Spatial Interpolation of Seasonal Precipitations Using Rain Gauge Data and Convection‐Permitting Regional Climate Model Simulations in a Complex Topographical Region DOI Creative Commons
Valentin Dura, Guillaume Évin, Anne‐Catherine Favre

et al.

International Journal of Climatology, Journal Year: 2024, Volume and Issue: 44(16), P. 5745 - 5760

Published: Oct. 27, 2024

ABSTRACT In mountainous areas, accurately estimating the long‐term climatology of seasonal precipitations is challenging due to lack high‐altitude rain gauges and complexity topography. This study addresses these challenges by interpolating precipitation data from 3189 across France over 1982–2018 period, using geographical coordinates, altitude. this study, an additional predictor provided simulations a Convection‐Permitting Regional Climate Model (CP‐RCM). The are averaged obtain climatology, which helps capture relationship between topography precipitation. Geostatistical machine learning models evaluated within cross‐validation framework determine most appropriate approach generate reference fields. Results indicate that best model uses interpolate ratio observations CP‐RCM simulations. method successfully reproduces both mean variance observed data, slightly outperforms geostatistical model. Moreover, incorporating outputs as explanatory variable significantly improves interpolation accuracy altitude extrapolation, especially when gauge density low. These results imply commonly used altitude‐precipitation may be insufficient derive simulations, increasingly available worldwide, present opportunity for improving interpolation, in sparse complex topographical regions.

Language: Английский

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
Multi-source precipitation estimation using machine learning: Clarification and benchmarking DOI
Yue Xu, Guoqiang Tang, Lingjie Li

et al.

Journal of Hydrology, Journal Year: 2024, Volume and Issue: 635, P. 131195 - 131195

Published: April 6, 2024

Language: Английский

Citations

5
A Random Forest-Based Precipitation Detection Algorithm for FY-3C/3D MWTS2 over Oceanic Regions DOI Creative Commons
Tengling Luo, Yi Yu, Gang Ma

et al.

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(9), P. 1566 - 1566

Published: April 28, 2025

Satellite microwave-sounding radiometer data assimilation under clear-sky conditions typically requires the exclusion of precipitation-affected field-of-view (FOV) regions. However, traditional scatter index (SI) and cloud liquid water path (CLWP)-based precipitation sounding algorithms from earlier NOAA microwave sounders are built on window channels which not available FY-3C/D MWTS-II. To address this limitation, study establishes a nonlinear relationship between multispectral visible/infrared FY-2F geostationary satellite using an artificial intelligence (AI)-driven approach. The methodology involves three key steps: (1) spatiotemporal integration VISSR-derived products with NOAA-19 AMSU-A brightness temperatures was achieved through GEO-LEO pixel fusion algorithm. (2) fused observations were used as training set input into random forest model. (3) performance RF_SI method evaluated by individual cases time series observations. Results demonstrate that effectively captures horizontal distribution scattering signals in deep convective systems. Compared those SI CLWP-based algorithms, accuracy rate exceed 94% 92%, respectively, error is less than 3%. Also, exhibits consistent across diverse temporal spatial domains, highlighting its robustness for cross-platform screening assimilation.

Language: Английский

Citations

0
Quantitative evaluation of uncertainty and interpretability in machine learning-based landslide susceptibility mapping through feature selection and explainable AI DOI Creative Commons
Xuan-Hien Le,

Chanul Choi,

Song Eu

et al.

Frontiers in Environmental Science, Journal Year: 2024, Volume and Issue: 12

Published: July 30, 2024

Landslide susceptibility mapping (LSM) is essential for determining risk regions and guiding mitigation strategies. Machine learning (ML) techniques have been broadly utilized, but the uncertainty interpretability of these models not well-studied. This study conducted a comparative analysis assessment five ML algorithms—Random Forest (RF), Light Gradient-Boosting (LGB), Extreme Gradient Boosting (XGB), K-Nearest Neighbor (KNN), Support Vector (SVM)—for LSM in Inje area, South Korea. We optimized using Bayesian optimization, method that refines model performance through probabilistic model-based tuning hyperparameters. The algorithms was evaluated accuracy, Kappa score, F 1 with accuracy detecting landslide-prone locations ranging from 0.916 to 0.947. Among them, tree-based (RF, LGB, XGB) showed competitive outperformed other models. Prediction quantified bootstrapping Monte Carlo simulation methods, latter providing more consistent estimate across Further, predictions analyzed sensitivity SHAP values. also expanded our investigation include both inclusion exclusion predictors, insights into each significant variable comprehensive analysis. paper provides predictive LSM, contributing future research Korea beyond.

Language: Английский

Citations

3
Evaluation of multi-source precipitation products for monitoring drought across China DOI Creative Commons

Yongyi Yuan,

Bingzhen Liao

Frontiers in Environmental Science, Journal Year: 2025, Volume and Issue: 13

Published: Jan. 24, 2025

Accurate precipitation data are crucial for effective drought monitoring, especially in China’s complex and diverse climatic regions. This study evaluates the performance of six multisource products-ERA5-Land, CMORPH CRT, GSMaP MVK, IMERG Late, Final-in detecting across China from 2009 to 2019, using ground station observations validation. By applying various evaluation indices timescales, this analysis captures short long-term climate variations, assessing each product’s accuracy Spatial temporal analyses revealed that Final closely aligns with observed precipitation, particularly high-rainfall areas like Yangtze River Basin, while MVK ERA5 tend overestimate arid semi-arid Discrepancies most pronounced terrains such as Qinghai-Tibet Plateau southwestern mountains, where sparse observational networks exacerbate errors. Drought indices, including SPEI-3 SPI-1, were used measure effectiveness intensity, frequency, duration. consistently showed highest correlation all levels (Light, Moderate, Severe), tended occurrences certain drought-prone areas. Hotspot CDD, PRCPTOT, R95p further confirmed Final’s identifying wet event patterns, reflecting measurements, whereas occasionally overestimated frequencies. In summary, emerged a relatively accurate reliable product showing strong applicability These findings aid correction, enhances understanding regional variability, integration strategies improve water resource management extreme monitoring.

Language: Английский

Citations

0
Advanced stepwise machine learning integration of near-real-time precipitation products in China's flood-prone basins DOI
Lingxue Liu,

Huajin Lei

Atmospheric Research, Journal Year: 2025, Volume and Issue: unknown, P. 108197 - 108197

Published: May 1, 2025

Language: Английский

Citations

0
Nueva base de datos de precipitaciones mensuales de la república Argentina (PMRAv1), 2000-2022 DOI Creative Commons
Juan Gaitán,

Lucio Biancari

Meteorologica, Journal Year: 2024, Volume and Issue: unknown, P. 032 - 032

Published: June 28, 2024

La representación precisa de los patrones espacio-temporales precipitación es un insumo esencial para numerosas aplicaciones ambientales. Sin embargo, la estimación derivados únicamente pluviómetros está sujeta a grandes incertidumbres, especialmente en regiones con escasez datos. Presentamos una nueva base datos Precipitaciones Mensuales República Argentina (PMRAv1) el período 2000-2022, 5 km resolución espacial. PMRAv1 utiliza metodología basada regresión bosques aleatorios (Regression Random Forest) combinar mensuales mediciones terrestres (entre 142 y 227 estaciones cada mes), cuatro productos globales estimada por satélite, modelos circulación atmosféricas o interpolación medidos terreno (TERRACLIMATE, ERA5-LAND, GPMv6 PERSIANN-CDR) objetivo mejorar las precipitaciones Argentina. desarrollada pudo espacio-temporal al permitir fusión múltiples fuentes información satelital terreno. validación realizada utilizando 30% mostró que mejora significativamente parámetros RMSE, MAE, EM R2 comparación precipitación. Además, producto resultó más estable predicción valores observados, presentar menor desvío estándar tres ajuste. se pone disposición usuarios varios formatos. El término ‘v1’ (versión 1) hace referencia considera este tendrá sucesivas versiones futuro permitan actualizarla precisión estimaciones. Asimismo, método presentado también podría ser utilizado otras variables climatológicas cuando disponga

Citations

1
Evaluating satellite-based precipitation products for spatiotemporal drought analysis DOI

Hussain Masood Khan,

Muhammad Fahim Aslam, Muhammad Waseem

et al.

Journal of Arid Environments, Journal Year: 2024, Volume and Issue: 224, P. 105225 - 105225

Published: July 14, 2024

Language: Английский

Citations

1
Underutilized Feature Extraction Methods for Burn Severity Mapping: A Comprehensive Evaluation DOI Creative Commons
Linh Nguyen Van, Giha Lee

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(22), P. 4339 - 4339

Published: Nov. 20, 2024

Wildfires increasingly threaten ecosystems and infrastructure, making accurate burn severity mapping (BSM) essential for effective disaster response environmental management. Machine learning (ML) models utilizing satellite-derived vegetation indices are crucial assessing wildfire damage; however, incorporating many can lead to multicollinearity, reducing classification accuracy. While principal component analysis (PCA) is commonly used address this issue, its effectiveness relative other feature extraction (FE) methods in BSM remains underexplored. This study aims enhance ML classifier accuracy by evaluating various FE techniques that mitigate multicollinearity among indices. Using composite index (CBI) data from the 2014 Carlton Complex fire United States as a case study, we extracted 118 seven Landsat-8 spectral bands. We applied compared 13 different techniques—including linear nonlinear such PCA, t-distributed stochastic neighbor embedding (t-SNE), discriminant (LDA), Isomap, uniform manifold approximation projection (UMAP), factor (FA), independent (ICA), multidimensional scaling (MDS), truncated singular value decomposition (TSVD), non-negative matrix factorization (NMF), locally (LLE), (SE), neighborhood components (NCA). The performance of these was benchmarked against six classifiers determine their improving Our results show alternative outperform computational efficiency. Techniques like LDA NCA effectively capture relationships critical BSM. contributes existing literature providing comprehensive comparison methods, highlighting potential benefits underutilized

Language: Английский

Citations

1
Enhancing Rainfall Estimates in the Indian Monsoon Season Using Multi-Source Satellite Data: Development and Evaluation DOI

Amarjyothi Kasimahanthi,

D. Preveen Kumar,

M. Reddy

et al.

Published: Jan. 1, 2024

Language: Английский

Citations

0