Two-Stream spectral-spatial convolutional capsule network for Hyperspectral image classification DOI Creative Commons
Han Zhai, Jie Zhao

International Journal of Applied Earth Observation and Geoinformation, Год журнала: 2024, Номер 127, С. 103614 - 103614

Опубликована: Янв. 5, 2024

Recently, the capsule network and its enhanced version named convolutional (Conv-CapsNet) were applied to hyperspectral image (HSI) classification delivered outstanding performance, attributing enriched capabilities of capsules for feature expression complex spatial knowledge excavation. However, restricted by locality convolutions in primary extraction, rich spectral-spatial diagnostic information found HSIs insufficiently utilized, which degrades performance a certain degree. In addition, most networks involve structure with many parameters, are haunted common bottleneck limited available training samples. To tackle these challenges, this paper presents novel two-stream (TSCCN) boost accurate small Based on advanced concept, Conv-CapsNet architecture is developed that reduces trainable parameters mitigates overfitting when samples insufficient. Specifically, 1D 2D sharing similar structures as accompanying streams fully excavate favorable features from continuous spectra raw cubes, separate fused generate joint deep classification. overcome convolutions, an efficient structural mining module (SIM) designed each stream, learn complementary cross-channel cross-kernel inter-dependencies subsequent learning further improve performance. Experiments three well-known HSI datasets demonstrate superiority proposed algorithm over nine representative state-of-the-art classifiers efficacy under sample conditions, where all methods implemented Tensor-Flow framework repeated 10 times.

Язык: Английский

Dynamic simulation of land use change and assessment of carbon storage based on climate change scenarios at the city level: A case study of Bortala, China DOI Creative Commons
Ziyao Wang, Xin Li,

Yueting Mao

и другие.

Ecological Indicators, Год журнала: 2021, Номер 134, С. 108499 - 108499

Опубликована: Дек. 23, 2021

Exploring future changes in land use and carbon storage (CS) under different climate scenarios is important for optimizing regional ecosystem service functions formulating sustainable socioeconomic development policies. We proposed a framework that integrates the system dynamics (SD) model, patch-generating simulation (PLUS) Integrated Valuation of Ecosystem Service Tradeoffs (InVEST) model to dynamically simulate use/cover change (LUCC) CS at city level based on SSP-RCP provided by CMIP6. The simulations were applied Bortala Mongol Autonomous Prefecture Xinjiang. Changes LUCC similar SSP126 SSP245 scenarios, but woodland expansion was more rapid scenario. SSP585 scenario from those other two this mainly caused continuous reduction area construction cultivated land. By 2050, results revealed highest (193.20 Tg), followed (192.75 Tg) (185.17 Tg). Overall, study suggest increases could be achieved controlling economic growth population growth, promoting an energy transition, expanding area.

Язык: Английский

Процитировано

271
Land use/land cover prediction and analysis of the middle reaches of the Yangtze River under different scenarios DOI
Shengqing Zhang, Peng Yang, Jun Xia

и другие.

The Science of The Total Environment, Год журнала: 2022, Номер 833, С. 155238 - 155238

Опубликована: Апрель 13, 2022

Язык: Английский

Процитировано

149
Land use optimization in Ningbo City with a coupled GA and PLUS model DOI
Xiang Li, Jingying Fu,

Dong Jiang

и другие.

Journal of Cleaner Production, Год журнала: 2022, Номер 375, С. 134004 - 134004

Опубликована: Сен. 7, 2022

Язык: Английский

Процитировано

112
The ecosystem service values simulation and driving force analysis based on land use/land cover: A case study in inland rivers in arid areas of the Aksu River Basin, China DOI Creative Commons
Ling Xie, Hongwei Wang, Suhong Liu

и другие.

Ecological Indicators, Год журнала: 2022, Номер 138, С. 108828 - 108828

Опубликована: Апрель 8, 2022

The ecosystem of inland river basin is great significance to the socio-economic stability in arid area. Therefore, evaluate service values (ESVs) necessary for monitor changes. In this paper, response land use/land cover (LULC) during 1990 2020 Aksu River Basin (ARB) ESV was explored. advanced equivalent factor which modified by biomass and used ESVs ARB. A patch-generating use simulation (PLUS) simulate spatial distribution considering influences temperature (TEM), precipitation (PRE), NDVI, DEM, Soil organic matter content (SOMC) Human Activity Intensity Land Surface (HAILS) ARB 2030. results show that total study area showed an increasing trend (1.63 × 1010 yuan 5.64 yuan) from 2020. grassland had highest ESV, accounting nearly 50% detection q value following explanatory power ESV: HAILS (0.332) > NDVI (0.126) TEM (0.125) PRE (0.108) DEM(0.096) SOMC(0.089)and interaction between effect 0.493 on ESV. shape index (SI) negatively correlated with correlation coefficient −0.794. aggregation (AI) Shannon's Diversity Index (SHDI) were positively coefficients 0.872 0.878, respectively. a rapid increase 2030, would still be largest, per unit plowland, forestland, unused water 20131.07 yuan/km2, 64743.29 3054.21 41398.54 This paper can help decision-makers achieve sustainable management develop land-use strategies basins oases.

Язык: Английский

Процитировано

105
Dynamic Simulation of Land Use/Cover Change and Assessment of Forest Ecosystem Carbon Storage under Climate Change Scenarios in Guangdong Province, China DOI Creative Commons
Lei Tian, Tao Yu, Wenxue Fu

и другие.

Remote Sensing, Год журнала: 2022, Номер 14(10), С. 2330 - 2330

Опубликована: Май 11, 2022

Exploring the spatial distribution of land use/cover change (LUCC) and ecosystem carbon storage under future climate scenarios can provide scientific basis for optimizing resource redistribution formulating policies sustainable socioeconomic development. We proposed a framework that integrates patch-generating use simulation (PLUS) model integrated valuation services tradeoffs (InVEST) to assess spatiotemporal dynamic changes in LUCC Guangdong based on shared pathways representative concentration (SSP-RCP) provided by Coupled Model Intercomparison Project 6 (CMIP6). The results showed patterns were similar SSP126 SSP245 scenarios, but artificial surface expanded more rapidly, increase forest slowed down SPP245 scenario. Conversely, SSP585 scenario, sharply resulted continuous decrease land. Under three population, elevation, temperature, distance water highest contributing driving factors growth cultivated land, grassland, surface, respectively. By 2060, terrestrial ecosystems increased from 240.89 Tg 2020 247.16 243.54 respectively, which 17.65 15.34 Tg, respectively; while it decreased 226.54 due destruction accounted 81.05% total storage. Overall, an important recommendation this study is be controlling population economic growth, balancing urban expansion ecological conservation, as well increasing area.

Язык: Английский

Процитировано

104
Comparison of multimodel simulations of land use and land cover change considering integrated constraints - A case study of the Fuxian Lake basin DOI Creative Commons
Zhiqiang Lin, Shuangyun Peng

Ecological Indicators, Год журнала: 2022, Номер 142, С. 109254 - 109254

Опубликована: Авг. 3, 2022

The simulation and future prediction of spatial temporal land use cover change (LUCC) have long been the focus (and challenge) research on sustainable development. However, existing lacks empirical comparisons among different models under same conditions in region; hence, most accurate results cannot be obtained through optimal model. In this paper, taking Fuxian Lake basin as an example, CA-Markov, FLUS, PLUS are adopted to explore model for simulating LUCC by integrating restricted areas LUCC. This study shows following. (1) more joint restrictions open water, prohibited development zones ecological protection red lines than a single restriction. (2) All three can simulate basin. Among these models, has highest accuracy at scale offers best match types terms their quantity distribution. (3) accurately predict changes LULC should help decision makers evaluate patterns provide reference formulation regional policies.

Язык: Английский

Процитировано

81
Simulation of future land use/cover change (LUCC) in typical watersheds of arid regions under multiple scenarios DOI
Qingzheng Wang, Qingyu Guan,

Yunfan Sun

и другие.

Journal of Environmental Management, Год журнала: 2023, Номер 335, С. 117543 - 117543

Опубликована: Фев. 26, 2023

Язык: Английский

Процитировано

71
Construction of ecological network in Suzhou based on the PLUS and MSPA models DOI Creative Commons

Xinlei Xu,

Siyuan Wang,

Wenzhuo Rong

и другие.

Ecological Indicators, Год журнала: 2023, Номер 154, С. 110740 - 110740

Опубликована: Авг. 1, 2023

Building regional ecological network can alleviate the notable contradiction between land use and development in process of rapid urbanization. Suzhou is center city Yangtze River Delta urban cluster a typical water city, but high intensity has fragmented its space. The study firstly introduced patch generation simulation (PLUS) model to simulate under priority scenario 2032, combining data 2002, 2012 2022 provide basis for construction later stage. Secondly, sources four periods were identified by morphological spatial pattern analysis (MSPA) landscape connectivity analysis, corridors nodes each period screened classified minimum cumulative resistance (MCR) model, gravity hydrological analysis. Then we superimposed sources, corridors, construct an that develop harmony with dynamics. Finally, degree optimization was verified structural evaluation comparison 2022. results indicate (1) largest area among types, good substrate. Under guidance green strategy, woodland grassland areas will slightly increase 2032. overall generally remain consistent 2022, fragmentation be mitigated. (2) A total 23 are identified, mainly located near Taihu Lake, Yangcheng Lake River. Among them, most important source. 76 screened, including 31 22 protected potential mostly corridors. 54 selected, divided into 21 general strategic points, 12 11 restorative points 10 break points. (3) plan forms "three cores, pieces, multiple sources" proposes corresponding refined management measures promote sustainable region. In addition, this aims propose framework coupled simulation, which new ideas similar regions affected urbanization around globe.

Язык: Английский

Процитировано

69
Spatial–temporal evolution analysis of multi-scenario land use and carbon storage based on PLUS-InVEST model: A case study in Dalian, China DOI Creative Commons
Qiang Wu, Wang Li, Tianyi Wang

и другие.

Ecological Indicators, Год журнала: 2024, Номер 166, С. 112448 - 112448

Опубликована: Авг. 3, 2024

Язык: Английский

Процитировано

28
Soil conservation ecosystem service supply-demand and multi scenario simulation in the Loess Plateau, China DOI Creative Commons

Zhen Jian,

Yingjun Sun, Fang Wang

и другие.

Global Ecology and Conservation, Год журнала: 2024, Номер 49, С. e02796 - e02796

Опубликована: Янв. 1, 2024

Ecosystem services are an essential foundation for sustainable human development. In view of the severe soil erosion on Loess Plateau (LP), it is important to explore spatial and temporal characteristics supply-demand balance conservation (SCSs). this study, ecosystem service ratio was innovatively proposed quantitatively assess supply demand conservation. First, InVEST model used spatially quantify (soil retention) erosion) (SCS) in LP from 2005 2020. Meanwhile, variation matching were analyzed. Second, predict future, InVEST-PLUS simulate under multiple scenarios 2030. The natural development (NDS), ecological protection (EPS) economic (EDS) coupled with shared socioeconomic pathways (SSPs) representative concentration (RCPs). results showed that: (1) 2020, annual average modulus increased by 60.58 (t/(hm²·a)); retention per unit area 12.89 t/hm2; both northwest southeast, pattern mainly low-low shifted low-high each year; (2) based LULC simulated PLUS model, kappa coefficient overall accuracy 86.29% 91.14%, respectively. A total 19.8% 22.97% built-up land expanded NDS EDS scenarios, respectively, a 17,515.65 km2 forestland grassland EPS scenario; (3) 2030.NDS-SSP245, 2030.EPS-SSP126 2030.EDS-SSP585 moderate high surpluses accounted 11.40%, 14.34% 10.80% surplus area, respectively; compared deficit 49.67%, 43.29% 56.65% study provided innovations coupling degree future model.

Язык: Английский

Процитировано

24