Geping Luo’s research while affiliated with State Key Laboratory of Desert and Oasis Ecology, Chinese Academy of Sciences and other places

Water resources are the lifelines of the agricultural development in Xinjiang. Currently, the problem of water shortage for agriculture in this region is becoming increasingly severe. Against this backdrop, predicting the changing trends of water supply and agricultural water use in Xinjiang and analyzing the supply and use relationship between them are of great practical significance for ensuring the sustainable development of regional agriculture. Firstly, we conducted an in-depth analysis of the water supply and agricultural water use patterns in Xinjiang over the past two decades. Secondly, we evaluated and compared several mainstream water resource prediction models, ultimately developing a novel GM(1,1)-NN essemble model. Validation results demonstrated that this model exhibits superior accuracy in forecasting water supply and agricultural water use compared to other existing models. Finally, we utilized the newly developed GM(1,1)-NN essemble model to predict short-term water supply and agricultural water use trends in Xinjiang. Based on these findings, we proposed recommendations for water resource conservation from both technological and regional planting perspectives. The key results are as follows: (1) There are significant regional disparities in water resources in Xinjiang, primarily attributed to uneven precipitation distribution and imbalanced economic development. (2) The GM(1,1)-NN essemble model demonstrates high short-term predictive accuracy for both water supply and agricultural water use in Xinjiang. (3) According to our GM(1,1)-NN essemble model’s projections, both water supply and agricultural water use in Xinjiang are expected to exhibit a downward trend in the coming years. The reduction in agricultural water use will help allocate more water resources to non-agricultural sectors. (4) Despite these improvements, the contradiction between water shortage and the high proportion of agricultural water use (approaching to 88%) remains unresolved. Therefore, it is recommended to reduce agricultural water use through the widespread adoption of water-saving facilities and the optimization of crop planting structures across different regions.

Grasslands on the Qinghai-Tibet Plateau play a crucial role in carbon sequestration and livestock farming. However, carbon storage and the environmental effects of grazing in Qinghai-Tibet Plateau grasslands are not well understood. Here, we utilised the advanced Biome-Biogeochemical Cycles with Multi-layer Soil Module model to evaluate carbon storage and responses to grazing in Qinghai-Tibet Plateau grasslands between 1979 and 2018. The average annual vegetation carbon density and soil organic carbon density were found to be 46.07 ± 7.19 gC/m² and 3,789.79 ± 17.08 gC/m², respectively. Grazing resulted in the loss of 21.63 tg of vegetation carbon, while soil organic carbon loss was 108.83 tg in 2018. Grazing was found to have reduced both vegetation carbon density and soil organic carbon density in most grasslands of the Qinghai-Tibet Plateau. The study findings highlight the extent of carbon loss caused by unreasonable grazing practices.

Plain Language Summary Over the past two decades, China has faced increasingly frequent extreme events driven by global climate change, alongside significant changes in land cover. However, assessing the impacts of these changes on carbon exchange remains challenging due to limited observational data. This study analyzed carbon exchange dynamics in China from 2003 to 2018 using a carbon flux product from Eurasian meteorological stations. The findings revealed that, on an annual scale, forests in southwestern China exhibited the highest carbon sink capacity, while forests in central China demonstrated the strongest capacity during the growing season. Notably, carbon sink capacity significantly increased at northern forest sites located within national forest parks. Furthermore, most sites situated in China's national ecological function zones achieved enhanced carbon sink capacity following land cover changes. These results highlight the important role of site‐level assessments in supporting China's dual carbon goals. Additionally, our analysis of post‐drought carbon exchange recovery emphasizes the uncertainty introduced by the selection of recovery thresholds when evaluating drought impacts. Although the uneven distribution of sites in this carbon flux product poses challenges for regional studies, it offers valuable insights for site‐level assessments.

The scarcity and uneven global distribution of eddy covariance (EC) towers are the key factors that contribute to significant uncertainties in carbon cycle studies of terrestrial ecosystems. To address this limitation of EC towers, Zhang et al. (2023b) developed a meteorological station-based net ecosystem exchange (NEE) dataset. This dataset includes 4674 global meteorological stations, representing a 22-fold increase compared to the 212 existing EC towers and covering a broader range of ecosystem types. Here, we propose a systematic framework for the comprehensive assessment of spatio-temporal representativeness and global uncertainty of the meteorological station-based carbon flux dataset. Meteorological stations effectively enhance the spatial representativeness of the EC towers and reduce the latitudinal variability of the spatial representativeness. In most regions, the temporal trends of carbon flux data from meteorological stations did not significantly differ from those observed by EC towers (p < 0.001). The global uncertainty of carbon fluxes from meteorological station is 0.37, followed by the VISIT and FLUXCOM products with uncertainties of 0.44 and 0.45, respectively. Overall, the carbon fluxes from meteorological stations exhibit higher spatial representativeness and better temporal representativeness compared to the EC tower observations and possess lower global uncertainties than the existing carbon flux gridded products. Consequently, the carbon flux data derived from meteorological stations is a trade-off dataset that addresses the low spatial representativeness of the EC towers and the high uncertainty of the gridded products. It effectively complements the existing EC tower data while ensuring accuracy. The development of this dataset will play an important role in reducing the uncertainty of global carbon sink-related studies.

The net ecosystem exchange (NEE), determining terrestrial carbon sequestration capacity, is strongly controlled by climate change and has exhibited substantial year-to-year fluctuations. How the increased frequency and intensity of warm extremes affect NEE variations remains unclear. Here, we combined multiple NEE datasets from atmospheric CO2 inversions, Earth system models, eddy-covariance data-driven methods and climate datasets to show that the terrestrial carbon sequestration capacity is weakened during warm extreme occurrences over the past 40 years, primarily contributed by tropical regions (81% ± 48%). The underlying mechanism can be rooted in the overwhelmingly decreased trend of gross primary productivity compared with terrestrial ecosystem respiration. Additionally, the weakened terrestrial carbon sequestration capacity is mainly driven by the transition from temperature or soil moisture control to vapour pressure deficit control, which is associated with the increasing intensity of warm extremes. Our findings suggest that warm extremes threaten the global carbon sequestration function of terrestrial ecosystems. Therefore, more attention should be given to the evolution of the increasing intensity of warm extremes in future climate projections.