Unraveling nonlinear and spatial non-stationary effects of urban form on surface urban heat islands using explainable spatial machine learning

Computers Environment and Urban Systems, Год журнала: 2024, Номер 114, С. 102200 - 102200

Опубликована: Окт. 4, 2024

Язык: Английский

---

Assessing the impact of adjacent urban morphology on street temperature: a multisource analysis using random forest and SHAP

Building and Environment, Год журнала: 2024, Номер unknown, С. 112326 - 112326

Опубликована: Ноя. 1, 2024

Язык: Английский

---

CLSER: A new indicator for the social-ecological resilience of coastal systems and sustainable management

Journal of Cleaner Production, Год журнала: 2024, Номер 435, С. 140564 - 140564

Опубликована: Янв. 1, 2024

Язык: Английский

---

A deep learning-based super-resolution method for building height estimation at 2.5 m spatial resolution in the Northern Hemisphere

Remote Sensing of Environment, Год журнала: 2024, Номер 310, С. 114241 - 114241

Опубликована: Июнь 4, 2024

Язык: Английский

---

3D-GloBFP: the first global three-dimensional building footprint dataset

Earth system science data, Год журнала: 2024, Номер 16(11), С. 5357 - 5374

Опубликована: Ноя. 25, 2024

Abstract. Understanding urban vertical structures, particularly building heights, is essential for examining the intricate interaction between humans and their environment. Such datasets are indispensable a variety of applications, including climate modeling, energy consumption analysis, socioeconomic activities. Despite importance this information, previous studies have primarily focused on estimating heights regionally at grid scale, often resulting in with limited coverage or spatial resolution. This limitation hampers comprehensive global analysis ability to generate actionable insights finer scales. In study, we developed height map footprint scale by leveraging Earth Observation (EO) advanced machine learning techniques. Our approach integrated multisource remote-sensing features morphology develop estimation models using extreme gradient boosting (XGBoost) regression method across diverse regions. methodology allowed us estimate individual buildings worldwide, culminating creation three-dimensional (3D) Global Building Footprints (3D-GloBFP) dataset year 2020. evaluation results show that perform exceptionally well R2 values ranging from 0.66 0.96 root-mean-square errors (RMSEs) 1.9 14.6 m 33 subregions. Comparisons other demonstrate 3D-GloBFP closely matches distribution pattern reference heights. derived 3D shows distinct regions, countries, cities, gradually decreasing city center surrounding rural areas. Furthermore, our findings indicate disparities built-up infrastructure (i.e., volume) different countries cities. China country most intensive total (5.28×1011 m3, accounting 23.9 % total), followed USA (3.90×1011 17.6 total). Shanghai has largest volume (2.1×1010 m3) all representative The building-footprint-scale reveals significant heterogeneity environments, providing valuable dynamics climatology. available https://doi.org/10.5281/zenodo.11319912 (Building Americas, Africa, Oceania 3D-GloBFP; Che et al., 2024c), https://doi.org/10.5281/zenodo.11397014 Asia 2024a), https://doi.org/10.5281/zenodo.11391076 Europe 2024b).

Язык: Английский

---

A 100-m gridded population dataset of China’s seventh census using ensemble learning and geospatial big data

Опубликована: Фев. 6, 2024

Abstract. China has undergone rapid urbanization and internal migration in past years its up-to-date gridded population datasets are essential for diverse applications. Existing China, however, suffer from either outdatedness or failure to incorporate the latest seventh national census data conducted 2020. In this study, we develop a novel downscaling approach that leverages stacking ensemble learning geospatial big produce grids at 100-m resolution both county town levels. The proposed employs random forest, XGBoost, LightGBM as base models delineates inhabited areas enhance estimation. Experimental results demonstrate exhibits best fit performance compared individual models. Meanwhile, out-of-sample town-level test set indicates estimated dataset (R2=0.8936) is more accurate than existing WorldPop (R2=0.7427) LandScan (R2=0.7165) products Furthermore, with enhancement, spatial distribution of reasonable intuitively two products. Hence, provides valuable option producing datasets. holds great significance future applications it publicly available https://figshare.com/s/d9dd5f9bb1a7f4fd3734 (Chen et al., 2024).

Язык: Английский

---

GABLE: A first fine-grained 3D building model of China on a national scale from very high resolution satellite imagery

Remote Sensing of Environment, Год журнала: 2024, Номер 305, С. 114057 - 114057

Опубликована: Фев. 27, 2024

Three-dimensional (3D) building models provide horizontal and vertical information of urban development patterns, which are significant to urbanization analysis, solar energy planning, carbon reduction sustainability. Despite that many popular products on a global or national scale proposed, these usually focus extraction height estimation at fairly coarse resolutions while categories not taken into consideration. In this study, we extend the previous work in two aspects involving introduction semantically fine-grained (i.e., 12 rooftop classes) spatially representations individual buildings with compact polygons. Specifically, develop novel framework for generation 3D models, including developing network joint classification, another parallel estimation, post-processing algorithm fusion results from independent networks. To train networks improve generalization, construct custom large-scale datasets addition existing Urban Building Classification (UBC) dataset 2023 IEEE Data Fusion Contest (DFC 2023) dataset. Finally, nation-scale fine-GrAined BuiLding modEl (GABLE) product is derived based Beijing-3 satellite images (0.5–0.8 m) our proposed framework. GABLE provides polygon, category value each instance. Further analyses conducted uncover distribution terms diversity, density. These demonstrate significance values GALBE, potentials far beyond these.

Язык: Английский

---

Mapping material stocks in buildings and infrastructures across the Beijing–Tianjin–Hebei urban agglomeration at high-resolution using multi-source geographical data

Resources Conservation and Recycling, Год журнала: 2024, Номер 205, С. 107561 - 107561

Опубликована: Март 22, 2024

Язык: Английский

---

Advancing the local climate zones framework: a critical review of methodological progress, persisting challenges, and future research prospects

Humanities and Social Sciences Communications, Год журнала: 2024, Номер 11(1)

Опубликована: Апрель 27, 2024

Abstract The local climate zones (LCZs) classification system has emerged as a more refined method for assessing the urban heat island (UHI) effect. However, few researchers have conducted systematic critical reviews and summaries of research on LCZs, particularly regarding significant advancements this field in recent years. This paper aims to bridge gap scientific by systematically reviewing evolution, current status, future trends LCZs framework research. Additionally, it critically assesses impact climate-responsive planning design. findings study highlight several key points. First, challenge large-scale, efficient, accurate mapping persists issue Despite challenge, universality, simplicity, objectivity make promising tool wide range applications future, especially realm In conclusion, makes substantial contribution advancement advocates broader adoption foster sustainable development. Furthermore, offers valuable insights practitioners engaged field.

Язык: Английский

---

3D-GloBFP: the first global three-dimensional building footprint dataset

Опубликована: Июнь 24, 2024

Abstract. Understanding urban vertical structures, particularly building heights, is essential for examining the intricate interaction between humans and their environment. Such datasets are indispensable a variety of applications, including climate modeling, energy consumption analysis, socioeconomic activities. Despite importance this information, previous studies have primarily focused on estimating heights regionally grid scale, often resulting in with limited coverage or spatial resolution. This limitation hampers comprehensive global analyses ability to generate actionable insights finer scales. In study, we developed height map (3D-GloBFP) at footprint scale by leveraging Earth Observation (EO) advanced machine learning techniques. Our approach integrated multisource remote sensing features morphology develop estimation models using eXtreme Gradient Boosting (XGBoost) regression method across diverse regions. methodology allowed us estimate individual buildings worldwide, culminating creation first three-dimensional (3-D) footprints (3D-GloBFP). evaluation results show that perform exceptionally well worldwide R2 ranging from 0.66 0.96 root mean square errors (RMSEs) 1.9 m 14.6 33 subregions. Comparisons other demonstrate our 3D-GloBFP closely matches distribution pattern reference heights. derived 3-D shows distinct regions, countries, cities, gradually decreasing city center surrounding rural areas. Furthermore, findings indicate disparities built-up infrastructure (i.e., volume) different countries cities. China country most intensive total (5.28×1011 m3, accounting 23.9 % total), followed United States (3.90×1011 17.6 total). Shanghai has largest volume (2.1×1010 m3) all representative The building-footprint reveals significant heterogeneity environments, providing valuable dynamics climatology. dataset available https://doi.org/10.5281/zenodo.11319913 (Building Americas, Africa, Oceania 3D-GloBFP) (Che et al., 2024a), https://doi.org/10.5281/zenodo.11397015 Asia 2024b), https://doi.org/10.5281/zenodo.11391077 Europe 2024c).

Язык: Английский

---