Assessing multi-source random forest classification and robustness of predictor variables in flooded areas mapping

Remote Sensing Applications Society and Environment, Год журнала: 2024, Номер 35, С. 101239 - 101239

Опубликована: Май 10, 2024

Flood extent delineation techniques have benefited from the increasing availability of remote sensing imagery, classification and introduction geomorphic descriptors derived Digital Elevation Models (DEM). On other hand, high-performing Machine Learning (ML) methods allowed for development accurate flood maps by integrating several predictor variables into supervised or unsupervised algorithms. Among others, Random Forest (RF) is a powerful widely applied ML classifier, providing predictions also with complex datasets varying parameters set. In present study, effectiveness this algorithm mapping flooded areas was evaluated. Various geospatial data sources were integrated, including morphological indicators, such as Geomorphic Index (GFI), Sentinel-2 bands, multispectral indices, Sentinel-1 polarizations. The reliability under different training sample sizes evaluated accuracy RF classifier assessed. Moreover, exploring ability to identify most important variables, predictors contributing identified their stability investigated. To gauge adaptability consistency these features, we our analyses study around World. results indicate that certain displayed remarkable across remained robust various parameters. However, some variability in structure features related specific complexities each considered case observed.

Язык: Английский

---

Understanding future hydrologic challenges: Modelling the impact of climate change on river runoff in central Italy

Environmental Challenges, Год журнала: 2024, Номер 15, С. 100899 - 100899

Опубликована: Март 24, 2024

Central Italy's diverse ecosystems and landscapes are susceptible to the Mediterranean climate change, affecting water resources riverine systems. Managing these is crucial for nation's sustainable development resilience. This research assesses potential long-term change impacts on river runoff in central highly regulated Aterno-Pescara River watershed. We simulate current future using Soil Water Assessment Tool (SWAT+). Climate projections from 5 Global Models (GCMs) under two emissions scenarios used quantify drought characteristic changes SWAT+ investigate (2015 – 2100) runoff. All GCMs predicted increasing daily temperature (up 0.6 °C decade−1 at 95% confidence level) decreasing precipitation trends (-16.4 mm decade−1), resulting negative (-0.036 m3s−1 decade−1). Uncertainties exist regarding variable magnitudes among scenarios. Analyzing 12-month standardized indices data revealed a strong correlation between (Pearson coefficient ranges 0.63 - 0.93 GCMs). The run-sum technique both showed frequent, severe, prolonged droughts, with meteorological droughts possibly lasting up 105 months (severity 163) hydrological exceeding 100 over 150). study provides insights policymakers, emphasizing need strategies addressing sustainability.

Язык: Английский

---

Spatial Assessment of Flood Susceptibility in Assam, India: A Comparative Study of Frequency Ratio and Shannon’s Entropy Models

Journal of the Indian Society of Remote Sensing, Год журнала: 2024, Номер 52(2), С. 343 - 358

Опубликована: Фев. 1, 2024

Язык: Английский

---

Fracture behavior of sandstone with partial filling flaw under mixed-mode loading: Three-point bending tests and discrete element method (DEM) numerical approach

Journal of Rock Mechanics and Geotechnical Engineering, Год журнала: 2024, Номер unknown

Опубликована: Апрель 1, 2024

Язык: Английский

---

Extreme Weather Events and their Socioeconomic Impacts: A Remote Sensing-Based Analysis of Flood Damages

Global and earth surface processes change., Год журнала: 2024, Номер 1, С. 100001 - 100001

Опубликована: Июнь 1, 2024

Язык: Английский

---

Changes in climate, vegetation cover and vegetation composition affect runoff generation in the Gulf of Guinea Basin

Hydrological Processes, Год журнала: 2024, Номер 38(3)

Опубликована: Март 1, 2024

Abstract Although considerable effort has been deployed to understand the impact of climate variability and vegetation change on runoff in major basins across Africa, such studies are scarce Gulf Guinea Basin (GGB). This study combines Budyko framework elasticity concept along with geospatial data fill this research gap 44 nested sub‐basins GGB. Annual rainfall from 1982 2021 show significant decreasing increasing trends northern southern parts GGB, respectively. potential evapotranspiration (PET) also shows higher magnitudes observed Changing variables corroborates shift arid wetter conditions north south, From 2000 2020 cover estimated using enhanced index (EVI) all including those experiencing a decline annual rainfall. Vegetation composition measured continuous fields (VCFs) an increase tree canopy (TC), short marginal changes bare ground (BG). Elasticity coefficients that 10% PET may lead 33% 24% runoff, On other hand, EVI 4% while TC, SV BG reduce by 3% 2%, Even though marginal, decomposing into different parameters VCFs hydrological effects which is one novelties be used for implementing nature‐based solutions. The demonstrates freely available together analytical methods promising approach understanding hydrology data‐scarce regions.

Язык: Английский

---

LightGBM hybrid model based DEM correction for forested areas

PLoS ONE, Год журнала: 2024, Номер 19(10), С. e0309025 - e0309025

Опубликована: Окт. 7, 2024

The accuracy of digital elevation models (DEMs) in forested areas plays a crucial role canopy height monitoring and ecological sensitivity analysis. Despite extensive research on DEMs recent years, significant errors still exist due to factors such as occlusion, terrain complexity, limited penetration, posing challenges for subsequent analyses based DEMs. Therefore, CNN-LightGBM hybrid model is proposed this paper, with four different types forests (tropical rainforest, coniferous forest, mixed broad-leaved forest) selected study sites validate the performance correcting COP30DEM forest area In choice was made use Densenet architecture CNN LightGBM primary model. This LightGBM’s leaf-growth strategy histogram linking methods, which are effective reducing data’s memory footprint utilising more data without sacrificing speed. uses values from ICESat-2 ground truth, covering several parameters including COP30DEM, height, coverage, slope, roughness relief amplitude. To superiority correction compared other models, test model, CNN-SVR SVR conducted within same sample space. prevent issues overfitting or underfitting during training, although common meta-heuristic optimisation algorithms can alleviate these problems certain extent, they have some shortcomings. overcome shortcomings, paper cites an improved SSA search algorithm that incorporates ingestion FA increase diversity solutions global capability, Firefly Algorithm-based Sparrow Search Optimization Algorithm (FA-SSA algorithm) introduced. By comparing multiple validating airborne LiDAR reference dataset, results show R 2 (R-Square) improves by than 0.05 performs better experiments. FA-SSA-CNN-LightGBM has highest accuracy, RMSE 1.09 meters, reduction 30% when models. Compared (such FABDEM GEDI), its 50%, significantly commonly used areas, indicating feasibility method importance advancing topographic mapping.

Язык: Английский

---

GIS-Based Modeling for Water Resource Monitoring and Management: A Critical Review

Springer geography, Год журнала: 2025, Номер unknown, С. 537 - 561

Опубликована: Янв. 1, 2025

Язык: Английский

---

Assessing Solar Energy Potential and CO2 Reduction through GIS-Based Spatial Analysis for Optimal Site Selection of Photovoltaic Systems in Jeddah, Saudi Arabia.

Remote Sensing Applications Society and Environment, Год журнала: 2025, Номер unknown, С. 101537 - 101537

Опубликована: Апрель 1, 2025

Язык: Английский

---

Streamflow Prediction with Time-Lag-Informed Random Forest and Its Performance Compared to SWAT in Diverse Catchments

Water, Год журнала: 2024, Номер 16(19), С. 2805 - 2805

Опубликована: Окт. 2, 2024

This study introduces a time-lag-informed Random Forest (RF) framework for streamflow time-series prediction across diverse catchments and compares its results against SWAT predictions. We found strong evidence of RF’s better performance by adding historical flows time-lags meteorological values over using only actual values. On daily scale, RF demonstrated robust (Nash–Sutcliffe efficiency [NSE] > 0.5), whereas generally yielded unsatisfactory (NSE < 0.5) tended to overestimate up 27% (PBIAS). However, provided monthly predictions, particularly in with irregular flow patterns. Although both models faced challenges predicting peak snow-influenced catchments, outperformed an arid catchment. also exhibited notable advantage terms computational efficiency. Overall, is good choice predictions limited data, preferable understanding hydrological processes depth.

Язык: Английский

---