Conference Agenda

The harmony of human-water relationships is a central issue in global sustainable development. In the Poyang Lake watershed, major challenges remain in advancing resilient wetlands and promoting harmonious human-water relationships. These challenges include identifying critical thresholds and evaluating wetland resilience under drought and flood conditions, simulating and estimating carbon cycle dynamics under the combined influences of global climate change and human activities, and improving understanding of human–water relationships across the watershed. In 2025, the project team achieved significant research progress in all these areas.

Several achievements have been obtained in the monitoring and evaluation of resilient wetland environments under flood and drought conditions. First, new parameter inversion algorithms were proposed through data fusion methods and the integration of supervised classification algorithms. Second, several new datasets and products for evaluating resilient wetland environments have been established through various methods. For example, a multispectral dataset of invasive species was constructed using UAV-based multispectral data, a long-term high-resolution nighttime light dataset was obtained using downscaling methods, and a daily water surface coverage dataset was constructed by integrating multi-source remote sensing images. Finally, we overcame the traditional limitations of equidistant division of thermal infrared broadband and the replication of satellite spectral band. To address the pathological coupling problem of temperature-emissivity, the Quantum Genetic Algorithm (QGA), combined with the Hyperspectral Reconstruction (HSR) model, was developed. At the same time, a clustering method based on temperature-emissivity correlation coefficients was proposed. These findings provide the theoretical foundation for the development of monitoring equipment.

To reveal the human-water relationship in the watershed and achieve harmonious development, significant progress has been made at the following aspects. First, the hydrological and water resource processes were explored at the watershed scale; Second, the impacts of human activities and vegetation greenness on total primary productivity were analyzed; Finally, geometric models of wetland vegetation, migratory birds, and watershed landscapes were constructed using digital twin technology to create multi-scale virtual scenes. GeoAI methods were further integrated to support the combination of virtual and real geographic experiments and management applications.

In addition to these scientific achievements, the project contributed significantly to the cultivation of young scientific talent. Seven young scientists were trained through the above research activities and played important roles in advancing the project’s major achievements. Among them, three young scientists from China have submitted two poster presentations (by Chen Liqiong and Tang Wenchao) and one oral presentation (Zheng Liang) to this mid-term symposium.

Grassland shrub encroachment is a widespread issue in global arid and semi-arid regions, posing a great threat to ecological environments and livestock production. Accurate shrub cover estimation is crucial for assessing encroachment extent and dynamics, particularly in severely impacted typical steppes. However, monitoring this process using coarse-resolution satellite sensors is challenging due to the typically low stature of shrubs and their intermixed growth with herbaceous vegetation. To address this, the Typical Steppe Shrub Encroachment Index (SSEI) was developed by mathematically transforming and integrating optical and Synthetic Aperture Radar (SAR) remote sensing features sensitive to shrub coverage. The SSEI leverages phenological differences between shrubs and herbaceous vegetation to identify the optimal monitoring window, effectively enhancing the weak signal of shrubs. Results show that the SSEI, achieves a significantly higher correlation (R = 0.76, p < 0.01, n = 758) and lower bias (RMSE = 3.5%, MAE = 2.75%) with shrub cover than individual remote sensing features and exhibits better sensitivity to changes in shrub cover. This index enables accurate discrimination between shrub-encroached grasslands and shrub-free grasslands when shrub cover reaches ≥10%. Furthermore, the relative vegetation volume scattering q-factor was introduced to complement the SSEI, effectively distinguishing shrub-encroached areas from haymaking fields via a simple threshold. This study demonstrates the value of fusing optical and microwave data for shrub cover extracting in heterogeneous landscapes at coarse resolutions. Requiring no training samples or complex parameterization, the SSEI provides a practical tool for large-scale monitoring of grassland shrub encroachment.