Stable decline in global surface relative humidity masks the distinct intensification of extremes

Global warming is profoundly reshaping the terrestrial water cycle. Relative Humidity (RH), serving as a critical nexus between the water and carbon cycles, plays a pivotal role in maintaining ecosystem stability. Although a consensus exists regarding the long-term decline in global surface RH, focusing exclusively on the mean state often masks the asymmetric amplification of extreme RH events in terms of frequency and intensity, potentially leading to an underestimation of future climate risks. Based on ERA5-Land reanalysis data from 1980–2023, this study systematically evaluates the spatiotemporal characteristics of extreme low (RH05d) and extreme high (RH95d) RH events and unravels their driving mechanisms using detrended partial correlation analysis. Our study find that the significant decreasing trend in global land surface RH (−0.49%/decade) is primarily driven by the surge in extreme low RH events. Over the past 44 years, the evolution of extreme RH events has exhibited distinct asymmetry: the frequency of extreme low RH events has increased significantly (0.22 days/year), a rate approximately three times that of the decrease in extreme high RH events. This intensification is statistically significant across 47.2% of global land pixels, particularly concentrated in the Amazon, Central Africa, and the mid-to-high latitudes of the Northern Hemisphere. Attribution analysis confirms that this asymmetry stems from a "mechanistic divergence": the intensification of extreme low RH events is dominantly driven by thermodynamic factors (temperature and radiation), reflecting the surge in "atmospheric water demand" caused by the exponential increase in Vapor Pressure Deficit (VPD) under warming. Conversely, extreme high RH events are strictly limited by "moisture supply constraints"; the supplementation rates of precipitation and soil moisture fail to keep pace with the rising thermodynamic demand, thereby suppressing the occurrence of high-humidity events in most regions. The "mechanistic divergence" framework proposed in this study elucidates the non-linear response of RH from its mean state to its extremes. This finding provides a novel physical perspective for understanding the evolution of extreme humidity under non-stationary climate conditions and offers a scientific basis for overcoming the limitations of the traditional mean-state perspective to accurately assess the asymmetric eco-hydrological risks under global warming.