Models overestimate ecosystem water use efficiency for northern permafrost regions

Understanding the carbon-water coupling over permafrost regions is essential to projecting global ecosystem carbon sequestration and water dynamics. Ecosystem water use efficiency (EWUE), defined as the ratio of gross primary productivity (GPP) and evapotranspiration (ET), reflects plant acclimation strategies with varying ecosystem functioning against environmental stress. Yet EWUE change and its potential drivers across the northern permafrost regions remain poorly quantified, hampering our understanding of permafrost carbon−climate feedback. Here, we compared and analyzed the difference using satellite observations and process-based models to estimate the spatio-temporal variations of EWUE in 1982–2018 over northern permafrost regions. Using flux measurements as truth data, satellite-derived EWUE was more reliable than model-based EWUE. Satellite-derived EWUE showed biome-dependent spatial patterns, with a steady temporal trend (0.01 g C mm−1 decade−1, P > 0.05) for spatially averaged EWUE over northern permafrost regions. Carbon dioxide (CO2) concentration and nitrogen deposition positively affected interannual variations of EWUE, while vapor pressure deficit and other climatic factors (i.e., temperature, precipitation, and radiation) negatively controlled EWUE. Compared to satellite-derived EWUE, we found that EWUEs derived from an ensemble of process-based carbon cycle models are overestimated in seven out of ten models, with an increasing trend of 0.11 g C mm−1 decade−1 (P < 0.001) for spatially averaged EWUE of the ensemble mean. The relationships between climatic factors and EWUE are partially misinterpreted in model estimates, especially with overstated CO2 sensitivity and the opposite temperature effect. The fluctuating sensitivities to climate over time and the diminishing effect of CO2 fertilization on gross primary productivity (GPP) may partially explain the discrepancy observed between satellite-derived and model-based estimates of EWUE. Thus, this study calls for caution concerning model-based EWUE and aids in understanding permafrost-climate feedbacks and projections of carbon and water cycles.