Chunqing Liu

Historic gardens represent evolving cultural landscapes shaped by human–environment interactions. However, coherent methods for evaluating their spatial and temporal continuity remain limited. This study integrates the Historic Land-Use Assessment (HLA) framework with deep mapping to construct an indicator system for assessing garden heritage value. By combining qualitative interpretation with quantitative spatial modeling, and using Nanchang, China, as a case study, we combined historical documents, cartographic archives, and ArcGIS-based spatial analysis to reconstruct garden evolution and produce a spatio-temporal continuity map. The results show that changes in garden function strongly influence temporal depth, while natural substrates promote long-term preservation through ecological–cultural anchoring. Together, these factors reveal the mechanisms of landscape persistence and transformation across historical periods. These findings highlight the multi-period layering of cultural and ecological processes that define historic gardens. The study provides a replicable framework and practical guidance for the conservation and planning of landscape gardens.

In recent years, the study of semantic segmentation of remote sensing images (RSI) has gained significant attention due to its critical role in geospatial analysis, agriculture, and forestry. However, existing remote sensing segmentation methods face several challenges: (1) limited dataset diversity and inadequate exploration of traditional village landscapes, resulting in a lack of geospatial representation for these unique environments; (2) inefficiencies in same-layer or cross-layer feature fusion when using convolutional neural networks (CNNs) or transformers, leading to either insufficient spatial modeling or excessive computational demands; and (3) multimodal approaches that improve modeling accuracy but introduce high parameter complexity and computational overhead. To address these issues, we propose the Mamba Prompt Learning Network (MPLNet) for efficient and accurate RSI segmentation, with a strong emphasis on spatial information extraction and GIS-based applications. First, we construct TV-RSI, a highly diverse large-scale data set specifically designed to capture the spatial structures, topographic variations, and land use patterns of traditional villages. Second, we develop the Mamba Fusion Module, which improves geospatial feature utilization by efficiently modeling both intralayer and interlayer spatial relationships, ensuring comprehensive feature extraction. Finally, we introduce prompt learning, which transfers bimodal geospatial knowledge from heavy-weight networks into a lightweight unimodal model, improving segmentation accuracy while maintaining computational efficiency. Extensive experiments on TV-RSI and two publicly available RSI datasets demonstrate that MPLNet achieves state-of-the-art performance with significantly reduced computational costs, making it an ideal solution for geospatial segmentation tasks in GIS-driven remote sensing applications.

Urban greenway trails serve as a vital link between urban populations and the natural environment, playing a key role in enhancing quality of life and promoting physical and mental well-being. We propose an interpretable machine learning framework applied to 424 geotagged footprint images from the Bada Shanren Meihu Scenic Area in China. Our main findings are as follows: (1) The key factors influencing trail landscape preferences include the Water Visibility Index (WVI), Building Landscape Index (BVI), Freedom Index, and Greenery Visibility Index (GVI). (2) For WVI, SHAP values significantly increase around the 0.05 threshold. BVI has a critical threshold of 0.17, with a strong influence below it and a reduced effect above it. The Freedom variable shows an inverse relationship, with minimal contribution below 0.21 and a sharp increase above this threshold. GVI maintains high SHAP values at lower levels (GVI ≤ 0.66), but its predictive utility decreases at higher values. (3) Landscape preferences are significantly positively correlated with naturalness, wildness, WVI, and openness, with water landscapes being the strongest driver. In contrast, artificial factors, V_Low, and H_Purple significantly suppress preferences. This suggests that human intervention and certain color tones may reduce the attractiveness of the landscape.

Traditional villages have evolved over time to adapt to their environmental characteristics, demonstrating high ecological suitability. Ideal village locations not only provide comfortable living spaces but also ensure safety and sustainability, reflecting the ancestors’ profound understanding of the natural environment and ecological wisdom. This study employs an artificial intelligence-based multi-model integration approach to evaluate the ecological suitability of 413 traditional village sites in Jiangxi Province. Key influencing factors are identified, and the ecological wisdom of ancestral site selection is analyzed, resulting in an ecological suitability evaluation map for traditional village locations in Jiangxi Province. The study draws upon environmental characteristic data of Jiangxi Province, including topography, climate, habitat quality, land use, air quality, vegetation cover, and river network density. GIS technology is utilized for spatial analysis and result visualization, with raster data being extracted and standardized. Machine learning methods, such as Random Forest, Support Vector Machine, and Gradient Boosting Decision Trees, along with deep learning methods like Convolutional Neural Networks and Multilayer Perceptrons, are applied. Multi-model integration techniques combine diverse predictive outputs, thereby enhancing the overall accuracy and robustness of ecological suitability evaluations. Experimental results indicate that elevation, slope, habitat quality, actual distance to water bodies, and average temperature are the main influencing factors for village site selection. The multi-model integration method performs excellently in evaluating ecological suitability, effectively identifying key ecological factors. The model’s accuracy and reliability are verified through confusion matrix, feature importance analysis, and ROC curve. By analyzing the impact weights of various ecological factors, this study constructs a Composite Suitability Index (CSI) and generates an ecological suitability evaluation map that clearly displays suitability levels. This provides a scientific basis for the protection and rational development of traditional villages and serves as a reference for ecological site selection studies in other regions.

With the rapid development of artificial intelligence technology, an increasing number of village-related modeling problems have been addressed. However, first, the exploration of village-related watershed fine-grained classification problems, particularly the multi-view watershed fine-grained classification problem, has been hindered by dataset collection limitations; Second, village-related modeling networks typically employ convolutional modules for attentional modeling to extract salient features, yet they lack global attentional feature modeling capabilities; Lastly, the extensive number of parameters and significant computational demands render village-related watershed fine-grained classification networks infeasible for end-device deployment. To tackle these challenges, we introduce a multi-view attention mechanism designed for precise watershed classification, leveraging knowledge distillation techniques, abbreviated as MANet-KD. Specifically, first, we have developed the inaugural multi-view watershed classification dataset, termed MVWD.Second, we introduce a cross-view attention module (CVAM), which models salient features from intersecting views with global attention, enhancing the accuracy and precision of watershed classification. This module enhances fine-grained classification accuracy. Based on the above proposed CVAM, we propose a heavyweight MANet-Teacher and a lightweight MANet-Student, and finally, we introduce an Attention Knowledge Distillation (AKD) strategy that effectively transfers critical feature knowledge from the teacher network to the student network, utilizing the AKD approach for enhanced learning outcomes. The experimental results show that the proposed MANet-Teacher achieves state-of-the-art performance with 78.51% accuracy, and the proposed MANet-Student achieves comparable performance to MANet-Teacher with 6.64M parameters and 1.68G computation. The proposed MANet-KD achieves a good balance of performance and efficiency in the multi-view fine-grained watershed classification task. To facilitate further research in multi-view fine-grained watershed classification, all datasets, codes, and benchmark outcomes will be made available to the public. https://github.com/Jack13026212687/MANet-KD.

Traditional villages have evolved over time to adapt to their environmental characteristics, demonstrating high ecological suitability. Ideal village locations not only provide comfortable living spaces but also ensure safety and sustainability, reflecting the ancestors’ profound understanding of the natural environment and ecological wisdom. This study employs a multi-model integration approach using artificial intelligence to evaluate the ecological suitability of 413 traditional village sites in Jiangxi Province. It identifies key influencing factors and analyzes the ecological wisdom of ancestral site selection, resulting in an ecological suitability evaluation map for traditional village locations in Jiangxi Province. The study is based on environmental characteristic data of Jiangxi Province, including topography, climate, habitat quality, land use, air quality, vegetation cover, and river network density. GIS is used for spatial analysis and result visualization, and raster data are extracted and standardized. Machine learning methods such as Random Forest, Support Vector Machine, and Gradient Boosting Decision Trees, as well as deep learning methods such as Convolutional Neural Networks and Multilayer Perceptrons, are applied. Multi-model integration techniques are used to combine the predictions of various models, enhancing the accuracy and robustness of the ecological suitability evaluation. Experimental results indicate that elevation, slope, habitat quality, actual distance to water bodies, and average temperature are the main influencing factors for village site selection. The multi-model integration method performs excellently in evaluating ecological suitability, effectively identifying key ecological factors. The model’s accuracy and reliability are verified through confusion matrix, feature importance analysis, and ROC curve. This study evaluates the ecological suitability of traditional village sites in Jiangxi Province using a multi-model integration approach. By analyzing the impact weights of various ecological factors, it constructs a Composite Suitability Index (CSI) and generates an ecological suitability evaluation map, displaying suitability levels. This provides a scientific basis for the protection and rational development of traditional villages and serves as a reference for ecological site selection studies in other regions.

The assessment of landscape visual quality (LVQ) holds significant importance in the preservation and advancement of traditional villages. One challenge in measuring human perception lies in establishing a connection between public preferences and landscape characteristics. This study conducted an analysis of social media data from Anyi traditional villages in China to address this issue and identified eight human perceptions: naturalness, ancientness, colorfulness, variety, uniqueness, ingenuity, vividness, and pleasantness. A total of thirty characteristic indicators with potential explanations for LVQ were determined by research group through field investigations. A questionnaire survey was developed to assess human’s preferences using 82 traditional village photos, and scores for the eight perceptions were obtained. The logistic regression was employed to establish distinct perception models, with perceptions serving as the dependent variables and characteristic indicators as the independent variables. Nomograms were subsequently utilized to visualize regression results and display the correlation between these two factors. The findings suggest that nomograms facilitate intuitive determination of the weights assigned to characteristic indicators in perceptual models, as well as their influence on LVQ. This work provides a reference for decision-making related to the adaptive protection and development of traditional villages, thereby helping to enhance the competitiveness of tourist destinations.

Introduction: As the global urbanization process accelerates, the contradiction between economic development demands and ecological protection becomes increasingly prominent.Methods: In this study, we simulated the evolution of the ecological security pattern (ESP) of Pingtan Island from 2000 to 2020 by extracting the ecological sources using Remote Sensing Ecological Index (RSEI), and identifying the ecological corridors and key nodes by combining with Linkage Mapping (LM) and Circuit Theory. In addition, Geodetector was utilized to identify these major determinants affecting RSEI.Results: The results showed 1) From 2000 to 2020, the ecological environmental quality (EEQ) of Pingtan Island continued to improve, and the mean value of RSEI gradually increased from 0.47 to 0.51. 2) Univariate analysis showed that elevation and slope were the most significant factors affecting the spatial variability of the RSEI, with the interaction between slope and proportion of built-up area having a significant effect on EEQ. 3) The number and extent of ecological sources were expanded year by year with significant spatial variability. At the same time, the number and range of ecological corridors also underwent phase adjustment. 4) Further exploration of ESP of Pingtan Island in 2020 identified 32 ecological pinch points (EPPs) and 52 ecological barrier points (EBPs), which were mainly located within or near the ecological corridors, indicating key areas for future ecological restoration efforts.Discussion: These insights help to enhance urban spatial planning and ecosystem restoration on Pingtan Island and provide a blueprint for ESP development in comparable island urban environments.

The composition and configuration of landscapes are critical important to design effective approaches to mitigate urban thermal environment in the urbanization process. In this research, land use maps and land surface temperature (LST) retrieval were derived in Nanchang city of central China based on product datasets and the thermal infrared band of Landsat. The results showed that the thermal environment of Nanchang had become worse over the past two decades, that is, the proportion of area of the extremely low temperature zone (ELTZ) decreased from 4.39 to 0.77% from 2001 to 2020, and that of medium temperature zone (MTZ) reduced by 20%, whereas those of the high temperature zone (HTZ) and the extremely high temperature zone (EHTZ) increased sharply after 2001, and by 2020, the area ratio increased by 11% and 7.16%, respectively. The agricultural land (AL) area decreased from 68.44 to 49.69%, was gradually replaced by construction land (CL). The CL occupied the largest proportion in EHTZ, HTZ and slight high temperature zone (SHTZ); water landscape (WL) and green land (GL) occupied the largest proportion in ELTZ, low temperature zone (LTZ); and AL occupied the largest proportion in SHTZ, MTZ, and slight low temperature zone (SLTZ). Landscape configuration also obviously impacted on LST. The model fitting was well (R = 0.87) between land use area and LST by multiple regression analysis. The significant correlation between LST and six landscape pattern indices of CL (p < 0.01) indicated that the larger percent (PLANT, R = 0.78) and the more concentrate (LPI, R = 0.73) of CL implied the higher LST, while the more fragment (NP, R = − 0.45), dispersed and complex shape (R = − 0.35) were benefit to relieve LST. Contrastively, the larger percent and the more concentrated and complex shape distribution of AL, GL and WL, the lower LST (p < 0.01). In addition, LST had closely correlation with landscape level indices such as aggregation degree (AI, R = 0.44) and diversity (SHDI, R = − 0.60) (p < 0.01).