As one of the main geographical elements in urban areas, buildings are closely related to the development of the city. Therefore, how to quickly and accurately extract building information from remote sensing images is of great significance for urban map updating, urban planning and construction,etc. Extracting building information around power facilities, especially obtaining this information from high-resolution images, has become one of the current hot topics in remote sensing technology research. This study made full use of the characteristics of GF-2 satellite remote sensing images, adopted an object-oriented classification method, combined with multi-scale segmentation technology and CART classification algorithm, and successfully extracted the buildings in the study area. The research results showed that the overall classification accuracy reached 89.5% and the Kappa coefficient was 0.86. Using the object-oriented CART classification algorithm for building extraction could be closer to actual ground objects and had higher accuracy. The extraction of buildings in the city contributed to urban development planning and provided decision support for management.

Effective underwater sensing is crucial for environmental protection and sustainable energy transitions, particularly as we face growing challenges in marine ecosystem monitoring, resource management, and the need for efficient energy infrastructure. To support these efforts, we propose a multimodal sensing approach that enhances underwater detection and distance estimation by combining affordable sonar technology with stereo vision-based depth cameras. Our method integrates the Ping 360 single-beam sonar for target detection and distance measurement with depth refinement from the Intel RealSense D455 camera. A promptable segmentation model automates sonar target detection, overcoming challenges such as acoustic noise and shadowing without requiring large labeled datasets. Depth images from the stereo camera are enhanced using a Depth-Anything model, addressing underwater-specific issues like noise, missing regions, and light attenuation, achieving accurate depth maps for distances up to 1.2 meters underwater. By leveraging multimodal sensing, this approach not only improves underwater robotics for navigation, manipulation, and exploration but also plays a key role in monitoring and maintaining energy infrastructure, such as offshore wind farms and underwater pipelines. Accurate, real-time sensing of these installations ensures more efficient operations, minimizes the environmental impact, and aids in the sustainable management of ocean resources. This enables better energy production and resource utilization, which are essential for a smarter and more sustainable energy transition.

In the continually evolving landscape of data-driven methodologies addressing car crash patterns, a holistic analysis remains critical to decode the complex nuances of this phenomenon. This study bridges this knowledge gap with a robust examination of car crash occurrence dynamics and the influencing variables in the Greater Melbourne area, Australia. We employed a comprehensive multi-model machine learning and geospatial analytics approach, unveiling the complicated interactions intrinsic to vehicular incidents. By harnessing Random Forest with SHAP (Shapley Additive Explanations), GLR (Generalized Linear Regression), and GWR (Geographically Weighted Regression), our research not only highlighted pivotal contributing elements but also enriched our findings by capturing often overlooked complexities. Using the Random Forest model, essential factors were emphasized, and with the aid of SHAP, we accessed the interaction of these factors. To complement our methodology, we incorporated hexagonalized geographic units, refining the granularity of crash density evaluations. In our multi-model study of car crash dynamics in Greater Melbourne, road geometry emerged as a key factor, with intersections showing a significant positive correlation with crashes. The average land surface temperature had variable significance across scales. Socio-economically, regions with a higher proportion of childless populations were identified as more prone to accidents. Public transit usage displayed a strong positive association with crashes, especially in densely populated areas. The convergence of insights from both Generalized Linear Regression and Random Forest's SHAP values offered a comprehensive understanding of underlying patterns, pinpointing high-risk zones and influential determinants. These findings offer pivotal insights for targeted safety interventions in Greater Melbourne, Australia.

With the continuous acceleration of urbanization, the resulting habitat fragmentation poses a serious threat to regional ecological security. Constructing an ecological network not only facilitates the positive circulation of regional materials and energy but also holds significant importance for achieving regional sustainable development. Taking the Guangxi Beibu Gulf Economic Zone as the study area, and supported by GIS and other technologies, this study uses 10-meter high-resolution land use data from 2017 and 2021 to select ecological sources based on MSPA and landscape connectivity. Using the MCR-gravity model, a higher precision ecological network for the Guangxi Beibu Gulf Economic Zone is constructed. This network is then optimized and comparatively analyzed using circuit theory and network analysis methods. The results showed that: ① Although the core area of ecological sources accounts for 95.16% of the total ecological source area, a comparison of the two-year analysis results shows an increasing degree of landscape fragmentation and a decline in landscape connectivity. ② The optimized ecological network of the Beibu Gulf Economic Zone of Guangxi comprises 11 ecological sources, 91 ecological corridors, and 71 ecological nodes. ③ The indices of the optimized network structure have all improved, enhancing the degree of ecological network connectivity within the study area.

Wetland landscapes have undergone tremendous changes and the spatial heterogeneity of wetlands has increased. It's a huge challenge to accurately assess the effect of human disturbance on the landscape patterns in such a complex environment. This paper, taking the Yellow River Delta as a case, proposed a new framework to evaluate the heterogeneity effect of human disturbances on landscape patterns. A pixel-based Human Disturbance Index (HDI) with the addition of ecological conditions and the buffer influence is first established to quantify the spatial difference of human disturbances. Besides, Geographically Weighted Regression (GWR) model was introduced to analyze the spatial correlation between HDI and landscape indices, i.e., Shannon's Diversity Index (SHDI), Contagion Index (CONTAG), and Area-Weighted Mean Shape Index ($SHAPE_{AM}$), which strongly correlated with HDI. The results show that HDI in the Yellow River Delta has increased gradually and its spatial heterogeneity has continued to increase in the past 30 years. The increase of human disturbances mainly occurred in coastal areas due to H-level human disturbances in Dongying Port and M-H level human disturbances along the coast. But in most areas inland of study area, the HDI reduced slightly benefit from the wetland conservation measurements. The landscape pattern in the Yellow River Delta wetland has changed tremendously. The spatial heterogeneity of landscapes is enhanced, and the connectivity is reduced. Patches tend to be regularized. Different levels of human disturbance have different effects on the landscape patterns. The M-H level HDI leads to low landscape different and more connected. While the H level HDI leads to stronger landscape heterogeneity, complex patch shapes and decreased spatial connectivity. These implicate the proposed framework is efficient for evaluating the heterogeneity effect of human disturbance and landscape patterns in a complex wetland ecosystem. These methods and findings will provide suggestions and guidance for wetland conservation and management.

Precisely monitoring the range of rice cultivation is an essential task for the government to dynamically supervise the red line of 180 million mu ($ 1 mu \approx 666.667 {m^2} $) of arable land. This study aims to address the issues of low efficiency, high cost, and insufficient accuracy in traditional rice cultivation range monitoring methods. Against the backdrop of the widespread application of UAV remote sensing and the maturity of deep learning technology, this paper constructs a high-precision UAV remote sensing image dataset for rice identification, which includes different growth stages of rice, different resolutions, and regions. It also utilizes deep learning semantic segmentation technology to study the models, remote sensing image resolutions, and model sample sizes suitable for precise monitoring of rice. The experimental results show that, on the basis of balancing cost, efficiency, and accuracy, the Deeplabv3+ and PSPNet models combined with remote sensing image data of 8 cm resolution are more suitable for monitoring and extraction of rice cultivation areas, and PSPNet has a stronger few-shot learning ability. In response to the strong model generalization ability under the dispersed rice cultivation areas and diversified features, this paper proposes a method of transfer learning with a small number of samples. This method has a more stable training process, and the IoU is 5% $\sim$ 10% higher than that of unsupervised transfer learning models and fully supervised models with a small number of samples.