PNNL

Abstract

Land surface relative humidity (RH) is a critical variable in the coupled land-atmosphere system whose changes have profound impacts on terrestrial hydroclimate and extremes. However, historical trends in land RH and their representations in Earth system models (ESMs) are not well understood, due to limited observations and the lack of a theoretical framework for interpreting land RH changes under the influence of multiple physical drivers. Here, we address both challenges by introducing a theoretical wetness index, the ratio of land precipitation (P) to potential evapotranspiration that assumes an asymptotic RH over unlimited water availability (PETo). The index, P/?PET?^o, strongly correlates with land RH, capturing its spatiotemporal variation and its distinct historical trends across observations, reanalyses, and ESMs. Being observationally derivable, the index allows us to calibrate RH in reanalyses and construct a reliable global, long-term RH record. The calibration corrects the overestimated land drying in reanalyses, which is consistently reflected in the exaggerated precipitation decline and surface warming. By linking land RH to fundamental physical factors, the index provides a quantitative framework for understanding RH trends in ESMs. ESMs simulate a wide range of historical land RH changes but systematically underestimate the land drying. The index explains both the model spread and bias, attributing them to trends in precipitation and PETo. The underestimated historical drying is primarily due to weaker subtropical precipitation decline, which is dynamically linked to muted subtropical high intensification. These biases extend into future projections, suggesting a substantially drier land and higher risks of RH-modulated extremes than ESM projections.