Current Earth System Models Overestimate Ecosystem Respiration in Mid‐To‐High Latitude Dryland Regions

Abstract

Abstract The inhibition of foliar respiration by light is a crucial yet often overlooked component in estimating ecosystem respiration. However, current estimations of the light inhibition of ecosystem respiration are biased by ignoring the effects of moisture factors. We developed a novel physics‐constrained machine learning method to quantify the extent of light inhibition (Reli) driven by multiple factors in global ecosystems. Our findings revealed significant seasonal variations in light inhibition rate aligned with vegetation growth. Temperature predominantly influenced variations in Reli, and the temperature‐Reli relationship was regulated by vapor pressure deficit rather than soil water content. A reassessment of global ecosystem respiration revealed that current Earth system models (ESMs) overestimate ecosystem respiration in mid‐to‐high latitude dryland regions, with a global average light inhibition strength of 0.51 (±0.16). Knowledge from this study provides an accurate understanding of light inhibition driven by temperature and moisture coupling in simulating carbon cycle.

Keywords

• ecosystem respiration
• light inhibition of respiration
• machine learning
• carbon cycle
• Earth system model