Ideas and perspectives: Research on ecosystem-atmosphere interactions in Asia: early career researcher opinion DOI Creative Commons
Sung‐Ching Lee, Hojin Lee, Tin W. Satriawan

et al.

Published: Aug. 22, 2024

Abstract. Due to a growing recognition of the need study how ecosystems and atmosphere interact with each other, many regional networks as well global network networks, FLUXNET, were formed. Since 1999, when AsiaFlux was established, scientists in region have been measuring flux densities energy, water vapor, greenhouse gas exchanges better evaluate ecosystem-atmosphere interactions understand their underlying mechanisms. The includes natural managed that span broad climatic ecological gradients, experience diverse management practices disturbances. In this ideas perspectives paper, from view early career researchers (ECRs), we synthesize key research foci recent years, focus on latest conferences, highlight selected discoveries. While achieving significant milestones, ECRs argue community should work together emphasize importance long-term observations, rejuvenate network’s shared open-access database, actively engage stakeholders. With unique ecosystem types Asian region, efforts expertise can provide critical insights into roles climate change, extreme weather events, soil properties, vegetation physiology structure, breathing biosphere. closing, hope paper inspire future generation Asia promote between across different cultures stages.

Language: Английский

Toward a better understanding of curve number and initial abstraction ratio values from a large sample of watersheds perspective DOI
Abderraman R. Amorim Brandão, Dimaghi Schwamback, André S. Ballarin

et al.

Journal of Hydrology, Journal Year: 2025, Volume and Issue: unknown, P. 132941 - 132941

Published: Feb. 1, 2025

Language: Английский

Citations

0
Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model DOI Creative Commons

Rui Wang,

Xingping Wang,

Zhonghu Zhang

et al.

Systems, Journal Year: 2025, Volume and Issue: 13(3), P. 187 - 187

Published: March 7, 2025

Within globalization, the significance of urban innovation cooperation has become increasingly evident. However, faces challenges due to various factors—social, economic, and spatial—making it difficult for traditional methods uncover intricate nonlinear relationships among them. Consequently, this research concentrates on cities within Yangtze River Delta region, employing an explainable machine learning model that integrates eXtreme Gradient Boosting (XGBoost), SHapley Additive exPlanations (SHAP), Partial Dependence Plots (PDPs) investigate interactive effects multidimensional factors impacting cooperation. The findings indicate XGBoost outperforms LR, SVR, RF, GBDT in terms accuracy effectiveness. Key results are summarized as follows: (1) Urban exhibits different phased characteristics. (2) There exist between factors, them, Scientific Technological dimension contributes most (30.59%) significant positive promoting effect later stage after surpassing a certain threshold. In Social Economic (23.61%), number Internet Users (IU) individually. Physical Space (20.46%) generally mutation points during early stages development, with overall predominantly characterized by trends. (3) Through application PDP, is further determined IU synergistic per capita Foreign Direct Investment (FDI), public library collections (LC), city night light data (NPP), while exhibiting negative antagonistic Average Annual Wage Staff (AAS) Enterprises above Designated Size Industry (EDS). (4) For at developmental stages, tailored development proposals should be formulated based single-factor contribution multifactor interaction effects. These insights enhance our understanding elucidate influencing factors.

Language: Английский

Citations

0
Flash Flood Regionalization for the Hengduan Mountains Region, China, Combining GNN and SHAP Methods DOI Creative Commons

Yifan Li,

Chendi Zhang, Peng Cui

et al.

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(6), P. 946 - 946

Published: March 7, 2025

The Hengduan Mountains region (HMR) is vulnerable to flash flood disasters, which account for the largest proportion of flood-related fatalities in China. Flash regionalization, divides a into homogeneous subdivisions based on flood-inducing factors, provides insights spatial distribution patterns risk, especially ungauged areas. However, existing methods regionalization have not fully reflected topology structure inputted geographical data. To address this issue, study proposed novel framework combining state-of-the-art unsupervised Graph Neural Network (GNN) method, Dink-Net, and Shapley Additive exPlanations (SHAP) HMR. A comprehensive dataset inducing factors was first established, covering geomorphology, climate, meteorology, hydrology, surface conditions. performances two classic machine learning (K-means Self-organizing feature map) three GNN (Deep Infomax (DGI), Deep Modularity Networks (DMoN), Dilation shrink (Dink-Net)) were compared flash-flood Dink-Net model outperformed others. SHAP then applied quantify impact all results by Dink-Net. newly developed captured interactions characterized factors. allowed be independent from historical data, would facilitate its application mountainous analysis highlights significant positive influence extreme rainfall floods across entire pronounced soil moisture saturated hydraulic conductivity areas with concentration events, together topography (elevation) transition zone Qinghai–Tibet Plateau Sichuan Basin, also been revealed. provide technical support scientific basis control disaster reduction measures mountain according local

Language: Английский

Citations

0
Interpretable machine learning reveals the importance of geography and landscape arrangement for surface water quality across China DOI
Kun Huo, Wangzheng Shen,

Junchong Wei

et al.

Water Research, Journal Year: 2025, Volume and Issue: unknown, P. 123578 - 123578

Published: March 1, 2025

Language: Английский

Citations

0
Transdisciplinary coordination is essential for advancing agricultural modeling with machine learning DOI Creative Commons
Lily‐belle Sweet, Ioannis N. Athanasiadis,

Ron van Bree

et al.

One Earth, Journal Year: 2025, Volume and Issue: 8(4), P. 101233 - 101233

Published: April 1, 2025

Language: Английский

Citations

0
Exploring the climate signal in the variation of winter wheat quality records in the North China Plain DOI

Weimo Zhou,

Syed Tahir Ata-Ul-Karim, Yoichiro Kato

et al.

Agricultural and Forest Meteorology, Journal Year: 2025, Volume and Issue: 369, P. 110567 - 110567

Published: April 22, 2025

Language: Английский

Citations

0
Spatial Heterogeneity of Driving Factors in Multi-Vegetation Indices RSEI Based on the XGBoost-SHAP Model: A Case Study of the Jinsha River Basin, Yunnan DOI Creative Commons
Jisheng Xia, Guoyou Zhang,

Shiping Ma

et al.

Land, Journal Year: 2025, Volume and Issue: 14(5), P. 925 - 925

Published: April 24, 2025

The Jinsha River Basin in Yunnan serves as a crucial ecological barrier southwestern China. Objective assessment and identification of key driving factors are essential for the region’s sustainable development. Remote Sensing Ecological Index (RSEI) has been widely applied assessments. In recent years, interpretable machine learning (IML) introduced novel approaches understanding complex mechanisms. This study employed Google Earth Engine (GEE) to calculate three vegetation indices—NDVI, SAVI, kNDVI—for area from 2000 2022, along with their corresponding RSEI models (NDVI-RSEI, SAVI-RSEI, kNDVI-RSEI). Additionally, it analyzed spatiotemporal variations these relationship indices. Furthermore, an IML model (XGBoost-SHAP) was interpret RSEI. results indicate that (1) levels 2022 were primarily moderate; (2) compared NDVI-RSEI, SAVI-RSEI is more susceptible soil factors, while kNDVI-RSEI exhibits lower saturation tendency; (3) potential evapotranspiration, land cover, elevation drivers variations, affecting environment western, southeastern, northeastern parts area. XGBoost-SHAP approach provides valuable insights promoting regional

Language: Английский

Citations

0
Long-term spatiotemporal mapping in lacustrine environment by remote sensing:Review with case study, challenges, and future directions DOI

Lai Lai,

Yuchen Liu, Yuchao Zhang

et al.

Water Research, Journal Year: 2024, Volume and Issue: 267, P. 122457 - 122457

Published: Sept. 16, 2024

Language: Английский

Citations

3
High-resolution soil temperature and soil moisture patterns in space, depth and time: An interpretable machine learning modelling approach DOI Creative Commons
Maiken Baumberger, Bettina Haas, S. Sivakumar

et al.

Geoderma, Journal Year: 2024, Volume and Issue: 451, P. 117049 - 117049

Published: Oct. 17, 2024

Language: Английский

Citations

2
Terrain Analysis According to Multiscale Surface Roughness in the Taklimakan Desert DOI Creative Commons
Sebastiano Trevisani, Peter L. Guth

Land, Journal Year: 2024, Volume and Issue: 13(11), P. 1843 - 1843

Published: Nov. 5, 2024

Surface roughness, interpreted in the wide sense of surface texture, is a generic term referring to variety aspects and scales spatial variability surfaces. The analysis solid earth roughness useful for understanding, characterizing, monitoring geomorphic factors at multiple spatiotemporal scales. different features characterizing landscape exhibit specific characteristics texture. capability selectively analyze metrics represents key tool geomorphometric analysis. This research presents simplified geostatistical approach multiscale or image texture case images, that highly informative interpretable. implemented able describe two main short-range roughness: omnidirectional anisotropy. Adopting simple upscaling approaches, it possible perform roughness. An overview information extraction potential shown portion Taklimakan desert (China) using 30 m resolution DEM derived from Copernicus Glo-30 DSM. indexes are used as input unsupervised supervised learning tasks. can be refined both perspective well relation considered. However, even its present, form, find direct applications contexts topics.

Language: Английский

Citations

1