The Republic of Togo, located in West Africa along the Gulf of Guinea, extends from latitudes 6˚N to 11˚N ( Figure 1 ). It shares its borders with Ghana to the west, Benin to the east, and Burkina Faso to the north. The country’s southern coastline stretches for 56 kilometers along the Gulf of Guinea. Encompassing a total area of 54,600 km², Togo exhibits a diverse topography, characterized by undulating hills in the northern regions, a central plateau in the south, and a low-lying coastal plain interspersed with extensive lagoons and marshlands . Togo faces significant socio-economic challenges, with approximately 69% of rural households currently living below the poverty line. The nation’s climate transitions from tropical in the south to savanna in the north. Rainfall patterns in the southern regions follow a bimodal distribution, with the first rainy season extending from mid-March to late July and the second occurring from early September early to mid-November. Conversely, the northern regions experience a single rainy season, typically from May to October . During the dry season, spanning from November to March, Togo is influenced by the Harmattan, a dry and dusty trade wind originating from the northeast, which brings cooler temperatures and arid conditions. Additionally, the northern regions are periodically affected by droughts.

This study utilizes two primary data sources. The first source consists of outputs from the CORDEX program, which provides simulations from a set of Regional Climate Models (RCMs) covering different regions worldwide. The CORDEX model domains are detailed at http://wcrp-cordex.ipsl.jussieu.fr/images/pdf/cordex_regions.pdf . For this analysis, datasets from the Africa-CORDEX domain were accessed via the Earth System Grid Federation at https://esgf-data.dkrz.de/projects/esgf-dkrz/ . These datasets have undergone quality control procedures to ensure reliability.

All CORDEX Africa RCMs operate on a rotated pole coordinate system with a spatial resolution of approximately 0.44˚ longitude by 0.44˚ latitude, corresponding to a quasi-uniform grid size of about 50 km × 50 km across the equatorial domain ( Table 1 ). The RCMs simulate climate dynamics based on Representative Concentration Pathway (RCP) scenarios, with detailed descriptions of their physical parameterizations and driving General Circulation Models (GCMs) provided by Nikulin et al. (2018) [19] . Among the driving GCMs are CNRM-CERFACS-CNRM-CM5 (CNRM), ICHEC-EC-EARTH (ICHEC), and MPI-M-MPI-ESM-LR (MPI), which have been thoroughly evaluated in previous studies (Nikulin et al., 2018). These CORDEX data have also been employed in prior research over the Mono Basin [18] .

The second data source consists of observed rain records obtained from the National Meteorological Agency of Togo (Météo Togo) for the period 1980-2024. These observations cover stations spatially distributed across the country. To ensure consistency with model outputs, the observational data were regridded onto a common 50 km spatial grid [18] .

The correlation coefficient (R), root mean square error (E), and standardized standard deviation (SD) were used to analyze the accuracy and deviation between the simulated and observed data, which were quantified as follow:

$\text{R}=\frac{\frac{1}{N}{\displaystyle {\sum }_{i=1}^N\left(f_n-\bar{f}\right)\left(r_n-\bar{r}\right)}}{{\sigma }_f{\sigma }_r}$

$\text{E}=\sqrt{\frac{1}{N}{\displaystyle {\sum }_{i=1}^N{\left(f_n-r_n\right)}^2}}$

$\text{SD}=\sqrt{\frac{1}{N}{\displaystyle {\sum }_{i=1}^N{\left(f_n-\bar{f}\right)}^2}}$

where, $r_n$ and $f_n$ are the simulated and observed data respectively. ${\sigma }_r$ and ${\sigma }_f$ are the standard deviation of $r_n$ and $f_n$. The simulated data were more accurate when the R was closer to 1. Similarly, when the E and SD were closer to the observation point on the x-axis, smaller deviations between the simulated and observed data were presented.

The relative changes from the reference period are assessed as shown by equation below:

$\text{I}=\left(\frac{\text{FF}-\text{HIST}}{\text{HIST}}\right)\cdot 100$

where FF and HIST represent respectively the mean for the near future (2021-2040) and the historical or reference period (1981-2010). These methods have been used over Mono Basin by Batablinle et al. (2018) [18] and Lawin et al. (2019) [20] .

The analysis was conducted using Python, an open-source programming language widely applied in climate and hydrological research for its robust data processing, statistical analysis, and visualization capabilities. Python’s scientific libraries—such as xarray, numpy, pandas, and matplotlib were employed to handle climate datasets, perform spatial and temporal aggregation, and generate the projection maps. This computational approach ensured reproducibility, flexibility in processing large datasets, and the ability to integrate multiple Regional Climate Model (RCM) outputs from the Africa-CORDEX archive into a multi-model ensemble for improved projection reliability.

Rainfall variability exerts a significant influence on water management, agriculture, and hydrological regimes in Togo. Figure 2 presents both the average monthly rainfall pattern and the annual rainfall trends, providing insights into the seasonal distribution and interannual fluctuations of precipitation across the country.

The analysis of average monthly rainfall in Togo over the 1983-2023 period reveals a rainfall regime characterized by a single, well-defined rainy season, peaking in August. From November to April, rainfall remains very low, often below 30 mm, marking a prolonged dry season. Beginning in May, a rapid increase in precipitation is observed, reaching its maximum in August with over 230 mm, before gradually declining towards November. This pattern reflects the seasonal migration of the Intertropical Convergence Zone (ITCZ) and has a direct influence on the country’s agricultural calendar.

In parallel, the examination of annual rainfall trends between 1983 and 2024 highlights a pronounced interannual variability around an average of 1500 mm/year. Certain years, such as 1990 and 2000, record significant surpluses, whereas others, notably 1998 and 2014, show marked deficits. Although no strong linear trend is immediately evident, a slight downward tendency can be observed after the mid-2000s. This variability, influenced by regional and global climate drivers such as ENSO events and oscillations within the West African monsoon system, poses considerable challenges for food security and sustainable water resource management.

Table 2 shows that precipitation variability across the wettest months of the year remains relatively low, with a constant coefficient of variation (CV) of 15% from May to October. This indicates a relatively uniform rainfall distribution during the main rainy season, reducing uncertainty for agricultural activities and water resource planning. May records an average of 122.13 mm, marking the effective onset of the rainy season. Rainfall then increases steadily through June (181.52 mm) and July (203.26 mm), peaking in August with 230.92 mm, before declining to 187.22 mm in September and dropping sharply to 98.85 mm in October.

Although variability remains stable, spatial and temporal differences are still evident. The highest maximum values are observed in August (300.23 mm), followed by July (264.27 mm), reflecting strong convective activity during the peak monsoon. Conversely, October’s lower average (98.85 mm) marks the retreat of the rainy season.

From a modelling perspective, interpolation accuracy varies by month. The highest R² values are recorded in October (0.93) and July (0.91), indicating strong agreement between observed and estimated values, whereas August shows a relatively lower R² (0.79), suggesting more complex spatial rainfall patterns during the peak season. The RMSE values are lowest in October (17.79 mm) and highest in August (41.57 mm), confirming that the latter month is more challenging to model due to localised heavy rainfall events.

Overall, the rainfall regime is typical of a tropical monsoon climate, with a well-defined peak in August and consistently low variability during the rainy season. Understanding this distribution is essential for optimizing crop calendars, managing hydrological systems, and anticipating potential flooding risks in peak months.).

The series of maps provides a detailed projection of the mean monthly precipitation across Togo during the rainy season from May to October, based on the CORDEX Africa mean model. To begin with, the spatial distribution highlights a latitude range roughly between 6.1˚N and 11.1˚N, and longitude from −0.15˚ to 1.8˚E, allowing us to observe how rainfall varies geographically within the country ( Figure 3 ). Looking at the precipitation scale, values range from about 4.5 mm/day in the driest areas to 9.5 mm/day in the wettest regions, represented by colors from deep blue to dark brown.

Focusing first on May (Month 5), rainfall is generally moderate to high in the central and southern parts of Togo, indicating the onset of the rainy season concentrated mainly in the southern half. Meanwhile, the northern regions experience relatively low precipitation levels. Moving into June (Month 6), precipitation intensifies particularly in the south-central areas, yet northern areas remain relatively dry, suggesting a gradual northward progression of the rains but maintaining a clear south-to-north gradient.

By July (Month 7), a notable increase in rainfall occurs in central and northern regions, marking the spread of the peak rainy season further north, although the southern parts see a slight decrease compared to June. Continuing into August (Month 8), precipitation remains high in central and northern areas, with some of the highest recorded values during the season, while the southern region experiences moderate rainfall. This pattern indicates that August likely corresponds to the peak intensity of the rainy season across much of Togo.

Subsequently, in September (Month 9), rainfall begins to diminish in the northern and southern areas but remains elevated in the central regions, signaling the start of the rainy season’s retreat. Finally, in October (Month 10), precipitation significantly decreases across the south and center, with minimal rainfall in the north, reflecting the transition toward drier conditions and the end of the rainy season.

In summary, the projection maps depict a classic West African monsoon progression: rainfall initiates in the south, advances northward to peak around August, and then gradually recedes by October. Overall, the spatial and temporal distribution of precipitation aligns well with known seasonal dynamics and climate patterns observed in Togo.

This 2020-2040 projection for Togo, derived from the multi-model mean of CORDEX-Africa simulations, shows relative (%) changes in precipitation from May to October compared to the 1983-2024 reference period ( Figure 4 ). The results highlight pronounced seasonal and spatial contrasts across the latitudinal (6˚N - 11˚N) and longitudinal (0˚ - 1˚E) gradients, reflecting the seasonal dynamics of the Intertropical Convergence Zone (ITCZ) and its differentiated regional impacts.

In May, the onset of the rainy season remains relatively stable, with deviations limited to ±5% in most areas. A localized increase of about +10% is projected near 8˚N/0˚, indicating micro-scale geographical variations. In June, precipitation is projected to increase more generally by +10% to +15%, with the most notable anomalies around 9˚N/1˚E, likely linked to an earlier and more intense arrival of the West African monsoon.

By July, the projection indicates a marked north-south contrast: the northern latitudes (10˚N - 11˚N) experience significant increases (+20% to +25%), while the south (6˚N - 7˚N) shows a slight decline (−5%). This shift reflects the northward displacement of the ITCZ, reducing rainfall in the southern part of the country. In August, the strongest projected anomaly appears at 10˚N/0˚ (+30%), while a decrease (−10%) is found at 7˚N/1˚E. Such concentration of rainfall in the north suggests the likelihood of intense, localized storm events.

In September, anomalies return closer to reference values (±5%), except at 8˚N/0˚, where a sustained increase of +15% persists. This indicates a transitional phase in the rainy season, although localized flood risks remain. By October, the projection shows a generalized decrease in precipitation (−10% to −15%), marking the end of the rainy season, except for 9˚N/1˚E, where rainfall levels remain stable.

Overall, the multi-model mean projection confirms a strong north-south gradient: northern regions (10˚N - 11˚N) are likely to experience substantial rainfall increases, while southern areas are expected to face persistent deficits. Recurrent “hotspots” near 8˚N/0˚ and 10˚N/0˚ appear particularly sensitive to climate variability and change. These trends have important implications: in the north, projected rainfall increases in July-August raise the risk of flooding, while in the south, declines from July to October heighten drought vulnerability.

To adapt to these projected conditions, targeted climate monitoring should focus on the 8˚N - 11˚N belt, agricultural calendars should be revised to incorporate drought-tolerant crop varieties in the south and improved drainage in the north, and further high-resolution studies should assess the influence of land-use change on rainfall variability.

Togo exhibits notable regional climatic variability, with a traditionally bimodal rainfall regime observed in the southern part of the country, characterized by two distinct rainy peaks annually. In contrast, the majority of the territory particularly the central and northern regions is dominated by a unimodal rainfall pattern, featuring a single well-defined rainy season from May to October. However, our recent analyses based on monthly precipitation series indicate a progressive transformation of this bimodal regime into a near-unimodal pattern on a national scale. This shift reflects a homogenization of the rainfall cycle, with the gradual disappearance of the second rainfall peak in the south, thereby aligning the southern region’s climatic characteristics more closely with those of the central and northern parts of the country.

Our analysis (1983-2024), combined with projections for 2020-2040, complements and expands upon earlier studies conducted over the 1961-2001 period [21] (The previous study highlighted a general decline in annual precipitation across nearly all agro-ecological zones in Togo, identifying the Maritime Region as the driest and Kouma-Konda as the wettest, with an average annual rainfall of 1714 mm. June was reported as the wettest month, and the Mann-Kendall test revealed a statistically significant downward trend in annual rainfall.

In contrast, our study leverages higher spatial and temporal resolution data, employing a multi-model ensemble from the CORDEX-Africa program and advanced statistical methods implemented in Python. While confirming substantial interannual and seasonal variability, we additionally observe the marked transformation of the southern region’s bimodal rainfall regime into a predominantly unimodal pattern across the entire country a phenomenon less evident in earlier analyses. Moreover, our projections indicate contrasting spatial trends, with increasing rainfall in northern Togo and decreasing precipitation in the south, suggesting more complex dynamics than the broad declining trend previously documented.

These differences likely reflect recent shifts in large-scale climate drivers such as ENSO and the Atlantic Multidecadal Oscillation, as well as methodological improvements that better capture localized extremes and variability [22] . The implications for water resource management and agriculture remain critical: whereas earlier findings emphasized risks associated with persistent precipitation deficits and drought, our results highlight the necessity of adapting to spatially differentiated changes and the gradual loss of bimodality, which impacts traditional agricultural calendars [2] . In sum, integrating historical observations with contemporary data and future projections offers a more nuanced and comprehensive understanding of rainfall variability in Togo. This enhanced perspective is vital for developing effective climate resilience strategies and ensuring food security in the face of evolving climatic conditions.