Enhanced soil moisture-temperature coupling could exacerbate drought under net-negative emissions

The study addresses how extreme droughts may evolve under climate mitigation pathways that aim for net-zero and net-negative CO₂ emissions. While the Paris Agreement emphasizes limiting warming via CO₂ reductions, the response of hydroclimatic extremes—particularly drought—to CO₂ removal remains uncertain. Drought is a high-impact hazard with rising frequency, severity, and spatial extent under warming. A key mechanistic uncertainty lies in land–atmosphere (LA) feedbacks, notably soil moisture–temperature (SM–T) coupling, which can intensify heat and dryness via reduced evapotranspiration, increased vapor pressure deficit, and altered precipitation. Plant physiological responses to CO₂ (e.g., stomatal conductance and leaf area changes) further modulate water and energy fluxes, complicating predictions. The study aims to quantify how global drought evolution and characteristics change under idealized net-zero and net-negative CO₂ pathways and to diagnose the role of SM–T coupling in drought onset and propagation, with attention to regional disparities.

Prior work highlights the importance of LA feedbacks in extreme events and the strong negative correlation between soil moisture and temperature, especially in dry regimes. SM deficits enhance sensible heating, increase atmospheric aridity (VPD), and can suppress precipitation, creating feedbacks that intensify droughts and heatwaves. Physiological effects of CO₂ on vegetation (stomatal closure, changes in transpiration, greening, albedo) can either cool via enhanced latent heat flux or warm via reduced transpiration and altered radiation balance; these effects influence humidity and precipitation patterns. Studies project intensified LA coupling with warming, and observations/models identify transitional climate regions (e.g., parts of Africa, South Asia, the Americas) as hotspots of strong coupling. The literature also notes that CO₂ has unique radiative properties and long lifetime; interest in negative emissions (afforestation, soil carbon, DAC, BECCS) is growing, but feedbacks during drought under CO₂ reduction phases are not well constrained. Previous modeling indicates potential declines in precipitation during CO₂ decrease phases in regions like the South Asian monsoon and parts of Africa, suggesting risk of regional drought despite global mitigation.

The study uses CESM2 to conduct idealized, emission-based simulations of CO₂ pathways featuring three phases: (1) linear increase in anthropogenic CO₂ emissions from 2000 to 2050 (~1.09 Gt CO₂ yr⁻¹ growth), (2) subsequent decrease until atmospheric CO₂ returns to the year-2000 level (383 ppm) by 2197, and (3) restoration thereafter. Net-zero and net-negative emission targets are achieved after 2123: net-zero emissions are maintained from 2124 to 2197; net-negative forcing implies net atmospheric CO₂ removal from 2124 to 2197. Simulations are transient and run at 1° resolution with nine ensemble members; ensemble means are analyzed. The primary analysis focuses on 60-year periods (2131–2190/2197) for net-zero and net-negative scenarios, compared to a reference period (2000–2030). Model variables include precipitation, temperatures, radiation, wind, relative humidity, evaporation, and soil moisture (SM). Reference-period CESM2 outputs were evaluated against ERA5 (2000–2020) and found broadly consistent. Robustness was assessed against CESM2 Large Ensemble (CESM2-LE; 50 members used for comparison) and CMIP6 CDRMIP CO₂-reversibility experiments (multi-model mean of CanESM5, CNRM-ESM2-1, GFDL-ESM4, MIROC-ES2L) for spatial patterns of precipitation, PET, and drought metrics. Drought metric: SPEI at 12-month scale (SPEI-12) computed from water balance D = P − PET, with PET estimated via a modified Penman–Monteith formulation that incorporates CO₂ effects on surface resistance. Drought characterization follows run theory: an event occurs when SPEI ≤ −1 for at least three consecutive months; duration (months per event), frequency (events per year), intensity (mean SPEI during event), and area coverage (maximum grid fraction in drought per period) are derived. SM–T coupling metric π quantifies LA coupling on monthly timescales: π = ε′ × T′ = [(R_n − λE)′ − (R_n − λE_p)′] × T′, where positive π indicates enhanced coupling associated with SM scarcity increasing sensible heat. Coupling trends and composites were examined for drought years (SPEI ≤ −1) versus non-drought years (SPEI > 0), and by drought severity (moderate, severe, extreme). Spatial trends were analyzed for precipitation, SM, and PET to interpret coupling–drought linkages. Statistical significance was assessed at 95% confidence; kernel density distributions and correlations among drivers were calculated. Limitations concerning PET estimation and autocorrelation in drought series were acknowledged.

• Global dryness trends: Under mitigation, drying is more extensive and stronger under net-negative than net-zero forcing. Total (significant) areas with decreasing SPEI-12 cover ~45% (~17%) of land for net-zero and ~47% (~38%) for net-negative. Hotspots include Africa (western, northeastern), South Asia, parts of Russia, Greenland, Central America, Western Australia. Regional disparities exist: South and North America tend to benefit more (weakened dryness) under net-negative, whereas Central America, Africa, and Asia benefit more under net-zero. CESM2 patterns during net-negative broadly match CMIP6 CDRMIP CO₂-decrease phase.
• Drought characteristics (relative to reference): Under net-negative emissions, global mean increases exceed 66% in duration, 68% in frequency, and 74% in intensity; under net-zero, increases are ~64%, 67%, and 71%, respectively. Severe intensification is projected in Western/Southern Africa, South/Central Asia, Middle East, Russian Far East, Mediterranean, Western Australia, Central America, and parts of the Americas. Extreme-drought frequency and area may decrease under net-negative, but severe-drought duration and intensity still increase.
• SM–T coupling: Coupling strengthens in drought-prone regions in both scenarios, with larger and more widespread increases under net-negative, particularly in western/northeastern Africa and South Asia. During drought years, coupling strength increases markedly compared to non-drought years—over 4-fold (net-zero) and over 8-fold (net-negative). Coupling strength increases with drought severity (moderate < severe < extreme), underscoring a decisive LA role in drought onset and intensification.
• Hydroclimate drivers: Future precipitation and soil moisture decrease broadly, especially under net-negative. Precipitation declines occur over ~75% (~38% significant) of land for net-negative and ~64% (~3% significant) for net-zero. SM decreases track precipitation reductions, with significantly decreased SM area expanding by ~20% (net-negative) versus ~3% (net-zero). PET increases where Pr and SM decline; area with increased PET is larger under net-zero (~21% of land) than net-negative (~10%), but the magnitude of PET increase is greater under net-negative. Correlations indicate tight coupling among drivers (e.g., higher global correlation between Pr and SM under net-negative). Spatial patterns of Pr and PET changes under net-negative align with CMIP6 CDRMIP CO₂-decrease phase.
• Temporal evolution: With CO₂ reduction, temperature and PET tend to decline, but vegetation physiological responses and enhanced evapotranspiration during net-negative can deplete SM, increasing evaporative stress and reinforcing SM–T coupling and drought risk in susceptible regions.
• Policy-relevant insight: Net-zero generally mitigates future drought risk more effectively than net-negative, though both pathways yield region-specific benefits; CO₂ mitigation alone may be insufficient to manage drought risk without complementary water management strategies.

The findings demonstrate that the hydroclimate response to CO₂ mitigation is not uniformly beneficial for drought risk. Although net-zero and net-negative reduce radiative forcing, SM–T coupling intensifies over key transitional and drought-prone regions, especially under net-negative forcing, leading to stronger drying trends and worsened drought characteristics. Mechanistically, reduced precipitation, enhanced PET, and vegetation physiological adjustments under changing CO₂ modify evapotranspiration and energy partitioning, lowering SM and amplifying surface heating. This positive feedback increases atmospheric aridity and can suppress precipitation, promoting drought onset and persistence. The strengthened coupling during drought years—particularly under net-negative—and its scaling with drought severity directly links LA feedbacks to drought evolution under mitigation. Regional heterogeneity is pronounced: while the Americas see more relief under net-negative, Central Africa and South Asia are more vulnerable, and many regions fare better under net-zero. These results suggest partial climate irreversibility and delayed recovery in soil moisture relative to atmospheric CO₂ reductions, raising equity concerns as developing regions may bear higher residual risks. Consequently, achieving CO₂ mitigation targets must be complemented by adaptive water management, drought preparedness, and land-use strategies tailored to regional feedback sensitivities.

This study shows that net-zero emissions are generally more effective than net-negative emissions at alleviating future drought risk, as net-negative pathways can intensify soil moisture–temperature coupling, enhance PET, and exacerbate drought characteristics in several vulnerable regions. Despite overall mitigation, drying trends, longer and more frequent droughts, and stronger coupling persist or worsen in parts of Africa and South Asia, while some regions in the Americas may benefit under net-negative. The work highlights the central role of LA feedbacks in shaping drought under CO₂ removal and underscores that CO₂ mitigation alone is insufficient to manage drought risks. Future research should refine PET estimation under varying CO₂ and vegetation states, explicitly address serial autocorrelation in drought metrics, better resolve regional processes with higher-resolution and larger-ensemble modeling, and investigate compound and flash drought dynamics and soil moisture memory under CO₂-reduction phases. Policy should pair mitigation with advanced, region-specific water management and adaptation strategies.

• PET estimation uses a modified Penman–Monteith approach that may not fully capture complex vegetation responses (stomatal conductance, leaf area dynamics, seasonality) to CO₂ changes.
• Drought time series exhibit strong autocorrelation; the study did not explicitly adjust for serial correlation, potentially affecting effective sample sizes and statistical robustness.
• Model-dependent uncertainties remain: CESM2’s relatively coarse resolution, limited ensemble size (nine members), transient experimental design, and differences in periods and forcings versus comparison datasets (CESM2-LE, CMIP6 CDRMIP) can contribute to regional discrepancies.
• Idealized, emission-based CO₂ pathways may not reflect real-world socioeconomic trajectories or concurrent non-CO₂ forcings.
• SPEI-based drought characterization depends on PET formulation; alternative drought indices or multi-index approaches could alter quantitative results.