The fast response of Sahel precipitation to climate change allows effective mitigation action

The study addresses why projections of Sahel precipitation under climate change remain highly uncertain and how the timing of different physical responses affects future rainfall. Prior CMIP simulations show both increases and decreases in Sahel rainfall by late 21st century, largely due to model-dependent atmospheric circulation responses and uncertainties in hemispheric differential warming and SST changes. A robust spatial pattern nevertheless emerges: increased precipitation over the central Sahel and decreased precipitation over the western Sahel, strengthening with warming across scenarios. The authors hypothesize that this zonal contrast arises from two mechanisms operating on distinct timescales: a fast atmospheric response to increased GHGs (stronger land and Northern Hemisphere warming, enhanced moisture flux, northward monsoon shift) and a slower ocean-mediated response via SST and circulation changes (notably Pacific- and Atlantic-related modes). They aim to temporally decompose and quantify these fast and slow components, assess their relative contributions to future changes and uncertainty, and evaluate the implications for effective mitigation actions in West Africa.

Previous CMIP generations (CMIP3/5/6) highlight large spread in Sahel rainfall projections due to differences in atmospheric circulation responses, uncertainty in SST patterns, and Northern vs Southern Hemisphere differential warming. Despite quantitative uncertainty, many studies find a consistent zonal contrast under warming: central Sahel wetting and western Sahel drying, robust across scenarios and strengthening with global warming. Mechanistic explanations include direct GHG-driven atmospheric effects (enhanced low-level moisture flux, northward monsoon shift, Sahel wetting) and indirect ocean-mediated SST responses (altered tropical Pacific and Atlantic patterns influencing monsoon circulation), which may superimpose to yield the observed contrast. Prior work suggested different timescales for these mechanisms, but their overlap in transient forcing has obscured attribution. This study builds on that literature by explicitly separating fast and slow timescale responses and linking them to mitigation relevance and model uncertainty.

- Experiments and datasets: The analysis uses CMIP6 multi-model ensembles. Idealized coupled experiments abrupt4xCO2 (CO2 abruptly quadrupled; first 150 years analyzed) are used to define fast and slow responses. Atmosphere-only AMIP-style experiments (amipC4IO/amip4C5/6/7/8 with CO2 forcing; amip-p4K with uniform +4 K SST warming) are used to diagnose land vs ocean forcing roles with prescribed SST and sea ice. Historical (1850–2014) and ScenarioMIP SSP585 (high forcing to ~8.5 W m−2 in 2100) projections assess transient evolution; SSP534_over represents a mitigation pathway tracking SSP585 until 2040 then rapidly reducing emissions to net negative, reaching ~3.4 W m−2 by 2100. - Seasons, regions, and variables: Focus on July–August–September (JAS) means, analyzing precipitation, surface temperature, surface winds, moisture flux convergence, and evaporation. Key regions are central Sahel [10°–20°N; 5°W–20°E] and western Sahel/eastern tropical Atlantic [10°–20°N; 30°W–10°W]. - Definition of fast and slow responses: In abrupt4xCO2, the fast response is the change averaged 5–10 years after CO2 quadrupling relative to preindustrial control; the slow response is the additional change between 120–140 years and 5–10 years after quadrupling. This separates rapid atmospheric/land-dominant adjustments from slower ocean-circulation-mediated SST responses. AMIP experiments corroborate attribution: elevated CO2 with fixed SSTs reproduces fast, land-warming-driven strengthening of monsoon energy gradients and precipitation increase; uniform +4 K SST warming with fixed CO2 reproduces a weakening of gradients, enhanced dry intrusions, and precipitation decrease. - Regression-based decomposition in transient scenarios: A spatial multiple linear regression expresses annual GMST anomalies (relative to 1960–1999) as a combination of patterns derived from fast and slow responses (from Fig. 1a,b), yielding coefficients F(t) and S(t) interpreted as contributions of fast and slow components to GMST in historical and SSP585/SSP534_over. Precipitation changes are reconstructed by scaling the fast and slow precipitation patterns (from Fig. 1c,d) by these coefficients. An intercept term r(t) is treated as residual. Interannual variability is reduced with a 21-year running mean; fourth-order polynomial fits are applied to further smooth fast/slow components at model level, minimizing internal variability effects. - Ensembles and robustness: 26 models for abrupt4xCO2 and for historical/SSP585; 10 models for AMIP; 7–8 models for SSP534_over. Robustness is assessed via stippling where ≥80% of models agree on sign. Pattern correlations are computed to compare reconstructed vs raw SSP585 responses. - Uncertainty quantification: Total uncertainty in 21-year mean precipitation is partitioned into fast-response model variance, slow-response model variance, and internal climate variability (estimated from preindustrial control 21-year running variances, assumed approximately stationary under warming). Additional checks with multi-member ensembles confirm that the method filters most internal variability.

- Temporal decomposition: The fast atmospheric response emerges within the first decade after increased CO2, driven by stronger land and Northern Hemisphere warming and enhanced low-level moisture flux, shifting the monsoon northward and increasing Sahel precipitation. The slow response develops over many decades to a century, linked to ocean circulation/SST changes (including El Niño-like Pacific patterns and Atlantic multidecadal variability), and generally opposes the fast response over the Sahel by weakening gradients, increasing dry intrusions, and reducing rainfall. - Land vs ocean forcing: AMIP experiments corroborate that the fast response is primarily land-warming/CO2-driven, while the slow response is ocean-warming/SST-driven. - Contribution to global warming: Under SSP585, the fast response grows rapidly and dominates GMST changes; the slow response contributes later and accounts for about 25% of the GMST anomaly by late 21st century. - Spatial rainfall pattern: For a 3 K GMST warming, the fast response increases and shifts Sahel rainfall northward, while the slow response shifts circulation southward and reduces rainfall over the western Sahel and tropical Atlantic. The reconstructed precipitation change (fast + slow) closely matches the SSP585 pattern, with a pattern correlation of 0.97 over [5°–25°N, 20°W–30°E]. Many grid points show robust agreement (≥80% of models) in sign. - Zonal contrast mechanism: The observed increase over the central Sahel and decrease over the western Sahel stem from antagonistic fast and slow mechanisms. Central Sahel wetting is dominated by the fast response; western Sahel drying is dominated by the slow, ocean-mediated response. Consequently, central Sahel changes are stronger and grow earlier than western Sahel changes. - Reconstruction skill: Central Sahel precipitation increases across the 20th–21st centuries in reconstructions aligned with SSP585; slow response yields moderate negative offsets. Over western Sahel, reconstruction is less accurate for weak signals/low SNR and may be affected by aerosols, but becomes more accurate for GMST warming >2 K (post-2050). - Mitigation effectiveness (SSP534_over): A sharp emission reduction after 2040 leads to a GMST decline after ~2060, mainly via fast-response reduction. Central Sahel precipitation stabilizes after ~2050–2060, evidencing rapid mitigation impact by damping the fast response. Western Sahel precipitation continues to decline due to the slow response, showing limited near-term mitigation benefit there. Differences between SSP585 and SSP534_over in western Sahel are associated with moisture flux convergence and local SST/circulation changes not fully captured by global fast/slow scaling. - Timescale for benefits: The study indicates mitigation benefits materialize within a few decades (on the order of 30 years) after emission reductions, particularly for central Sahel monthly precipitation, and likely for extremes via reduced temperature gradients. - Uncertainty partitioning: The dominant source of uncertainty in central Sahel precipitation change is linked to the fast response and is model-dependent; internal variability is substantial but largely filtered with the methodology and remains approximately constant in variance under warming in the 21-year mean framework.

By separating fast atmospheric and slow ocean-mediated components of the Sahel rainfall response, the study clarifies the temporal evolution and mechanisms behind the robust zonal contrast projected under climate change. The fast response, which dominates early to mid-21st-century changes and strongly influences central Sahel wetting, is sensitive to model physics and thus the primary contributor to projection uncertainty. Conversely, western Sahel drying depends more on the slower ocean response and is less immediately influenced by rapid GHG mitigation. These insights explain why different parts of the Sahel respond differently over time and why mitigation can yield tangible regional benefits within decades, especially in central Sahel. The decomposition improves interpretability of multi-model projections, supports targeted adaptation planning by region and timeframe, and highlights the importance of modeling the fast atmospheric adjustments accurately to reduce uncertainty. Additionally, while the analysis focuses on monthly means, the physical link between temperature gradients and convective storms suggests mitigation could also reduce extremes, an important implication for risk management.

The paper demonstrates that Sahel precipitation changes can be effectively decomposed into fast CO2/land-warming-driven atmospheric adjustments and slower ocean-mediated SST responses. This framework reproduces the robust zonal contrast: increased central Sahel rainfall and decreased western Sahel rainfall. Crucially, the fast response dominates near-term changes and uncertainties and can be rapidly attenuated by aggressive GHG mitigation, delivering benefits within a few decades, notably stabilizing central Sahel rainfall under SSP534_over. Western Sahel drying, tied to the slow ocean response, is less responsive to near-term mitigation. The approach enhances confidence in projections by attributing mechanisms and timescales, and underscores the urgency and efficacy of immediate emissions reductions for West Africa. Future research should refine representation of fast atmospheric adjustments across models, quantify aerosol influences and their seasonality, resolve regional SST pattern effects (especially in the tropical Atlantic and Pacific), extend analysis to precipitation extremes and subseasonal dynamics, and explore policy-relevant mitigation and adaptation pathways under different emissions trajectories.

- Scaling assumption: The reconstruction assumes Sahel precipitation scales linearly with GMST via fixed fast/slow patterns, which may overestimate magnitudes and miss nonlinearity or state dependence. - Definition sensitivity and interaction: Fast/slow separation depends on chosen time windows in abrupt4xCO2; fast and slow processes can interact and may not be strictly independent. - Aerosols not explicitly accounted for: Anthropogenic aerosols can affect Sahel rainfall seasonality and regional patterns, contributing to reconstruction errors, especially in western Sahel. - Regional SST pattern dependencies: Local tropical Atlantic SST changes and circulation features may not scale well with global fast/slow patterns, limiting reconstruction fidelity in western Sahel. - Data and model spread: The dominant uncertainty arises from model dependence of the fast response; limited ensemble sizes for some scenarios (e.g., SSP534_over with 7–8 models) constrain robustness. - Focus on monthly means: The study does not directly analyze extremes; implications for high-frequency events are inferred rather than explicitly quantified.