Hydroclimatic trends, breakpoints and future projection analysis of the Ogun River Basin in West Africa

Hydroclimatic trends play a vital importance in the field of water resources and are very complex because they depend upon many factors such as atmosphere, hydrosphere, cryosphere, geosphere and biosphere (Gebre & Ludwig 2014). Changes in one of these factors alter the hydrological cycle, for example, the concentration of greenhouse gases (GHGs) especially carbon dioxide has increased over the last few decades resulting in global warming (Mahmood & Jia 2017). The level of increasing GHG emission due to anthropogenic activities in the past decades has caused unequivocal global climate system changes and one of the most significant consequences of global warming would be an increase in the magnitude and frequency of extreme weather events (Solomon et al. 2007; Stocker et al. 2013). The fifth report of the Intergovernmental Panel on Climate Change (IPCC) reported the annual mean temperature from 2003 to 2012 increased by 0.78 °C compared with that from 1850 to 1900 (IPCC 2013). The most obvious impacts of climate change have resulted in the rise in global temperature and altered precipitation patterns, due to the inseparable linkage between the climatic and hydrological cycle, global warming is reported to accelerate water circulation, which will lead to spatial-temporal redistribution of water resources at global and regional scales (Yang et al. 2020; Zhao et al. 2021).

Temperature and precipitation are fundamental components of climate, and the analysis of changes in these variables characterizes key tasks in detecting climate changes (Okafor et al. 2017). In fact, the analysis of rainfall and temperature has continued to receive attention from researchers as a way to predict the occurrence and water resources management for various uses (Ahmad et al. 2015; Meshram et al. 2017; Hu et al. 2019), particularly in arid and semi-arid regions, which usually have higher evaporation and minimal precipitation (Forootan 2019). This is mostly the case because evaporation and rainfall are the key climate-related elements influencing freshwater resources (IPCC 2014).

One of the challenges posed by climate change is the ascertainment, identification and quantification of trends in these climatic variables and their implications on river flows so as to assist in the formulation of adaptation measures through appropriate strategies for water resources planning and management (Taxak et al. 2014; Akinsanola & Ogunjobi 2017).

Several studies on the analysis of rainfall time series have been carried out at different temporal scales and in different parts of the globe. Existing analyses of rainfall time series show for some areas a positive trend and a tendency towards higher frequencies of heavy and extreme rainfall in the last few decades (Houghton et al. 1996). In Alberta Canada, Newton et al. (2021) reported the cumulative effect of increasing temperatures, shortening the duration of winter and advancing the onset of temperature which in turn shifts the pattern of precipitation. Yang & Xing (2022) reported a magnitude increase in air temperature and a decrease in rainfall and runoff signifies reduced water resources availability in upper Yangtze, China. Also, Mohan & Rajeevan (2017) reported a consistent increase in hydroclimatic intensity due to an increase in precipitation intensity during the period of 1951–2010 and future changes in dry spell length in the Indian monsoon region. Lutz et al. (2016) reported a decrease in precipitation, which caused a decline in streamflow due to increased temperature in some Mediterranean rivers.

In West Africa, studies focused on river discharge over several basins and show that trends in rainfall influenced the variation of river discharge. Oguntunde et al. (2006) reported a decrease in rainfall and an increase in temperature which may be due to climate warming in the Volta River basin. Lawin et al. (2019) reported a decreased trend in rainfall while an increasing trend in temperature in Burundi. Adeyeri et al. (2019) also reported an increase in precipitation indices which might have been caused by the Sahelian rainfall recovery and a significant warming trend in temperature in the Komadugu-Yobe basin, Lake Chad region.

In Nigeria, significant increases in rainfall were found in the northern region between 1953 and 2002 (Ati et al. 2002), in the southwest between 1973 and 2008 (Oguntunde et al. 2011), and in the Ogun River for the future period (Awe et al. 2020).

So far, research on the change in hydroclimatic variables of the Ogun River Basin is very limited. Therefore, there is a need to understand the trends and variation of historical and future hydroclimatic variables which is pertinent for sustainable management of water resources, water and energy cycles and the increasing demand for water due to population and economic growth (Sankarasubramanian & Vogel 2002; Oguntunde et al. 2006). Therefore, knowledge of precipitation patterns is of crucial importance in hydrology, to derive precipitation-runoff relationships, flood/drought assessment and mitigation measures (Chattopadhyay & Edwards 2016).

Hence, this study seeks to improve the understanding of the past precipitation, river discharge, minimum and maximum temperatures and the future trends over the Ogun River Basin under different RCP scenarios.

The major tributaries of the Ogun River are the Ofiki and Opeki rivers. The rainfall patterns in the basin are two distinguishable seasons; a dry season from November to March and a wet season between April and October. The two major vegetation zones that can be identified within the watershed are the high forest vegetation in the North and Central parts, and the swamp/mangrove forests that cover the southern coastal and flood plains, next to the lagoon (OORBD 1982; Awe et al. 2020).

The mean wet season rainfall varies from 1,020 to 1,520 mm while the dry season rainfall is 178–254 mm (OORBD 1982; Fox & Rockström 2003). The climate of the basin can be described as the Maritime South West Monsoon wind and Dry North East Trade Wind. The mean annual temperature is about 30 °C, which can vary depending on the location and time of the year (Ifabiyi 2005). The months of February and March are the hottest in the basin and temperatures are high over the entire basin during this period.

Water resources in the Ogun River Basin include surface water and groundwater. The overlying sedimentary rock sequences are from cretaceous to recent; the oldest of them, the Abeokuta formation, consists of grey sandstones intercalated with clay. It is overlain by the Ewekoro Formation, which typically contains thick limestone layers at its base (Oke et al. 2015).

The study utilizes data from the Nigeria Meteorological Agency (NiMet) for the climatic stations and the Ogun-Osun River Basin Development Authority (OORBD) for the hydrological stations. Seven synoptic stations for precipitation and temperature data over the Ogun River Basin have been used between the period of 1984–2019 for historical and 2020–2060 for future predictions and three hydrological stations were used, respectively. The seasonal and annual trend analyses were performed on the climatic and hydrological stations for the entire basin. The statistical indicator was used for the validation of the climatic data with W5E5 data. W5E5 is a merged dataset which combines WFDE5 data over land with ERA5 over the ocean. W5E5 are based on temporally and spatially (0.25°–0.5° resolution) aggregated ERA5 data (Lange 2019a, 2019b). The list of observed climatic and hydrological stations with available data is shown in Table 1.

Global Climate Models, also known as General Circulation Models (GCMs), have been used in many studies all over the world to predict rainfall, temperature and other climate properties. GCM are models that generate the meteorological variables such as precipitation (rainfall), temperature, relative humidity, wind speed and solar radiation (Nunez & McGregor 2007). These models provide insights into future climate, projecting regional climate change, assessing climate risks or planning adaptation policies (IPCC 2014). The Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) is a dynamically downscaled and bias-adjusted version of the 20th-century reanalysis (Compo et al. 2011). The ISIMIP was designed as a framework to assess the impacts of climate change in different sectors and at different scales (Schellnhuber et al. 2014). ISIMIP (ISIMIP3 (GSWP3-W5E5) dataset has a daily temporal and 0.5° × 0.5° spatial resolution of w5e5 v2.0 for the period of 1984–2019 for the historical trend analysis and ISIMIP2 for the future prediction for the period of 2020–2060. Four GCMs were selected in the fast track ISIMIP: GFDL-ESM2, HadGEM2-ES, IPSL-CM5A-LR and MIROC 5 (Table 2). The future scenario proposed in the Coupled Model Intercomparison Project Phase 6 (CMIP6), as a Representative Concentration Pathway (RCP) of 2.6 and 8.5 were considered (Moss et al. 2010). RCP 2.6 is a very stringent pathway where radiative forcing emission peaks at approximately 3 W m⁻³ and declines back to zero by 2100 and RCP is likely to keep global temperature rising below 2 °C by 2100 (IPCC 2013) and RCP 8.5 which is business as usual scenario in which absence of climate change policies leads to higher future GHG emissions (Riahi et al. 2011). The RCP 2.6 is a mitigation scenario leading to a very low forcing level while RCP 8.5 is a high baseline emission scenario leading to high greenhouse concentration levels (van Vuuren et al. 2011).

Trend analysis is used to examine whether the trend is increasing, or decreasing or if there is no trend data value. Modified Mann-Kendall (MMK) test and Sen's Slope estimator were used to detect the various trends in climate variables and river discharge. The homogeneity test was also carried out using Pettitt's test (Pettitt 1979), Standard Normal Homogeneity Test (SNHT) by Alexandersson (1986) and Buishand's test (Buishand 1982).

The MMK test was then computed using the software R package ‘modifiedmk’ (Patakamuri & O'Brien 2021) and the null hypothesis (H_o) corresponding to ‘no trend’ and the alternative hypothesis (H₁) corresponding to the presence of a trend in the series at a significant level of 5%.

Theil-Sen estimator, also known as Sen's Slope estimator, is a procedure for robustly fitting a line to sample points in a simple linear regression by selecting the median of all line slopes through pairs of points (Gilbert 1987; El-Shaarawi et al. 2001). Sen (1968) proposed the over-whitening (OW) approach, which adds white noise with zero mean and specific standard deviation to eliminate the serial correlation.

For climatological research to accurately reflect the actual fluctuations in weather and climate, homogeneity testing is absolutely essential. Inhomogeneity in climate data can be caused by a variety of things, such as instrumentation errors, changes in the instrument's adjacent surroundings, and improper human management. The findings of trend analysis will reveal erroneous trends if the homogeneity cannot be established first.

A climatic series is said to be homogeneous if the observed variation is resulting from fluctuations in weather and climate exclusively (Domonkos & Coll 2017). The detection of homogeneity and break years in rainfall data has frequently been accomplished by researchers using various statistical techniques. Wijngaard et al. (2003) tested the homogeneity of the climate in Europe using the SNHT, Buishand range test (BRT), Pettitt's test and Von Neumann ratio (VNR) test. Given that different approaches have different breakpoints with varying sensitivity, this hybrid method produces a good outcome. It is important to carry out different homogeneity tests. In this current study, three homogeneities will be considered, which are the Pettitt's test (Pettitt 1979), SNHT by Alexandersson (1986) and Buishand's test (Buishand 1982).

Pettitt's test is a rank-based test for detecting significant changes in the mean of time series data when the exact time of change is unknown (Pettitt 1979). This method is used to determine the occurrence of a change point and it has been widely adopted to detect the abrupt change points in hydrological as well as climatic records (Mavromatis & Stathis 2011). This test is sensitive to the detection of breaks at the beginning and the end of the series and is considered robust to changes in the distributional form of the time series.

If P < 0.05, a significant change point exists, the time series is divided into two parts at the location of the change point.

This test was computed on the software R package using the package trend (Thorsten 2020) ‘pettit.test’ under the null hypothesis (H_o): ‘No change point’ and the alternative hypothesis (H₁): ‘Presence of change point’ defined at a significance level of 5%.

These methods have been used previously for analysing climate data and also to investigate hydroclimatological signals of climate change and variability (Taxak et al. 2014).

On the annual scale, there was a significant positive trend for Iseyin while all the other stations had a non-significant positive trend for precipitation. The wet season precipitation reveals a significant positive trend for Iseyin while there was a non-significant negative trend for Ibadan, Osogbo and Abeokuta while the dry season precipitation reveals a non-significant negative trend for Ilorin, Akure and Osogbo.

The annual scale (Table 4) result for the hydrological variable(discharge) shows there were no significant trends for all the discharge stations (Ilaji-ile, New-bridge and Mokoloki), similar results were found for the wet season.

Table 5 presents the monthly observed precipitation statistical indicator which reveals Ibadan (0.90), Akure (0.90) and Osogbo (0.91) have a positive linear correlation coefficient between the observed and simulated data. Akure, Ibadan and Oshodi (48.11, 48.23 and 59.99) indicate a better model simulation performance for RMSE. Ibadan and Oshodi (11.00 and 9.30) indicate model underestimation bias while Abeokuta and Ilorin (−13.20 and −8.50) indicate model overestimation bias. The NSE indicates an accepted level of performance for Abeokuta, Ibadan, Oshodi, Akure and Osogbo (0.63, 0.71, 0.57, 0.76 and 0.77) while Ilorin (−1.15) indicates that the mean observed value is a better predictor than the simulated value which indicates unacceptable performances.

Table 6 presents the validation of the monthly observed and simulated precipitation which reveals Ibadan (0.90), Akure (0.90) and Osogbo (0.91) have a positive linear correlation coefficient between the observed and simulated data. Akure, Ibadan and Oshodi (48.11, 48.23 and 59.99) indicate a better model simulation performance for RMSE. Ibadan and Oshodi (11.00 and 9.30) indicate model underestimation bias while Abeokuta and Ilorin (−13.20 and −8.50) indicate model overestimation bias. The NSE indicates an accepted level of performance for Abeokuta, Ibadan, Oshodi, Akure and Osogbo (0.63, 0.71, 0.57, 0.76 and 0.77) while Ilorin (−1.15) indicates that the mean observed value is a better predictor than the simulated value which indicates unacceptable performances. This could be attributed to the fact that Ilorin is tending towards the Sahel (Northern) region of the basin.

The validation of monthly minimum temperature reveals that Abeokuta and Osogbo (0.91 and 0.98) have a positive linear correlation coefficient between the observed and the simulated data. Abeokuta and Oshodi (0.65 and 0.93) indicate a better model simulation performance for RMSE. Oshodi (0.00) indicates an optimal level of performance while Abeokuta, Ibadan, Ilorin, Osogbo and Akure (2.10, 3.00, 3.30, 4.70 and 5.00) indicate model underestimation bias. NSE indicates an accepted level of performance for all the stations.

The validation of monthly maximum temperature reveals that Abeokuta, Ilorin and Osogbo (0.97, 0.90 and 0.99) have a positive linear correlation coefficient between the observed and the simulated data. Oshodi and Osogbo (1.21 and 1.05) indicate a better model simulation performance for RMSE. The Pbias indicates model underestimation bias for all the stations while NSE indicates an acceptable level of performance for all the stations.

Table 7 presents the change point probability and detection of homogeneity of the data series using Pettitt's test, SNHT test and Buishand's test. The applied breakpoint tests for the annual precipitation series show that all the stations are homogeneous except Iseyin which indicated a significant breakpoint in the year 2009.

For the maximum temperature, Pettitt's homogeneity indicated change points in the years 1996, 1997, 2006, 2007 and 2010 for Ilorin, Oshodi and Osogbo, Abeokuta, Ibadan and Iseyin, respectively. This implies a significant difference in the mean before and after the identified breaking point. There was no significant change detected in the maximum temperature for the Akure station.

Change points in the minimum temperature were detected in the years 1996, 2000, 2006 and 2008 for Oshodi, Osogbo and Abeokuta, Akure and Ibadan, respectively. In Contrast, the Iseyin station did not show any change in the minimum temperature.

For most cases, the three tests agreed on the existence of the change points in annual rainfall, and annual mean of the minimum and maximum temperatures. The three tests mostly detected similar breakpoint years for the different variables and stations. This increases our confidence in the obtained results as far as the breakpoint analysis is concerned.

Table 8 shows the change point probability for annual hydrological variables (discharge). For most of the variables, there was no significant change point detected at the annual scale and wet season.

For the dry season, breakpoints were detected for the three stations, two by the Pettitt's test and the third by the SNHT test. There is particularly no convergence between the three tests applied for the detected breakpoints.

Table 9 presents the trend in climatic models using the MMK analysis, (Z statistic) and Sen's Slope for the time series between 2020 and 2060 at annual time scales. Sen's Slope estimator determines the magnitude of statistically significant trends.

Understanding the trends and variation of hydroclimatic variables which is pertinent to the development and sustainable management of water resources and energy cycles and the increasing demand for water due to population and economic growth is crucial. The findings of this paper reveal trends and understanding of the rainfall and temperature pattern occurrences in the basin. Several studies have shown similar rainfall and temperature patterns in south-western Nigeria. For instance, Oyamakin et al. (2010) revealed a seasonal change in the rainfall pattern in south-western Nigeria. In Odeda local government of Ogun state, Nigeria, Balogun et al. (2016) reported a non-significant increase in monthly rainfall of 0.20–2.36 mm/year and seasonal rainfall of 1.59–6.7 mm/season. Ogunsola & Yaya (2019) findings showed a significant trend for minimum temperature in south-western Nigeria. Aiyelokun & Odekoya (2016) also indicated a gradual increase in temperature, while Owolabi (2016) revealed a steady increase in yearly temperature, which implies the impacts of global warming are being experienced in part of the region.

This study showed that there are variabilities in the precipitation and temperature over time during the period between 1984 and 2020. There was no significant increasing trend for the precipitation for some stations in the basin, this agrees with other findings (Nashwan et al. 2019a; Beyene et al. 2021; Ngoma et al. 2021). In the past century, the south-western region of Nigeria has had considerable mean temperature increases of up to 0.6 °C, according to Oguntunde et al. (2012). This increasing temperature has put stress on water resources and has led to the modification in the hydrological cycle, water quality and quantity (Abubakar et al. 2017).

A non-significant increasing trend was mostly found for the discharge data. This is in agreement with the findings of Durowoju et al. (2017) who revealed a non-significant trend in the extreme rainfall and discharge of Lagos Metropolis due to certain environmental factors. However, there were significant decreasing and increasing trends in the dry season discharge which correlates with the findings of Zhang et al. (2012). The impact of changing land use and land cover as a result of rapid urbanization and the expansion of agricultural activities as a result of population growth, industrialization and water pollution can also be said to be a contributing factor to the basin's non-significant increasing trend in discharge (Awoniran et al. 2014). Additionally, the change of wetlands has an impact on the formation of soil and the lowering of groundwater levels, which have an impact on water resources (Ballut-Dajud et al. 2022; Tobore & Bamidele 2022).

The homogeneous result of the change points may be changes in station location and environment, changes in instruments, station network intensity and structure as well as observation methods (Keevallik & Vint 2012). Other factors may include changes in land use land cover and anthropogenic activities (Adeyeri et al. 2017a; Ige et al. 2017) reduction in forest cover and human activities (Adeyeri et al. 2017b).

The implications of peri-urbanization, population growth and industrialization have a negative impact on the region's water resources ((Tobore et al. 2021), the expansion of land use and socioeconomic activities depend on changes in climatic trends (Awoniran et al. 2014). Urbanization is increasing at an alarming rate, especially in the lower basin, which is having a significant negative impact on water resources. Changes in precipitation and temperature have an immediate impact on evapotranspiration, which in turn affects the quantity and quality of runoff components of the water balance. It is anticipated that these changes in climatic trends will modify the basin's hydrological regime and put more strain on the availability of water. According to Onanuga et al. (2022), changes in climate can affect water resources due to urban development.

The result of RCPs was based on two different scenarios (RCP 2.6 and RCP 8.5) which were used in four ensembled models (GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR and MIROC5) to project the future rainfall for a period 2020–2060 with a baseline period 1984–2019. The World Meteorological Organization recommends a 30-year period as the climate normal standard when compared with the future period and this should be maintained as a reference period for monitoring long-term variability (WMO 2007, 2014; Liersch et al. 2020). Above this, a regularly updated 30-year baseline period should be employed to give people a more recent context for understanding weather and climate extremes and forecasts (Hawkins & Sutton 2016; WMO 2017).

The trends result for annual rainfall shows a non-significant trend for both RCP scenarios and an expected mean annual rainfall under scenario RCP 2.6 of 1,871 mm when compared with RCP 8.5 scenario of 1,892 mm. This result correlates with the IPCC (2014) report which indicates that changes in precipitation in a warming world will not be uniform and the high latitude and the equatorial Pacific are likely to experience an increase in annual mean precipitation by the end of this century under the RCP 8.5 scenario.

Annual temperature is expected to increase between 1.5 and 2.5 °C by 2060 under RCP 2.6, on the contrary temperature is expected to increase by 2 and 3.5 °C under RCP 8.5 by 2060. This finding is similar to the result of Moazami Goudarzi et al. (2020) which projected an increase in annual temperature ranges of 1.2–5.5 °C under RCP 8.5 scenarios in Iran. In addition, Vijayakumar et al. (2021) projected an increase in annual temperature ranges between 1.7 and 3.4 °C under RCP 8.5 in India. Aziz & Obuobie (2017) projected annual mean temperature between 3.5 and 3.7 °C under RCP 8.5 in the black Volta Basin West Africa. Riede et al. (2016) reported temperature projections over West Africa from global climate simulation ranges between 3 and 6 °C above the late 20th-century baseline depending on the emission scenario. The uncertainties associated with RCP 2.6 is to keep global warming below 2 °C above pre-industrial temperature and it assumes a consistent decrease in emissions and GHG mitigation while RCP 8.5 is a baseline scenario (worst-case climate change scenarios) which does not include any climate mitigation target (IPCC 2014).

According to the IPCC (2018), GHG emissions, which are primarily driven by population expansion, socioeconomic development and technological processes, are the key emission scenarios that dictate the future climate pattern. The intensity and frequency of floods may rise with more rainfall, which will have an impact on the aquatic and terrestrial ecosystems (Tabari 2020). The amount of water in the basin will decrease as a result of increased evaporation brought on by warmer temperatures (Lange 2019a, 2019b). Global sea level rise has been influenced by climate change, which will have a significant negative impact on the ecology (Griggs & Reguero 2021; Setzer & Higham 2022). Early warning systems, ecosystem restoration and building climate-resilient infrastructure are exemplary adaptation techniques to combat future climate change.

The study investigated the historical trends of hydroclimatic variables as well as the relationship between discharge, minimum and maximum temperatures, and precipitation time series for the annual, wet and dry seasons over the Ogun River Basin between 1984 and 2020. It was found that there was a decrease in rainfall and an increase in temperature which can be attributed to the northern part of the region closer to the Sahel region of the country. Significant increase and decrease in the river discharge for some stations during the dry season can be attributed to the rainfall pattern of the basin. The excess water from heavy precipitation from the southern part of the basin closer to the Lagos Lagoon as a result of an increasing temperature could lead to flooding which is one of the major problems in the basin. This could have a significant impact on the water management and resources and also affect the socioeconomic activities in the basin. The model projected increased precipitation when compared with the baseline mean annual precipitation of 1,355 mm and also an increase in temperature for both RCPs. Adequate and appropriate measures and relevant practices should be put in place to mitigate the warming trend.

Hence, this study's output will help to understand the past and future climate change in the region and also to undertake climate-smart adaptation and mitigation options. It is recommended that there should be coordinated efforts from the stakeholders in a participatory dialogue with the farmers in the region to understand and tackle climate and environmental issues in the basin. Efforts should be made to concentrate on afforestation and conservative practices, policy to mitigate the effect of flood occurrence such as rainwater harvesting, development of early warning climate systems, land regulations and controlling deforestation and encroachment of forest reserve areas. This will promote good management of climate extreme events like floods, land and water resources without compromising the sustainability of dynamic ecosystems in the basin.

Nevertheless, limited hydrological and climatic data of sufficient quality have hindered more research and in-depth investigation of the entire basin's hydrology. A potential drawback is the issue of significant data gaps in some parts of the sub-catchments, which made it difficult to conduct in-depth analyses for the whole basin. Future study is advised to replace missing data with in situ or satellite data, and improve data quality while taking anthropogenic activities that can affect the basin's hydrological processes into account. It is also suggested that non-functional gauges should be rehabilitated for a more in-depth investigation of the entire basin.

This study forms part of a PhD research under the West African Science Service Centre on Climate Change and Adapted Land Use (WASCAL) program on Climate Change and Water Resources at the University of Abomey-Calavi, Republic of Benin and funded by the German Ministry of Education and Research (BMBF).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.