Researchers led by CAS member ZHU Jiaojun from the Institute of Applied Ecology of the Chinese Academy of Sciences (CAS) have reported a series of studies clarifying how complex wind regimes in mountainous forests affect carbon flux measurements.

Drawing on long-term observations from eddy covariance (EC) flux towers within a 536-hectare temperate forest watershed, these studies address long-standing methodological challenges associated with applying the EC technique in complex terrain.

The EC method is the only approach that can directly measure the exchange of carbon dioxide, water vapor, and energy between ecosystems and the atmosphere. Although it is widely used in flat terrain, its application in mountainous forests has been uncertain. In complex terrain, highly heterogeneous wind fields, driven by topography and thermal processes, complicate data interpretation and quality control. Consequently, measurements from a single flux tower often fail to adequately represent the spatial structure of airflow in mountainous forests.

To address these limitations, the researchers conducted a series of investigations at the Qingyuan-Ker Towers, a forested watershed covering approximately 536.4 ha with three observation towers. Each tower is located in a distinct forest type—a natural deciduous mixed forest, a natural Mongolian oak forest, and a larch plantation—providing a unique opportunity to examine wind regimes and carbon flux dynamics across contrasting forest ecosystems within the same mountainous watershed.

In the first study, the researchers evaluated the applicability of commonly used wind speed profile models in mountainous forests. These models are largely derived from Monin–Obukhov similarity theory and are commonly used in flat terrain. They are also critical for estimating the aerodynamic parameters required for energy-conservation (EC)-based flux calculations.

The results showed that the models performed poorly at all three forest towers in Qingyuan, exhibiting low explanatory power and large inaccuracies, while they worked well at a cropland site located in the flat Panjin Plain. The discrepancies were primarily attributed to topographic influences, with seasonal differences between leaf-on and leaf-off periods also contributing.

Key aerodynamic parameters, including zero-plane displacement height and surface roughness length, were systematically underestimated in complex terrain, resulting in inaccurate estimates of aerodynamic height. The findings, published in Chinese Journal of Applied Ecology, provide guidance for improving parameter calibration in EC studies conducted in mountainous regions.

In the second study published in Agricultural and Forest Meteorology, the researchers used data from the three towers to characterize wind regimes and their driving forces. The results indicated that thermally driven mountain–valley wind systems were strongly modulated by regional weather conditions.

During the day, downslope winds dominated above the canopy at two towers, while upslope winds prevailed at the third tower. At night, each site showed a distinct dominant flow regime. Pronounced vertical wind direction shear was frequently observed, suggesting a decoupling between the airflow above the canopy and within the canopy layer. Classical diurnal reversals of mountain–valley winds occurred on only about 10% of days, reflecting the combined influence of synoptic-scale winds and local thermal circulations, which weakened the idealized pattern.

Further analysis revealed that nighttime drainage flows occurred more frequently at the larch plantation, which could lead to an underestimation of net ecosystem CO₂ exchange. Such effects were weaker at the other two towers. These findings provide a scientific basis for improving nighttime flux correction in mountainous forest ecosystems.

Based on these insights, the researchers proposed an improved strategy for processing and gap-filling eddy covariance CO₂ flux data in complex terrain. Due to the strong directional heterogeneity of the surrounding CO₂ sources and sinks, conventional data processing approaches often introduce substantial bias.

Dividing wind directions into discrete sectors and estimating friction velocity thresholds separately for each sector markedly improved the accuracy of nighttime flux filtering. This sector-based approach significantly reduced bias in the final flux estimates. Of the gap-filling methods evaluated, random forest models demonstrated the highest accuracy and robustness, particularly for annual net ecosystem exchange estimates.

This work, published in Science China Earth Sciences, highlights the critical importance of adopting terrain-sensitive data processing strategies when assessing forest carbon sinks at regional and global scales.

The studies above systematically elucidate the impact of the wind regime on flux observations from three perspectives: wind speed prediction, driving mechanisms, and advances in carbon flux estimation methods. These efforts provide new insights into the wind regime in mountainous forests and improve the accuracy of carbon flux data.