A Multi-Layer Perceptron Approach to Downscaling Geostationary Land Surface Temperature in Urban Areas DOI Creative Commons
Alexandra Hurduc, Sofia L. Ermida, Carlos C. DaCamara

и другие.

Remote Sensing, Год журнала: 2024, Номер 17(1), С. 45 - 45

Опубликована: Дек. 27, 2024

Remote sensing of land surface temperature (LST) is a fundamental variable in analyzing variability urban areas. Geostationary sensors provide sufficient observations throughout the day for diurnal analysis temperature, however, lack spatial resolution needed highly heterogeneous areas such as cities. Polar orbiting have advantage higher resolution, enabling better characterization while only providing one to two per day. This work aims at using multi-layer perceptron-based method downscale geostationary-derived LST based on polar-orbit-derived one. The model trained pixel-by-pixel basis, which reduces complexity requiring fewer auxiliary data characterize conditions. Results show that able successfully city Madrid, from approximately 4.5 km 750 m. Performance metrics between training and validation datasets no overfitting. was applied different time period compared derived three additional sensors, were not used any stage process, yielding R2 0.99, root mean square errors 1.45 1.58 absolute ranging 1.07 1.15. downscaled shown improve representation both temporal heterogeneity when geostationary- individually. resulting take high observation frequency geostationary data, combined with polar may be added value study seasonal patterns environments.

Язык: Английский

Quantitative Assessment of Non-Stationary Relationship Between Multi-Scale Urban Morphology and Urban Heat DOI Creative Commons
Deniz Erdem Okumuş, Mert Akay

Building and Environment, Год журнала: 2025, Номер unknown, С. 112669 - 112669

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

0
Machine learning approaches for resilient modulus modeling of cement-stabilized magnetite and hematite iron ore tailings DOI Creative Commons
Farzad Safi Jahanshahi, Ali Reza Ghanizadeh

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Фев. 10, 2025

Resilient modulus (Mr) is key for understanding the stiffness and stress‒strain properties of road materials flexible pavement design. Measuring Mr in a laboratory requires conducting dynamic triaxial loading tests with varying confining deviatoric stresses, which can be costly time-consuming process. This study evaluates various machine learning (ML) models to predict cement-stabilized magnetite hematite iron ore tailings based on multiple variables such as cement content, curing time, bulk stress, are considered input parameters. For developing ML models, set data from experimental studies was collected. After comparison, Gaussian Process Regression outperformed other methods predicting both MIOT HIOT. HIOT materials, R2 values were 0.9936 0.9876, 0.9893 0.9825 train test datasets, respectively. The sensitivity analysis revealed that time least important variable, whereas Portland percentage most effective tailings. Additionally, parametric undertaken investigate impact each variable Mr.

Язык: Английский

Процитировано

0
Decoding urban emissions: the overlooked impact of commercial and public service zones across regions and seasons DOI Creative Commons

Zhiyu Yi,

Yuebin Wang, Xiaojing Yao

и другие.

GIScience & Remote Sensing, Год журнала: 2025, Номер 62(1)

Опубликована: Апрель 18, 2025

Язык: Английский

Процитировано

0
The Relationship between Maternal Environmental Temperature Exposure and Preterm Birth: A Risk Prediction Based on Machine Learning DOI
Yuxiao Wang,

Xing Bi,

Yang Cheng

и другие.

Sustainable Cities and Society, Год журнала: 2024, Номер 115, С. 105814 - 105814

Опубликована: Сен. 11, 2024

Язык: Английский

Процитировано

3
Skip or not: Hybrid machine learning for decision support in strategic port-skipping behavior to enhance liner shipping reliability DOI

Xingcan Fan,

Jing Lyu, Lingye Zhang

и другие.

Ocean Engineering, Год журнала: 2025, Номер 324, С. 120730 - 120730

Опубликована: Фев. 22, 2025

Язык: Английский

Процитировано

0
A Comprehensive Study on the Estimation of Concrete Compressive Strength Using Machine Learning Models DOI Creative Commons
Yusuf Tahir ALTUNCI

Buildings, Год журнала: 2024, Номер 14(12), С. 3851 - 3851

Опубликована: Ноя. 30, 2024

Conducting comprehensive analyses to predict concrete compressive strength is crucial for enhancing safety in field applications and optimizing work processes. There an extensive body of research the literature focusing on predicting mechanical properties concrete, such as strength. Summarizing key contributions these studies will serve a guide future research. To this end, study aims conduct scientometric analysis that utilize machine learning (ML) models strength, assess models, provide insights developing optimal solutions. Additionally, it seeks offer researchers information prominent themes, trends, gaps regarding prediction. For purpose, 2319 articles addressing prediction published between 2000 19 August 2024, were identified through Scopus Database. Scientometric conducted using VOSviewer software. The evaluation relevant demonstrates ML are frequently used advantages limitations examined, with particular emphasis considerations when working complex datasets. A their practical distinguishes from existing This contributes significantly by examining leading institutions, countries, authors, sources field, synthesizing data, identifying areas, gaps, trends It establishes strong foundation design ML-supported, reliable, sustainable, optimized structural systems civil engineering, building materials, industry.

Язык: Английский

Процитировано

2
Combined Drought Index Using High-Resolution Hydrological Models and Explainable Artificial Intelligence Techniques in Türkiye DOI Creative Commons
Eyyup Ensar Başakın, Paul C. Stoy, Mehmet Cüneyd Demirel

и другие.

Remote Sensing, Год журнала: 2024, Номер 16(20), С. 3799 - 3799

Опубликована: Окт. 12, 2024

We developed a combined drought index to better monitor agricultural events. To develop the index, different combinations of temperature condition precipitation vegetation soil moisture gross primary productivity, and normalized difference water were used obtain single severity index. more effective results, mesoscale hydrologic model was values. The SHapley Additive exPlanations (SHAP) algorithm calculate weights for provide input SHAP model, crop yield predicted using machine learning with training set yielding correlation coefficient (R) 0.8, while test values calculated be 0.68. representativeness new in situations compared established indices, including Standardized Precipitation-Evapotranspiration Index (SPEI) Self-Calibrated Palmer Drought Severity (scPDSI). showed highest an R-value 0.82, followed by SPEI 0.7 scPDSI 0.48. This study contributes perspective detection integration increased volume data from remote sensing systems technological advances could facilitate development significantly efficient monitoring systems.

Язык: Английский

Процитировано

1
A Multi-Layer Perceptron Approach to Downscaling Geostationary Land Surface Temperature in Urban Areas DOI Creative Commons
Alexandra Hurduc, Sofia L. Ermida, Carlos C. DaCamara

и другие.

Remote Sensing, Год журнала: 2024, Номер 17(1), С. 45 - 45

Опубликована: Дек. 27, 2024

Remote sensing of land surface temperature (LST) is a fundamental variable in analyzing variability urban areas. Geostationary sensors provide sufficient observations throughout the day for diurnal analysis temperature, however, lack spatial resolution needed highly heterogeneous areas such as cities. Polar orbiting have advantage higher resolution, enabling better characterization while only providing one to two per day. This work aims at using multi-layer perceptron-based method downscale geostationary-derived LST based on polar-orbit-derived one. The model trained pixel-by-pixel basis, which reduces complexity requiring fewer auxiliary data characterize conditions. Results show that able successfully city Madrid, from approximately 4.5 km 750 m. Performance metrics between training and validation datasets no overfitting. was applied different time period compared derived three additional sensors, were not used any stage process, yielding R2 0.99, root mean square errors 1.45 1.58 absolute ranging 1.07 1.15. downscaled shown improve representation both temporal heterogeneity when geostationary- individually. resulting take high observation frequency geostationary data, combined with polar may be added value study seasonal patterns environments.

Язык: Английский

Процитировано

1