A novel multi-model fusion framework diagnoses the complex variation characteristics of ecological indicators and quantitatively reveals their driving mechanism.

Systematic analysis of the change law and driving mechanism of ecological indicators (GPP, ET, WUE), as well as the study of maximum threshold of water resources benefit changing with ecological benefit, are important prerequisites for realizing the scientific allocation and efficient utilization of water resources in desert riparian forests. However, previous studies have defects in the detailed description of the change characteristics of ecological indicators. How to accurately diagnose the characteristics of a site, mutation year, pattern (linear, exponential, logarithmic, etc.), duration of change, future change trends of ecological indicators in a desert riparian environment, as well as quantitatively revealing their driving mechanisms, are major scientific problems that need to be solved urgently. In this regard, an ensemble function coupling a logistic function and an asymmetric Gaussian function was creatively adopted, a novel framework was created to integrate the time-series trajectory fitting method and the sensitivity analysis method, and the arid and ecologically fragile Tarim River Basin was taken as a typical area. The results showed that with enhanced water resource management in the Tarim River Basin, GPP, ET, and WUE all showed patterns of increasing change and could be expected to continue to rise or to remain at a high-level stable state. The longest continuous period of GPP change was 15 years, showing that ecological restoration is a long-term process. The years of GPP mutation were consistent with the implementation periods of major measures in the Tarim River Basin (1990, 2001, and 2011), indicating the reliability of this framework. More importantly, when GPP increased to 216.44 gCm-2, the maximum WUE threshold of 0.93 gCm-2mm-1 occurred. This threshold can be used as a reference criterion for efficient utilization of ecological water in the basin. Among the ecological indicators studied, GPP was the most sensitive to environmental change, but GPP, with 80.60% of pixel area, showed a weak memory effect（α < 0.4). Besides, GPP was the most sensitive to the leaf area index (LAI) and had the strongest correlation with it (p < 0.001). Therefore, LAI can be used as the main control factor for judging plant growth. This research can provide important scientific guidance and reference for the analysis of ecological indicator changes and the sustainable utilization of water resources in arid areas.

Past and projected future patterns of fractional vegetation coverage in China.

With the intensifying climate change and the strengthening ecosystem management, quantifying the past and predicting the future influence of these two factors on vegetation change patterns in China need to be analyzed urgently. By constructing a framework model to accurately identify fractional vegetation coverage (FVC) change patterns, we found that FVC in China from 1982 to 2018 mainly showed linear increase (29.5 %) or Gaussian decrease (27.4 %). FVC variation was mainly affected by soil moisture in the Qi-North region and by vapor pressure deficit in other regions. The influence of environmental change on FVC, except for Yang-Qi region in the southwest (-2.0 %), played a positive role, and weakened from the middle (Hu-Yang region: 2.7 %) to the northwest (Qi-North region: 2.4 %) to the east (Hu-East region: 0.8 %). Based on five machine learning algorithms, it was predicted that under four Shared Socioeconomic Pathways (SSPs, including SSP126、SSP245、SSP370、SSP585) from 2019 to 2060, FVC would maintain an upward trend, except for the east, where FVC would rapidly decline after 2039. FVC in the eastern region experienced a transition from past growth to future decline, suggesting that the focus of future ecosystem management should be on this region.