1 March 2020 in ISPRS Journal of Photogrammetry and Remote Sensing
1 August 2020 in ISPRS Journal of Photogrammetry and Remote Sensing
30 September 2020 in Journal of Geophysical Research: Biogeosciences
Terrestrial gross primary production (GPP) is a key indicator of the ecosystem response to climate change and land use/cover change (LUCC) in arid areas. The available global GPP data sets cannot meet the demands for local applications in arid areas due to sparse vegetation and extreme climate conditions. Here, we developed a novel GPP estimation model for the Heihe River Basin (HRB), the second largest inland river basin in northern China, and disentangled the impacts of climate change and LUCC on GPP. First, we calibrated the vegetation photosynthesis model (VPM) using CO2 flux observations from multiple stations in the HRB and developed a modified local GPP model (HRB‐VPM). Then, we decoupled the joint effects of LUCC and climate change on GPP based on the log and differential transformation method. The results showed that HRB‐VPM outperformed Moderate Resolution Imaging Spectroradiometer and VPM GPP models in arid ecosystems. The root mean square error of HRB‐VPM was 4.9 and 1.5 gCm−2 day−1 lower than those of the Moderate Resolution Imaging Spectroradiometer and VPM models, respectively. We concluded that the underlying driving forces of the GPP changes were distinct across the HRB. In the upper reach, climate change accounted for 65.8% of GPP changes, while in the middle and lower reaches, LUCC contributed to 75.1% of GPP changes. Our research provides an effective way to monitor arid ecosystem degradation and is useful for mitigating the negative impacts of human activities and future climate change.
1 January 2026 in Journal of Remote Sensing
Center-pivot irrigation systems (CPISs) represent a water-efficient irrigation technology, yet their rapid expansion can exacerbate the trade-off between food production and groundwater sustainability in water-scarce regions. In China, the absence of continuous, long-term observations of CPISs has limited our ability to quantify this food–water trade-off. To address the challenges of CPIS extraction in heterogeneous agricultural landscapes, we propose an EfficientNet–UNet framework equipped with a focal Tversky loss and a PointRend (Point-based Rendering) refinement module (ETP-UNet). Experimental results demonstrate that ETP-UNet achieved an intersection over union (IoU) of 81.0%, an F1-score of 89.5%, and a boundary IoU of 66.0%. Using ETP-UNet, we generated a long-term China CPIS (CCPIS) dataset at 3-year intervals from 2000 to 2024, achieving overall accuracies above 0.88 and macro-F1-scores exceeding 0.87. A sample-based area estimator indicates that the CPIS area expanded from 23 ± 18 km 2 in 2000 to 2,202 ± 64 km 2 in 2024, representing an approximately 96-fold increase. This expansion was concentrated in semiarid regions and was most rapid during 2009 to 2015, coinciding with strengthened policy incentives for irrigation development. New CPISs were established predominantly on bare land (53%) and grassland (32%), with a smaller proportion from existing cropland (15%). CCPIS provides a critical foundation for jointly assessing CPIS-driven production gains and groundwater-depletion risks, supporting strategies that strengthen both food security and freshwater sustainability in arid and semiarid China.