), Standardized Precipitation Evapotranspiration Index (SPEI, Vicente-Serrano et al. 2010), Palmer Drought Severity Index (PDSI, Palmer 1965), Weighted Anomaly Standardized Precipitation Index (WASP, Lyon 2004), and Effective Drought Index (EDI, Byun and Wilhite 1999) can be used for drought detection; however, each indicator has its advantages and disadvantages. While some can flexibly choose time scales (e.g., SPI, SPEI, EDI, and WASP), other indicators account not only for precipitation but also for other variables of the hydrological water cycle, like evapotranspiration (e.g., SPEI, PDSI). However, the complexity of computation can be very different. Furthermore, it is known that not all indicators perform equally well for different climate regimes. For example, the WASP is known to have difficulties in detecting drought events in desert areas (Lyon 2004) and several studies showed that the SPI might underestimate the drought intensity in different climate regions when compared to the SPEI (Naresh Kumar et al. 2009; Li et al. 2020; Pei et al. 2020).

In this review, for showcasing the detection of meteorological drought, the SPEI was chosen as described in Vicente-Serrano et al. (2010). This indicator is not only based on precipitation but also considers temperature (global ERA5 reanalysis data were used) to determine potential evapotranspiration. ERA5 provides detailed records of the global atmosphere, land surface, and ocean variables from 1950 onwards; it was developed by the European Centre for Medium-Range Weather Forecasts (ECMWF, Copernicus Climate Change Service 2017; Hersbach et al. 2020). The reanalysis data is produced by using 4D-Var data assimilation on the Integrated Forecasting System (IFS) and it provides more quality and level of detail compared to its predecessor ERA-Interim (Hersbach et al. 2019). SPEI is one of the most widely used meteorological drought indicators and its computation (see “Appendix”) is simple, since gridded remote sensing data of precipitation and temperature are widely distributed, and several data sets exist on a global scale. The use of temperature covers the quantification of evaporative demand, which in times of climate change becomes more and more important (Ajayi and Ilori 2020). This indicator allows the analyst to select a time scale of relevance and it has been used to quantify droughts in many studies (e.g., Meza 2013; Chen and Sun 2015). It can be computed for different precipitation accumulation periods; here six months were selected and that allows to effectively filter out shorter dry spells. Gerdener et al. (2020) showed that short-term fluctuations can disrupt the detection of drought, an unwanted feature, which was mitigated by choosing a longer accumulation period. Indeed, 2011 was a year of severe drought in West Africa. Many countries including Burkina Faso, Mali, Mauritania, Senegal, and Niger were affected (EM-DAT 2021). Figure 1 shows the SPEI 6 (6 for a six-month accumulation period) in the western part of Africa from April to December 2011.

Fig. 1
figure 1

Monthly SPEI6 in West Africa from April (top left) to December (bottom right) 2011. SPEI6 shows drought severity classes ranging from none to extreme drought

Full size image

At the beginning of that year, there was not an identified drought (not displayed) but starting in April, it increased to October; thereafter the intensity slightly decreased. Drought impacted large portions of West Africa but the most affected were Mali and the Niger basin, one of the largest basins in West Africa.

In addition, Fig. 2 shows a closer look at affected parts of the Niger basin as noticed by a specific SPEI6 drought severity class (e.g., extreme drought). Indeed, up to 80% of this basin was touched in 2011; and even 10% were identified as extreme drought. However, other phases of extreme drought in the Niger basin were identified with the SPEI6 (and confirmed by the EM-DAT database) in 2005, 2009, and 2015.

Fig. 2
figure 2

Area of the Niger basin (%) that is affected by one of the four SPEI6 drought severity classes (extreme, severe, moderate, or mild) shown from 2003 to 2018

Full size image

It is important to note that apart from data scarcity (observation networks have deteriorated but have seen some refurbishment via, e.g., the African Monsoon Multidisciplinary Analysis—AMMA and the West African Science Service Centre on Climate Change and Adapted Land Use—WASCAL programs), modeling is challenging since water and energy cycles are sensitive to small-scale soil moisture gradients (Xue et al. 2012). The initiation of mesoscale convective systems can be caused by gradients over a few tens of kilometers (Taylor et al. 2011b). Intercomparisons have revealed that models have problems representing teleconnections and atmospheric bridges (linking SST across different oceans via sea-air and air-sea coupling). Regional climate models predict an increase in precipitation, while global climate models do not. There is no consensus on whether land degradation in the south affects precipitation in the Sahel/Sudan via water recycling; one hypothesis is that there is always sufficient moisture and the trigger controls rainfall, while the competing hypothesis assumes that degradation over the last decades has been so severe that it reduces moisture convergence in the Sahel. The observed mean and extreme precipitation trends are inconsistent with each other (Lausier and Jain 2018). Sylla et al. (2018) suggest that river basins will face severe freshwater shortages; they predict a decline in irrigation potential for 2 °C global warming with accompanying demands for crop water and irrigation. Mamalakis et al. (2021) address future zonally varying responses in the position of the intertropical convergence zone (ITCZ) to climate change with potentially dramatic consequences for the livelihoods and food security of billions of people. However, they point to large uncertainties in climate model simulations.

In contrast, the detection of hydrological drought is usually based on observable surface water like gauged streamflow or reservoirs (Tabari et al. 2013; Gusyev et al. 2015; Jahangir and Yarahmadi 2020). However, gauge data is often not accessible. Furthermore, drought can be defined via subsurface (groundwater) availability; but in practice this is rare. Indeed, well data is sparse, and irregularly distributed in space and time, as well as political and other reasons often limit access. Thus, it is generally difficult to derive temporally and spatially consistent observations of hydrological variables. A new opportunity was provided by the satellite mission Gravity Recovery And Climate Experiment (GRACE), which observed, for the first time, the temporal variations of the total water storage remotely from space with global resolution. GRACE was managed by NASA and German Aerospace Center (DLR). It globally observed the whole surface and subsurface water changes concerning a long-term mean, which is commonly termed Total Water Storage Anomalies (TWSA, Wahr et al. 1998). The GRACE mission started in April 2002 and ended in June 2017. It was succeeded by GRACE-Follow On (FO) in May 2018.

In the last decades, TWSA data derived from GRACE and GRACE-FO have strongly contributed to our knowledge of water storage variability in West Africa, including drought and flood quantification, and they were used also to study other climate variables. Thus, Scanlon et al. (2022) combined GRACE TWSA with reservoir levels and MODIS imagery for West Africa, and they found rising TWS trends in aquifers, attributed to increased recharge from land use change and cropland expansion. Oussou et al. (2022) studied hydrogeological regimes in fractured aquifers from multi-satellite data over Benin, including GRACE water storage maps. Furthermore, GRACE data were used to identify shortcomings in model process representation. For example, Grippa et al. (2011) compared GRACE-derived soil moisture with nine land surface models. Hassan and Jin (2016) compared TWSA from three different GRACE solutions to the Global Land Data Assimilation System (GLDAS) and the Tropical Rainfall Measuring Mission (TRMM) rainfall data at seasonal and secular timescales, to understand the causes of water storage increase and decline for major African river basins. However, GRACE and GRACE-FO had and still have limitations; the most important one being their limited spatial resolution of about 300 km because of the satellite orbital configuration. Additionally, the temporal coverage is restricted to the lifetime of the missions, e.g., the time series contains a gap between GRACE and GRACE-FO (July 2017 to May 2018.)

To overcome these limitations, some research groups developed downscaled TWSA data products through data assimilation techniques (Zaitchik et al. 2008; Forman et al. 2012; Tangdamrongsub et al. 2015). The study of Zaitchick et al. (2008) was the first one to consider GRACE assimilation within a land surface model; they found an increased correlation between hydrometeorological variables after assimilation. This study was followed by many others, who further developed different assimilation schemes on different spatial scales, and who developed validation approaches. To name a few, Li et al. (2019) provide a critical assessment of GRACE TWSA assimilation into the Catchments Land Surface Model CLSM for groundwater monitoring and compared simulations to in situ data from about 4000 wells. Jung et al. (2019) found improved soil moisture estimates in West Africa when comparing their assimilation results against satellite data records from the ASCAT (Advanced Scatterometer), SMOS (Soil Moisture and Ocean Salinity), and SMAP (Soil Moisture Active Passive) instruments. At the University of Bonn, the first assimilation framework was developed and that enables the integration of gridded GRACE and GRACE-FO TWSA maps (Eicker et al. 2014; Schumacher et al. 2016) into the WaterGAP global Hydrological Model (WGHM 2.2d/e, Schmied et al. 2021). This approach uses the Ensemble Kalman Filter (Evensen 1994, Nerger et al. 2007) and it accounts for the complex GRACE error structure. Meanwhile, this originally regional approach has been developed into a global data assimilation system. This framework enables spatial downscaling and vertical disaggregation of the GRACE-FO data while improving the model’s realism, and is one of only a few worldwide able to globally assimilate GRACE. Note that it is out of the scope of this manuscript to provide a full description; however, information can be found in Gerdener et al. (2020) and Gerdener et al. (2022), and the GRACE TWSA data used here can be downloaded from https://www.apmg.uni-bonn.de/daten-und-modelle/grace_level3_monthly_solutions (last access 2022-09-07).

Quantification of water deficit, which is required for indicators of hydrological drought, was derived either as the sum of all storages, i.e., TWSA, or as below separately for surface water, soil moisture, and groundwater storage. Several GRACE-based hydrological drought indicators, such as the Drought Severity Index (DSI, Zhao et al. 2017) and the Drought Index (DI, Houborg et al. 2012) were developed. Gerdener et al. (2020) showed how drought visible in GRACE data propagates maps into indicators, and they also explained in detail how signals contained in GRACE data can bias drought detection, e.g., noise or linear trends (see also Humphrey et al. 2016; Hosseini-Moghari et al. 2020). For example, some drought indicators, such as the DI, are very sensitive to typical GRACE noise in the time series and they will identify artifact drought months. It is possible to mitigate noise in input TWSA time series by using—similar to meteorological indicators—hydrological indicators that accumulate deficit over time, as the DSI with accumulation (DSIA, Gerdener et al. 2020) does. However, this has also limitations, as the drought period might be detected later than it occurred. Real linear trends in GRACE TWSA, e.g., due to lowering or rising groundwater levels, can lead drought indicators to erroneously detect a negative phase, as was found for example for the DI, the DSI, or an indicator provided by. The DSI when modified to use temporal storage differences (DSID, Gerdener et al. 2020) and the Modified Total Water Storage Deficit Index (MTWSDI, Hosseini-Moghari 2020) where the trend must be removed in advance, seem to be less affected by trends. However, due to time-differencing, the DSID is prone to amplify GRACE noise, and removing the trend in advance may also bias drought water deficits that appear actually at the beginning or at the end of the time series. On balance, there is not yet an agreed ‘best’ hydrological indicator to be applied with observed water storage variability, and in this article, the DSI (see “Appendix”) will simply be used for analyzing drought events in the last two decades, based on downscaled storage information derived from the global assimilation of GRACE data into the WGHM model.

Figure 3 shows the drought severity index DSI from April to December 2011 for West Africa, which was computed based on GRACE TWSA data as discussed above. Corresponding to what was shown with the meteorological drought indicators, it was noticed an increase in drought severity during the year. The peak of the meteorological drought was mainly identified in September, while the hydrological drought was most severe in December for large parts of West Africa along the Gulf of Guinea coast. The precipitation deficit (Figs. 1, 2) is extending to the water storage, which emphasizes the need for a deeper look into the different water storage.

Fig. 3
figure 3

Monthly DSI in West Africa from April 2011 (04/2011, top left) to December 2011 (12/2012, bottom right). The DSI is showing drought severity classes ranging from no (none) to exceptional drought and was computed using assimilation-based TWSA

Full size image

Moreover, Fig. 4 shows the area of the Niger basin in percentage that is identified in the DSI drought classes, e.g., affected by extreme drought. The exact delineation of the Niger basin considered in this study is shown in Fig. 5. Due to the assimilation into a hydrological model, it is possible to distinguish between surface water (top), soil moisture (middle), and groundwater-DSI (bottom). As discussed above, drought detection with the DSI might be affected by linear trends. Here, none of the drought indices for the individual storage compartments show a strong trend because drought periods are not visible at the beginning or the end of the time series. Thus, the DSI for all storage is not overly affected by trends. Furthermore, it was found that the 2011 drought was visible in all three water storages, but the highest drought severity and duration were identified in the groundwater storage. About 40% of the area of the Niger basin was touched by groundwater drought. This lasted from mid-2011 to mid-2012. Then, in 2012, it was classified about 10% of the area was exceptionally dry in the groundwater. For surface water deficit, it can be noted a similar drought period and affected area but with a slightly lower drought severity, where only at most 5% of the area is touched by exceptional drought. For soil moisture, the drought started in early 2011 and lasted until the end of 2011. This means that the occurrence of the drought was first affecting the soil moisture and later appeared in surface water and groundwater, thus, a shift of three to four months between soil moisture and groundwater/surface water is identified. This is, in general, an expected result that hydrological models can simulate, provided their representation of soil properties, river networks, reservoirs, and other processes is correct. It is important to mention that Gerdener et al. (2022) in their study of South Africa found significant differences between the observed and modeled temporal rainfall-storage-vegetation dynamics. When looking at the start of the drought in 2011 in SPEI6 (precipitation), it is possible to confirm this sequence of precipitation deficit; in other words, meteorological drought affects water storage with a temporal delay: the main start of the drought period found with SPEI6 was detected at the beginning of 2011, contributing to soil moisture drought in early 2011 and again contributing to surface water and groundwater drought mid-2011.

Fig. 4
figure 4

Area of the Niger basin (%) that is affected by one of the five DSI drought severity classes (Exceptional, extreme, severe, moderate, or near-normal) shown from 2003 to 2018. The DSI was computed using an assimilation-based (Eicker et al. 2014; Schumacher et al. 2016) surface water (top), soil moisture (middle), and groundwater (bottom)

Full size image
Fig. 5
figure 5

Polygon of the Niger basin as considered in this study

Full size image

Further, Meza et al. (2020) combined hydrological drought indicators from global hydrological modeling with remote sensing, crop modeling, and vulnerability indicators to derive maps of drought risk, separated for irrigated and rainfed cropping systems, and as a combined risk indicator for countries where both irrigated and rainfed agriculture prevail. Indeed, for many West African countries with a gradient in climatic conditions, irrigated crops are cultivated in more arid areas, while rainfed crops are predominantly grown in the more humid regions. Still, vulnerability for irrigated cropping is generally higher in Sub-Saharan Africa. As a result, drought risk is higher for irrigated systems and lower for rainfed cropping in Burkina Faso and Côte d’Ivoire (Meza et al. 2020). This global assessment reported that seven out of ten countries with the highest overall drought risks are located on the African continent.

3.2 Hydroclimatic Droughts in the Hydro-system of the Bandama Watershed: Indicators of Water Resources Shortage Risks in Côte d’Ivoire

The IPCC (2014) noticed that despite uncertainties and controversies surrounding climate change, global warming of the Earth as well as increased climate variability and extreme hydrometeorological events such as droughts, heat waves, cyclones, thunderstorms, etc. are visible and conclusive. Thus, the intensification of rainfall variability is linked to components of the water cycle such as runoff, infiltration, water table, evapotranspiration, etc. Subsequently, several factors have contributed to the mutation into a hydroclimatic change (Noufé 2011; N’Da 2016). Indeed, these changes, which have been established in the intertropical zone, are exposed by the rising temperatures and disruption of rainfall patterns and scarcity (Servat et al. 1997; Hubert et al. 1998; Sarr 2008; Dibi Kangah 2010; Lausier and Jain 2018; Sylla et al. 2018).

With a focus on this climate change issue, this study aims to present indicators of risks of water resource shortages in the Bandama watershed. The objective is to evaluate, within a case study, the balance between water input (rainfall) and water loss (evaporation) as well as reveal the difficulties faced by the population to access water. It is important to note that this area was chosen because it is the largest watershed in Côte d’Ivoire that encompasses several major towns, including Bocanda, Dimbokro, Bouaké, and Korhogo (Fig. 6). These areas have faced unprecedented shortages of running water for the past three decades (M’Bra 2013; Dibi Kangah 2010).

Fig. 6
figure 6

(Source: Dibi Kangah et al. 2016)

Homogeneous rainfall zones in the Bandama watershed, Côte d’Ivoire

Full size image

3.2.1 Data used and Applied Methods

Based on a hydroclimatic approach, three types of data were collected, including rainfall, potential evapotranspiration (PET), and decadal useful soil reserves (USR). The first two records were gathered at the national weather service (Direction de la Météorologie Nationale-DMN) of Côte d'Ivoire and covered the period from 1970 to 2015. The third parameter (USR) was derived from a review of related literature (Perraud 1979; Dibi Kangah 2010; Noufé 2011; N’Da 2016). The USR corresponds to the amount of water that the soil can absorb and/or return. Thus, three levels of USR were used based on decreasing regional gradients in this watershed. These ranks are 100 mm, 60 mm, and 30 mm corresponding respectively to the southern, central, and northern zones of the study area.

Moreover, socio-anthropological data were gathered through individual and group interviews via survey sheets (Dibi Kangah et al. 2016). It is important to stress that the questions focused mainly on the perception of rainfall variability, difficulties in accessing drinking water, and the causes and reactions of rural populations. Data were obtained from interviewing 390 individuals, which were selected from a non-probability formula. The surveys were conducted in six villages, including Ahua and Krokokro in Dimbokro (Zone III), Kongodékro, and Sessenouan in Bouaké (Zone II), Waraniéné, and Torgokaha in Korhogo (Zone I). These investigation sites were chosen nearby synoptic stations where climatic records are complete and reliable.

The hydroclimatic approach builds upon various processing methods (computation of the hydrological drought balance and index, meteorological drought indices) to characterize the risks of soil water resource shortages in the Bandama hydro-system. The analyses started with the computation of rainfall indices (McKee et al. 1993), which determine the intensity of droughts, rainfall deficits, and excesses on monthly, seasonal, and annual time series. These indices are deviations from the mean. Indeed, cumulative analysis of the indices reveals broad trends and removes small fluctuations. Thus, when the cumulative indices increase, the trend is wet and vice versa (Nezzal and Iftini-Belaid 2013). Also, a drought occurs when the SPI is consecutively negative, and its value reaches an intensity of − 1 or less and ends when the SPI becomes positive (Dibi Kangah 2010; Noufé 2011; N’Da 2016).

It is important to note that the methods used to assess indicators of hydrological drought are the water balance or soil water reserve (SWR) and the hydrological drought index (HDI). In general, a soil water reserve at the end of a particular period (SWRi) is equal to the water reserve at the beginning of that period (SWRi-1), which is increased by rainfall (Ri) and decreased by water lost by real evapotranspiration (RETi) and by the rapid drainage (RDi) during the period studied (Noufé 2011; N’Da 2016). Drainage occurs when, at the end of the period (SWRi), the cumulative rainfall Ri and the water reserve at the beginning of the period (SWRi-1), from which the RETi has been deducted, is greater than the value of the useful reserve at the end of the period (USRi). Further, the Hydrological Drought Index (HDI) indicates how rainfall relates to RET. It also reflects the fraction of rainfall (R) that is evapo-transpired. It is obtained by dividing the cumulative rainfall (Rcum) by the cumulative actual evapotranspiration (RETcum) on a time scale of a month; it is expressed in percentage. It is a measure of the amount of water that remains available for deep percolation or surface runoff. Therefore, a HDI lesser than 100% indicates that the RET was higher than rainfall during the month; accordingly, the stored water (in biomass, soil, and surface water) was depleted. A HDI greater than 100% means that rainfall was higher than RET during the month; thus, there was deep runoff or percolation (Noufé 2011; N’Da 2016).

3.2.2 Meteorological Droughts as Indicators of Risk of Water Scarcity

Interannual Trend in Rainfall Analysis of the cumulative deviation from mean curves of the rainfall series shows alternative wet (excess) and dry (deficit) years from 1970 to 2013. Figure 7 shows trends toward drought or wetness. In Korhogo (Zone I in northern Côte d’Ivoire), the period 1971–1974 registered a humid trend. It is followed by consecutive years of droughts, notably in 1975–1977 and 1981–1983. Further, 1998–2013 was marked by a relapse in rainfall. Overall, the years remained continuously dry and affected this region. It should also be noted that localities in the north, such as Ferkessédougou, have experienced intense drought since 1970.

Fig. 7
figure 7

Cumulative gaps from the mean of rainfall series from 1970 to 2013

Full size image

Furthermore, in the equatorial transition zone, which includes Bouaké (Zone II) and Dimbokro (Zone III), there are two important trends. The first was a long period of drought, which extends from 1970 to 1999 in Bouaké and from 1970 to 2001 in Dimbokro. The second phase was characterized by a humid tendency starting in 2000 for Bouaké and 2001 for Dimbokro. However, in Grand-Lahou (Zone IV), on the coast, rainfall showed four trends. The first period (1970 to 1980) was humid, the second (1980 to 1993) was dry, and the third was marked by a positive recovery from 1993 to 1997, followed by a drop in rainfall from 1998 to 2013 (Fig. 7).

Intra-annual Change in Rainfall Water balances, before and after a change in rainfall trend, were performed to estimate the impact of rainfall variability on soil water resources. Figure 8 show some stability in soil water reserves, deficits, and excesses in the Bandama watershed. On the coast (Grand-Lahou), the most deficient period is observed between December and March. Between May and July, the soil water excess (or “effective infiltration”) is very high, around 100 mm in May and July, and 400 mm in June. In Dimbokro, the water deficit is a month longer (November) than that of the coast. In these two hydroclimatic subregions, October is slightly an excess. In Bouaké and Korhogo, the excess period runs from June to October between 1999 and 2013; it was shorter (August-October) and less significant in the period before (1971–1998). The critical months are from December to March.

Fig. 8
figure 8

Monthly water balances before and after a break

Full size image

3.2.3 Changes in Soil Water Reserves as Indicators of Water Scarcity

Consistent with previous results as well as satellite observations that showed dry trends after 2000, the average hydrological dryness indices (HDI) were determined for each station before and after that date. This method allows for observing the evolution of soil water reserves, which are important indicators of water scarcity (Fig. 9).

Fig. 9
figure 9

Monthly water balances before and after the break

Full size image

Figure 9 shows that the period of high risk of shortage is between November and March. However, beyond the 8th degree north, this period extends into April. Before the year 2000, only the period November–February had this degree of risk below this parallel. But after this date, the month of March becomes critical in many southern regions of Côte d’Ivoire. Moreover, the localities in the south of the country are less stable than those in the north. Indeed, the poor distribution of precipitated water does not ensure the continuous availability of water in the soil over a long period. Faced with numerous dry spells and reduced rainfall, the water table sinks. Above all, when the dry seasons become longer, the water resources (ponds, lakes, rivers, wells) become scarcer.

3.2.4 Experiences of Rural Societies on Water Shortages as Indicators of Drought

Of the six villages surveyed, only Ahua in Dimbokro and Kongodékro in Bouaké have access to water from the Ivorian water company, the Société de Distribution d'Eau en Côte d'Ivoire (SODECI), or piped water (Table 1). More than 75% (n = 65/village) of farmers use it. While very few respondents in Ahua add this water to other sources, many in Kongodékro use the pump (20%), wells (44.6%), and river.

Table 1 Consumers’ water resources (%)
Full size table

In Krokokro (Dimbokro) and Séssénouan (Bouaké), since they do not have access to water from SODECI, more than 95% of households use village pumps as a priority. In contrast to the Baoulé villages, more than 90% of farmers in the Sénoufo localities of Waraniéné and Torgokaha prefer to use well water rather than pump water. About 50% of them use both sources of water simultaneously.

The impact of climatic variation on soil water resources is strongly felt by villagers. The farmers’ testimonies are proving to be good indicators of rainfall variability and its effects on water resources (Table 2).

Table 2 Water scarcity and social impacts in rural areas (%)
Full size table

Table 2 illustrates this well. Indeed, lack of water is felt more through stress and drying up of plants (more than 75%), wells, surface water, rivers, and village pumps. Thus, from Kongodékro to Waraniéné, over 50% of respondents mentioned the drying up of wells and rivers. The two villages served by SODECI also experience water shortages but to a lesser extent. To reduce their vulnerability, community leaders are asking the administration (Ivorian government) to increase the number of village pumps (from two to four) according to the population. Households, on the other hand, are forced to over-create wells. The soil water reserve is thus the first source to fall victim to the new climate deal as already predicted by Dugué et al. (2012) and the IPCC (2014). The consequences on agriculture, surface water (ponds, lakes, and rivers), and groundwater (surface water table) are counted. Almost everywhere, this resource is becoming scarcer.

4 Conclusion and Perspectives

The changes in climatic conditions are considerable both in their interannual and seasonal distributions. Indeed, the climate in West Africa exhibits large spatial and temporal variabilities and droughts are therefore recurrent. This hydrometeorological extreme event varies from lack of rain in one season (dry spells) and lasts for many years. Remote sensing and in situ observations reveal that the African continent is highly vulnerable to droughts, which have a heavy toll on people and the economy due to a large dependency on rainfed agriculture. This is visible at the local scale. In this paper, rainfall in the Bandama watershed has been characterized by a trend toward drought since the rainfall decline in the 1970s. The available water in the soil fluctuates at the rate of rainfall variation. Indeed, a reduction in rainfall; especially the lengthening of dry seasons accentuates water deficit. As a result, water in the soil (ponds, lakes, streams, wells, etc.) is drying up. This deficit is very much felt by the populations, animals, environment, and economy. The continuous degradation of climatic and hydrological conditions has compromised access to water and represents important indicators to evaluate and anticipate the risks of water resource shortage. The planning and management of this hydrometeorological extreme event require a change of paradigm. Indeed, awareness and preparedness should shift from crisis management toward risk management. Thus, comprehensive studies on drought events are more than necessary. Certainly, policies could be better designed and written to significantly improve planning and tackling mitigation strategies for droughts.