The African continent, covering latitude 35˚00'S to 52˚00'N and longitudes from 24˚00'W to 32˚00'E, was subdivided into nine sub-regions in the recent Intergovernmental Panel on Climate Change Sixth Assessment Report (IPCC AR6) ([23]). Some studies considered three divisions for Africa ([36]). In this study, four domains were used based on hydroclimatic zones, namely, North Africa (15.5˚N-37.5˚N, 18˚W-37.5˚E), West Africa (11.5˚S-15˚N, 18˚W–30˚E), East Africa (11.5˚S-15˚N, 30.5˚-52˚E), and Southern Africa (35˚-12˚S, 11˚-52˚E), as illustrated in Figure 1(a). These divisions enable us to detect how each region responds to climate conditions without generalizing continent patterns. North Africa comprises the hyper-arid Sahara Desert, regarded as a climate change hotspot characterized by high temperatures, water scarcity, and frequent drought ([39]). West Africa, as defined in this study to include the Central African domain, is dominated by the West African Monsoon and the influence of the Congo Rainforest Basin, driving a north to south rainfall gradient and high annual rainfall over the equatorial region ([33]). Rainfall in the southern and equatorial parts of the region is influenced by strong local moisture recycling and large-scale circulation patterns, driving moisture from the Atlantic and Indian Oceans ([16]). East Africa has a bimodal rainfall pattern, from March to May, known as the “long rain” period, and October to December “short rain”, which is becoming wetter. Approximately 75% of the land is classified as arid to semi-arid, impacting various sectors ([6]). In addition to West Africa, the Southern African climate is largely controlled by the oscillation of the Intertropical Convergence Zone (ITCZ) and ENSO variability, with El Niño typically inducing drought conditions ([9]). This regional division provides the essential spatial context for analyzing the drivers and impacts of aridity changes across the continent’s diverse climates.

The study employed precipitation data obtained from the Global Precipitation Climatology Centre (GPCC) full data Monthly Version 2023 ([40]), providing a gridded dataset at a resolution of 0.25˚ × 0.25˚ covering the entire African continent for the study period from 1951 to 2020. Potential evapotranspiration (PET) was computed from surface net radiation (R_n), relative humidity (RH), and wind speed (U), which we obtained from the ERA5-Land reanalysis dataset available at 0.1˚ × 0.1˚ resolution ([32]). The precipitation was regridded from 0.25˚ × 0.25˚ resolution to match the ERA5-Land grid using linear interpolation for aridity index calculations. This approach of regridding data to a common grid minimizes scale mismatches when working with data from different sources, types, or resolutions. Several studies in climate science research use this practice; for example, temperature and multiple precipitation were regridded to get a common grid for trend analysis and other statistics in [20]. Similarly, [35] used the ERA5-Land reanalysis dataset as a reference while downscaling and regridding precipitation and temperature datasets to obtain higher resolution data from 0.25˚ × 0.25˚ to 0.1˚ × 0.1˚. These datasets were collectively used to assess aridity trends, spatial distribution, and corresponding contributions for annual and seasonal time scales. The data were aggregated from monthly values to standard seasons (DJF, MAM, JJA, SON) and annual means for further trend analysis across space and time.

The AI was used in this study as a standardized measure of dryness, and was defined as the ratio of precipitation to PET, representing water supply and atmospheric water demand, respectively:

In Equation (1), P is the total precipitation (mm) and PET is the potential evapotranspiration (mm). The AI classification follows the United Nations Environment Programme (UNEP) scheme ([29]) and is divided into Hyper-arid (AI < 0.05), Arid (0.05 ≤ AI < 0.20), Semi-arid (0.20 ≤ AI < 0.50), Dry subhumid (0.50 ≤ AI < 0.65), and Humid (AI ≥ 0.65) ([29]; [31]). This categorization has been extensively adopted in many studies ([13]; [17]; [26]; [38]).

PET was estimated using the FAO Penman-Monteith method shown in Equation (2), a standardized method for reference evapotranspiration by [4]. Monthly gridded datasets were used for the required meteorological variables in this calculation as follows: T, R_n, RH, and U, as shown in the general form of Equation (2) below.

In Equation (2), PET is potential evapotranspiration (mm/day), Δ is the slope of the vapor pressure curve (kPa∙˚C⁻¹), R_n is net radiation at the surface (MJ∙m⁻²∙day⁻¹), G is soil heat flux density (MJ∙m⁻²∙day⁻¹, assumed = 0), γ is the psychrometric constant (kPa∙˚C⁻¹), T is mean air temperature (˚C), U₂ is wind speed at 2 m height (m∙s⁻¹),eₛ is saturation vapor pressure (kPa), and eₐ is actual vapor pressure (kPa) replaced by RH.

This method is preferably used since it takes into account both energy balance and aerodynamic factors, making it suitable to be used across diverse climatic zones to provide reliable results. The PET was further decomposed into contributions from its driving meteorological variables in order to assess their regional and seasonal influence on drying trends.

Linear trends in AI, precipitation, and PET were calculated using Ordinary Least Squares (OLS) at each grid cell, and statistical significance was assessed at p < 0.05. In this method, we did not apply adjustment for serial autocorrelation (e.g., pre-whitening or Modified Mann-Kendall) since the analysis focused on estimating the magnitude and spatial pattern of long-term trends at continental scale rather than testing individual grid-cell significance. This approach follows previous studies that utilized a similar approach ([2]; [10]). Regional trends were derived from the aggregated time series, with 5-year running means applied to visualize decadal variability. PET changes were further decomposed into partial contributions of its primary drivers, namely T, RH, U, and available energy (AE), which is the difference between surface net radiation and soil heat flux density. This decomposition approach was adopted from [27], where we treated the contributions of each variable separately to assess its influence on PET change while holding other variable constant. The methodology assumes interaction between meteorological variables are negligible and the total change in PET is the sum of these individual effects. The physical decomposition of the change in PET is expressed as follows: ΔPET = ΔT + ΔAE + ΔRH + ΔU, while for aridity index ΔAI = ΔP + ΔPET.

The spatial distribution of aridity across Africa from 1951 to 2020 (Figure 1(a))

Figure 1. African Aridity Baseline (1951-2020). Spatial patterns of mean aridity and its key drivers: (a) Aridity classification based on the UNEP scheme using the mean annual Aridity Index. (b) Mean annual rainfall (mm/year). (c) Mean annual potential evapotranspiration (mm/year). (d) Area coverage (%) of each aridity class.

shows the mean annual state of AI, with classification based on the UNEP scheme. This provides a baseline for this study, highlighting long-term patterns of aridity and regional variations. Similarly, Figure 1(b), Figure 1(c) present mean annual precipitation and potential evapotranspiration (PET), highlighting distinct moisture availability and its variations across regions in Africa, while the extent of each aridity class and areal coverage is quantified in percentages (Figure 1(d)). In North Africa, which is classified as a hyper-arid region, minimal mean annual precipitation (~89 mm) is received, and the highest PET (~1902 mm/year) on the continent is experienced. Conversely, West Africa is identified as a humid region, where high rainfall reaches ~1261 mm/year and moderate PET (~1595 mm/year) is experienced. This gives a representation of how aridity is distributed across Africa to identify areas likely vulnerable to aridification.

The widespread drying conditions indicated by negative annual AI trends (Figure 2(a)) are prominent across large parts of Africa, covering about 88.9% of the total continent area, with 43% showing statistically significant drying, especially over the West African region. The mean decline in AI over the continent is −0.0012 yr⁻¹, corresponding to an 8.4% reduction over a 70-year period. Across all seasons, there is a consistent drying pattern, but the exception is SON, which registered the strongest mean trend of −0.0014 yr⁻¹, representing 29.1% significant area drying (Figure 2(m)). This significant drying pattern experienced over the seasons predominantly covers West Africa and part of East Africa across the seasons. The PET annual trend dominates Africa with 85.7% significant trends, while precipitation accounts for only 20.3% (Figure 2(b) and Figure 2(c)). This indicates the dominance of the PET-driven AI trend rather than precipitation in most places. On a seasonal scale, PET significant trends are 67.8% - 82.3%, while precipitation only covers 8.7% to 14.6% of the African continent (Figure 2). This suggests that PET spatially overshadows precipitation influence.

All four regions display universal drying trends but at varying intensities (Figures 3(a)-(t)). West Africa stands out, experiencing the most significant annual drying, with a trend of ΔAI = −0.0025 yr⁻¹ (p < 0.01; Figure 3(b)), and this pattern is persistent across all seasons. East Africa presents the next significant decreasing pattern during the annual and three of the seasons (JJA, MAM, and DJF). The annual trend here is ΔAI = −0.0011 yr⁻¹ (p = 0.004; Figure 3(c)). Southern Africa depicts a different pattern from East Africa, with a lower annual drying trend of ΔAI = −0.0009 yr⁻¹ (p = 0.015; Figure 3(d)). The season with the most significant seasonal drying trend for the region is SON (Figure 3(t)). However, North Africa exhibits the lowest annual AI drying trend compared to the three other regions (Figure 3(a)).

Overall, the analysis reveals the highest decreasing trend over western Africa, followed by East Africa, as compared to other regions.

Figure 2. Seasonal trends (1951-2020) Aridity index (left column), Precipitation (middle), and potential evapotranspiration (right) across Africa. Rows show Annual, DJF, MAM, JJA, and SON. Hatching denotes changes significant at the 95% confidence level, indicated by percentage values (bottom-left).

Figure 3. Regional Aridity Index (AI) trends across Africa (1951-2020) from (a) to (t) show time series of annual AI values (blue), linear trends (black dashed), 5-year running means (dark red), and 95% confidence intervals around the trend (gray shading). Each panel includes an annotation of the annual trend magnitude (yr⁻¹), percentage change per year, and the corresponding p-value. Statistical significance is indicated by asterisks in each panel, where * denotes p < 0.05, and ** denotes p < 0.01.

We further analysed ΔP and ΔPET to evaluate their influence on aridity across the continent, as shown in Figure 4. Spatially, PET emerges as the most coherent driver of African aridification during 1951-2020, covering a wide area during annual and all seasonal periods, representing 86.1%-98.5% of Africa (Figure 4). This widespread increase in the PET signal aligns closely with the increase in temperature, explaining about 82.1%-98.4% of the rising atmospheric moisture demand. However, precipitation changes exhibit an uneven distribution pattern, revealing greater spatial heterogeneity. Here, precipitation only explains 20.3% of the drying signals annually, whereas individual seasons register 8.7%-14.6%.

Figure 4. Contributions to AI changes (1951-2020) from precipitation change (ΔP), potential evapotranspiration change (ΔPET), temperature change (ΔT), available energy change (ΔAE), relative humidity change (ΔRH), and wind speed change (ΔU). The columns show annual and seasonal arrangements from left to right (Annual, DJF, MAM, JJA, SON). Hatching and percentages (bottom-left) indicate grid cells with statistically significant contributions (p < 0.05). Physical decomposition follows: ΔPET = ΔT + ΔAE + ΔRH + ΔU, ΔAI = ΔP + ΔPET.

We extended the analysis to examine the relative contributions of precipitation and PET at the regional level. The results vary across the four study regions (Figure 5), where over West Africa, annual aridification is induced by increasing PET, followed closely by declining precipitation (Figure 5(b)). In this region, the PET increase is consistent with reduced available energy and rising temperature,

Figure 5. Regional mean contributions of climate drivers to Aridity Index (AI) trends (1951-2020). Rows: seasons (Annual, DJF, MAM, JJA, SON); columns: regions. Bars show average contribution percentages: ΔP (blue), ΔPET (orange), ΔT (green), ΔAE (red), ΔRH (purple), ΔU (brown). Negative values: drying contributions; positive values: moistening. Red asterisks (*) indicate sub-regional contributions where ≥50% of the grid cells are statistically significant (p < 0.05).

suggesting a warming environment driving increased atmospheric moisture demand. Here, the influence of RH and wind speed is very minimal, suggesting a strong influence of thermodynamic factors, unlike aerodynamic factors driving rising PET; however, precipitation change remains an important factor in the region.

Similarly, East Africa drying patterns indicate the influence of PET across all seasons, explaining close to one-half of the overall drying. Unlike West Africa, this drying arises from the combined effects of rising temperature and declining relative humidity. The situation is different during the SON season, where precipitation plays a moderating role on the effect of the strong PET contribution, partially proving moistening, resulting in minimizing net drying only to −4.9%.

In Southern Africa, the results are different as precipitation deficits here control much of the annual drying in the region. The strongest drying effect from precipitation decline is observed during SON, with PET contributing less than half of it (Figure 5(t)). This dominance is similarly visible during other seasons. PET increases contribute to additional drying, but their influence is considerably weaker compared to declining precipitation. Precipitation dominance in driving aridification over Southern Africa is consistent across all seasons. DJF shows elevated atmospheric demand, marked by rising temperature overshadowing the precipitation effect in regulating aridification.

Over North Africa, annual aridification is strongly driven by precipitation deficits, compared with −28.9% from ΔPET (Figure 5(a)). This mechanism is similar to that in Southern Africa. However, during the SON season, increases in PET exceed the precipitation effect on aridity, reflecting the growing role of PET (Figure 5(q)). This seasonal dominance of PET was observed to be driven by a rise in temperature and a decline in humidity in the region.

Overall, these results reveal that aridification in different regions of Africa has its own mechanisms. West and East Africa are influenced by rising PET, highlighting their sensitivity to warming and atmospheric drying, whereas Southern Africa is dominated by precipitation deficits, reflecting its vulnerability to rainfall change. Similar dependence on precipitation is also observed in North Africa, with a seasonal shift in aridification contribution from precipitation-dominated to PET-driven drying in SON. Therefore, based on these patterns, there is a need for adaptation strategies at the regional level, addressing both declining water availability and rising atmospheric moisture demand.

Analysis of the shift in aridity categories between the early (1951-1984) and late (1985-2020) periods across Africa (Figure 6) reveals a highly heterogeneous pattern defined by regional and seasonal distinctions. We reveal that Southern Africa emerges as a hotspot for widespread aridity intensification, shifting from semi-arid to arid conditions, while dry sub-humid regions degrade towards semi-arid conditions (Figure 6(e)). The hyper-arid Namib Desert is also expanding, highlighting aridification across the region. Despite these aridification patterns being

Figure 6. Seasonal aridity category transitions across Africa between 1951-1984 and 1985-2020. Panels show (a) Annual and seasonal (b) to (c) for DJF, MAM, JJA, and SON) shifts among five aridity classes (Hyper-arid to Humid). Colors indicate drying (red to orange) and wetting (blues) transitions of varying severity.

visible across other seasons, DJF showed a contrasting signal of seasonal recovery as the rainy season begins over Southern Africa (Figure 6(b)). This demonstrates the critical role of the inter-seasonal cycle in modulating the drying shifts in the regions.

Similar complexity is observed in East Africa, where seasonal divergence challenges a narrative of uniform aridification. Despite the annual signal showing a net shift toward drier conditions (Figure 6(a)), the SON season exhibits a widespread moistening pattern, characterized by a transition from the semi-arid to dry sub-humid category during the later period (Figure 6(e)). Another similar moistening signal was also detected in East Africa but was confined to the Horn of Africa during MAM (Figure 6(c)).

In West Africa, drying pattern transitions were observed from predominantly humid to dry sub-humid, while semi-arid conditions shifted to arid conditions, reaching their peak during the dry DJF season, signifying the expansion of the region toward more arid conditions. However, the rainy JJA season presents seasonal relief from aridification in the region (Figure 6(d)).

North Africa, mainly characterized by the hyper-arid Sahara Desert, maintains its persistent climatic conditions. The results here reveal a dominant transition category within the arid spectrum from arid to hyper-arid, signaling an intensification and spatial expansion of existing extreme dryness. We identified minimal seasonal modulation to aridification in the region due to the rainy JJA season shifting northwards, causing hyper-arid regions to become arid (Figure 6(d)).