UReslham: Radar reflectivity inversion for smart agriculture with spatial federated learning over geostationary satellite observations DOI Creative Commons

Zhengyong Jin,

Xiaolong Xu, Muhammad Bilal

et al.

Computational Intelligence, Journal Year: 2024, Volume and Issue: 40(3)

Published: June 1, 2024

Abstract The frequent occurrence of severe convective weather has certain adverse effects on the smart agriculture industry. To enhance prediction weather, inversion model effectively fills radar reflectivity data gaps by leveraging geostationary satellite data, offering more comprehensive and accurate support for meteorological information in systems. Nevertheless, collaborative cross‐regional driven dispersed faces challenges efficiency, privacy, accuracy. this end, we employ an U‐shaped residual network with embedded light hybrid attention mechanism utilize a federated averaging algorithm efficient distributed training across multiple devices which could preserve privacy from different locations, thereby improving performance. In addition, to address unbalanced nature weighted loss function is designed model's sensitivity high reflectivity. Experimental results demonstrate that proposed exhibits level improvement evaluating performance thresholds compared other models, thus substantiating superiority approach.

Language: Английский

Advances in geocomputation and geospatial artificial intelligence (GeoAI) for mapping DOI Creative Commons
Yongze Song, Margaret Kalácska, Mateo Gašparović

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2023, Volume and Issue: 120, P. 103300 - 103300

Published: April 28, 2023

Geocomputation and geospatial artificial intelligence (GeoAI) have essential roles in advancing geographic information science (GIS) Earth observation to a new stage. GeoAI has enhanced traditional analysis mapping, altering the methods for understanding managing complex human–natural systems. However, there are still challenges various aspects of applications related natural, built, social environments, integrating unique features into models. Meanwhile, data critical components geocomputation studies, as they can effectively reveal patterns, factors, relationships, decision-making processes. This editorial provides comprehensive overview classifying them four categories: (i) buildings infrastructure, (ii) land use analysis, (iii) natural environment hazards, (iv) issues human activities. In addition, summarizes case studies seven categories, including in-situ data, datasets, crowdsourced (i.e., big data), remote sensing photogrammetry LiDAR, statistical data. Finally, presents opportunities future research.

Language: Английский

Citations

41
Improving rainfall-runoff modeling in the Mekong river basin using bias-corrected satellite precipitation products by convolutional neural networks DOI Creative Commons
Xuan-Hien Le, Younghun Kim, Đoàn Văn Bình

et al.

Journal of Hydrology, Journal Year: 2024, Volume and Issue: 630, P. 130762 - 130762

Published: Jan. 24, 2024

Accurate rainfall-runoff (RR) modeling is crucial for effective Mekong River Basin (MRB) water resource management. Satellite precipitation products (SPPs) can offer valuable data such modeling; however, these often exhibit biases that may adversely affect hydrological simulations. This study aimed to improve RR using bias-corrected SPPs and the Soil Water Assessment Tool (SWAT) model MRB. A convolutional neural network-based deep learning framework was employed correct in four (TRMM, PERSIANN-CDR, CHIRPS, CMORPH), resulting respective (ADJ_TRMM, ADJ_CDR, ADJ_CHIR, ADJ_CMOR). The were compared against a gauge-based dataset terms of rainfall analysis, their performance within SWAT assessed over calibration (2004-2013) validation (2014-2015). Bias-corrected demonstrated superior with ADJ_TRMM outperforming other products. results showed satisfactory across all stations, Nash-Sutcliffe Efficiency (NSE) ranging from [0.76-0.87]. Integrating into significantly increased simulations MRB, indicated by higher NSE values [0.72-0.85] uncorrected [-0.37 0.85] at Kratie station. Besides, inconsistent between analysis observed, ADJ_CDR model. These highlight significance applications, especially areas limited ground-based data, need further research refine bias correction methods address limitations

Language: Английский

Citations

13
Enhancing wildfire mapping accuracy using mono-temporal Sentinel-2 data: A novel approach through qualitative and quantitative feature selection with explainable AI DOI Creative Commons
Linh Nguyen Van, Vinh Ngoc Tran, Giang V. Nguyen

et al.

Ecological Informatics, Journal Year: 2024, Volume and Issue: 81, P. 102601 - 102601

Published: April 16, 2024

Accurate wildfire severity mapping (WSM) is crucial in environmental damage assessment and recovery strategies. Machine learning (ML) remote sensing technologies are extensively integrated employed as powerful tools for WSM. However, the intricate nature of ML algorithms often leads to 'black box' systems, obscuring decision-making process significantly limiting stakeholders' ability comprehend basis predictions. This opacity hinders efforts enhance performance risks exacerbating overfitting. present study proposes an innovative WSM approach that incorporates qualitative quantitative feature selection techniques within Explainable AI (XAI) framework. The methodology aims precision provide insights into factors contributing model decisions, thereby increasing interpretability predictions streamlining models improve performance. To achieve this objective, we SHapley Additive exPlanations (SHAP)-Forward Stepwise Selection (FSS) method demonstrate its efficacy elucidating impacts predictors on algorithm performance, accuracy, designed Utilizing post-fire imagery from Sentinel-2 (S2), analyzed ten bands generate 225 unique spectral indices utilizing five different calculations: normalized, algebraic sum, difference, ratio, product forms. Combined with original S2 bands, resulted 235 potential classifications. A random forest was subsequently developed using these optimized through extensive hyperparameter tuning, achieving overall accuracy (OA) 0.917 a Kappa statistic 0.896. most influential were identified SHAP values, FSS narrowing them down 12 critical effective WSM, evidenced by stabilized OA values (0.904 0.881, respectively). Further validation ninefold spatial cross-validation technique demonstrated method's consistent across data partitions, ranging 0.705 0.894 0.607 0.867. By providing more accurate comprehensible XAI-based research contributes broader field monitoring disaster response, underscoring analysis models' capabilities.

Language: Английский

Citations

10
Development of Multi-Source Weighted-Ensemble Precipitation: Influence of bias correction based on recurrent convolutional neural networks DOI Creative Commons

Yung‐Cheng Kao,

Hsiang-En Tsou,

Chia‐Jeng Chen

et al.

Journal of Hydrology, Journal Year: 2024, Volume and Issue: 629, P. 130621 - 130621

Published: Jan. 6, 2024

Accurate precipitation information is the cornerstone of regional hydroclimatic studies, and merging data from various sources provides a means to enhance accuracy. This study aims apply technique referred as Multi-Source Weighted-Ensemble Precipitation (MSWEP) merge gauge-, satellite-, model-based products for Taiwan (MSWEP_TW). To correct known biases in satellite precipitation, long short-term memory (LSTM) emerging convolutional (ConvLSTM) networks are employed. Afterward, how correction influences performance merged assessed. The reveals that LSTM with spatial coherence scheme can show similar effectiveness ConvLSTM increasing correlations gauge-based by ∼10%. MSWEP_TW proven outperform original product (i.e., MSWEP version 2.8), higher weights satellite- over gauge-scarce regions verified. Further, this confirms provide more accurate than gauge-only interpolation, promoting advantage ungauged areas. Lastly, enhancement made directly contributes development satellite-only low latency, suggesting their usefulness near real-time applications.

Language: Английский

Citations

6
Bias correction of the hourly satellite precipitation product using machine learning methods enhanced with high-resolution WRF meteorological simulations DOI

Yao Nan,

Jinyin Ye, Shuai Wang

et al.

Atmospheric Research, Journal Year: 2024, Volume and Issue: 310, P. 107637 - 107637

Published: Aug. 13, 2024

Language: Английский

Citations

4
From bias to accuracy: Transforming satellite precipitation data in arid regions with machine learning and topographical insights DOI Creative Commons
Faisal Baig, Luqman Ali, Muhammad Abrar Faiz

et al.

Journal of Hydrology, Journal Year: 2025, Volume and Issue: unknown, P. 132801 - 132801

Published: Feb. 1, 2025

Language: Английский

Citations

0
Unveiling environmental drivers of soil erosion in South Korea through SHAP-informed machine learning DOI
Linh Nguyen Van, Giang V. Nguyen, Minho Yeon

et al.

Land Use Policy, Journal Year: 2025, Volume and Issue: 155, P. 107592 - 107592

Published: May 9, 2025

Language: Английский

Citations

0
Deep neural network-based discharge prediction for upstream hydrological stations: a comparative study DOI
Xuan-Hien Le, Duc Hai Nguyen, Sungho Jung

et al.

Earth Science Informatics, Journal Year: 2023, Volume and Issue: 16(4), P. 3113 - 3124

Published: Aug. 21, 2023

Language: Английский

Citations

9
Integrated Evaluation and error decomposition of four gridded precipitation products using dense rain gauge observations over the Yunnan-Kweichow Plateau, China DOI Creative Commons
Tianjian Lu,

Qingquan Xiao,

Hanyu Lu

et al.

European Journal of Remote Sensing, Journal Year: 2024, Volume and Issue: 57(1)

Published: March 11, 2024

Evaluating the precision and applicability of high-quality precipitation products in distinctive terrain intricate climate Yunnan-Kweichow Plateau (YKP) is pivotal for research. This study comprehensively assesses four gridded datasets (AERA5-Asia, AIMERG, ERA5-Land, IMERG-Final) using China Meteorological Administration's surface data. It employs eight statistical indicators error decomposition methods at various spatiotemporal scales. The main findings are as follows: (1) AERA5-Asia, IMERG-Final show similar patterns, with ERA5-Land overestimating. While all display minor seasonal variations, AERA5-Asia underestimates summer rain. tends to overstate, whereas AIMERG generally accurate but slightly undervalued southern YKP. (2) Hourly analysis reveals leads performance metrics (CC: 0.23, MAE: 0.49 mm/hour, RMSE: 0.18 CSI: 0.27). In contrast, lags, marked by lowest BIAS (35.39%), FAR (0.74), FBI (2.85). comparable results underperform CC (0.16, 0.13), POD (0.31, 0.30), CSI (0.19, 0.18). (3) False bias significantly contributes total products. mitigate enhance false situations through calibration algorithms, albeit introducing missed central region offer valuable insights into YKP precipitation, informing development grid-based fusion algorithms region's complex terrain.

Language: Английский

Citations

2
An Innovative Correction–Fusion Approach for Multi-Satellite Precipitation Products Conditioned by Gauge Background Fields over the Lancang River Basin DOI Creative Commons

Linjiang Nan,

Mingxiang Yang, Hao Wang

et al.

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(11), P. 1824 - 1824

Published: May 21, 2024

Satellite precipitation products can help improve estimates where ground-based observations are lacking; however, their relative accuracy and applicability in data-scarce areas remain unclear. Here, we evaluated the of different satellite datasets for Lancang River Basin, Western China, including Tropical Rainfall Measuring Mission (TRMM) 3B42RT, Global Precipitation Measurement Integrated Multi-satellitE Retrievals (GPM IMERG), Fengyun 2G (FY-2G) datasets. The results showed that GPM IMERG FY-2G superior to TRMM 3B42RT meeting local research needs. A subsequent bias correction on these two significantly increased correlation coefficient probability detection reduced error indices such as root mean square absolute error. To further data quality, proposed a novel correction–fusion method based window sliding Bayesian fusion. Specifically, corrected dataset was merged with Early, Late, Final Runs. resulting FY-Early, FY-Late, FY-Final fusion high coefficients, strong performances, few observation errors, thereby effectively extending sources. this study provide scientific basis rational use areas, well reliable support forecasting water resource management Basin.

Language: Английский

Citations

2