Assessing the Turbulence Kinetic Energy Budget in the Boundary Layer Using WRF-LES: Impact of Momentum Perturbation DOI
Mukesh Kumar, Tirtha Banerjee, Alexandra Jonko

и другие.

Опубликована: Март 4, 2021

<p>Mesoscale-to-Large Eddy Simulation (LES) grid nesting is an important tool for many atmospheric model applications, ranging from wind energy to wildfire spread studies. Different techniques are used in such applications accelerate the development of turbulence LES domain. Here, we explore impact a simple and computationally efficient Stochastic Cell Perturbation method (SCPM) generation Weather Research Forecasting (WRF) on Turbulence Kinetic Energy (TKE) budget. In convective boundary layer, study variation TKE budget terms under initial conditions Scaled Wind Farm Technology (SWiFT) facility located West Texas. this study, WRF with horizontal resolution 12 m, one-way nested within idealized mesoscale It crucial understand how forced perturbation shifts balance between quantify shear production, buoyant production unstable case. Since additional introduced SCPM method, investigate dissipation term TKE. addition, also turbulent transport. Generally, it integrates over height null planar homogeneous case without subsidence, indicating positive some heights negative other heights. Furthermore, transport after extending random up certain height. The findings will provide better understanding contribution different simulation.</p>

Язык: Английский

LD-YOLO: A Lightweight Dynamic Forest Fire and Smoke Detection Model with Dysample and Spatial Context Awareness Module DOI Open Access
Zhenyu Lin, Bensheng Yun, Yanan Zheng

и другие.

Forests, Год журнала: 2024, Номер 15(9), С. 1630 - 1630

Опубликована: Сен. 15, 2024

The threat of forest fires to human life and property causes significant damage society. Early signs, such as small smoke, are often difficult detect. As a consequence, early detection smoke is crucial. Traditional fire models have shortcomings, including low accuracy efficiency. YOLOv8 model exhibits robust capabilities in detecting smoke. However, it struggles balance accuracy, complexity, speed. This paper proposes LD-YOLO, lightweight dynamic based on the YOLOv8, detect Firstly, GhostConv introduced generate more feature maps through low-cost linear transformations, while maintaining high reducing parameters. Secondly, we propose C2f-Ghost-DynamicConv an effective tool for increasing extraction representing from fires. method aims optimize use computing resources. Thirdly, introduce DySample address loss fine-grained detail initial images. A point-based sampling utilized enhance resolution small-target images without imposing additional computational burden. Fourthly, Spatial Context Awareness Module (SCAM) insufficient representation background interference. Also, self-attention head (SADH) designed capture global features. Lastly, Shape-IoU, which emphasizes importance boundaries’ shape scale, used improve experimental results show that LD-YOLO realizes mAP0.5 86.3% custom dataset, 4.2% better than original model, with 36.79% fewer parameters, 48.24% lower FLOPs, 15.99% higher FPS. Therefore, indicates fast speed, complexity. crucial timely

Язык: Английский

Процитировано

7
An Improved Fire and Smoke Detection Method Based on YOLOv8n for Smart Factories DOI Creative Commons
Ziyang Zhang,

Lingye Tan,

Tiong Lee Kong Robert

и другие.

Sensors, Год журнала: 2024, Номер 24(15), С. 4786 - 4786

Опубликована: Июль 24, 2024

Factories play a crucial role in economic and social development. However, fire disasters factories greatly threaten both human lives properties. Previous studies about detection using deep learning mostly focused on wildfire ignored the fires that happened factories. In addition, lots of focus detection, while smoke, important derivative disaster, is not detected by such algorithms. To better help smart monitor disasters, this paper proposes an improved smoke method based YOLOv8n. ensure quality algorithm training process, self-made dataset including more than 5000 images their corresponding labels created. Then, nine advanced algorithms are selected tested dataset. YOLOv8n exhibits best results terms accuracy speed. ConNeXtV2 then inserted into backbone to enhance inter-channel feature competition. RepBlock SimConv replace original Conv improve computational ability memory bandwidth. For loss function, CIoU replaced MPDIoU efficient accurate bounding box. Ablation tests show our achieves performance all four metrics reflecting accuracy: precision, recall, F1, mAP@50. Compared with model, whose approximately 90%, modified above 95%. mAP@50 particular reaches 95.6%, exhibiting improvement 4.5%. Although complexity improves, requirements real-time monitoring satisfied.

Язык: Английский

Процитировано

4
Forest fire regimes in the Northwestern Himalayas: unravelling microlevel impact of topography, weather, and human activity on fire behaviour DOI

B Alton Paul,

U.C. Dumka, Somnath Bar

и другие.

International Journal of Remote Sensing, Год журнала: 2025, Номер unknown, С. 1 - 26

Опубликована: Апрель 20, 2025

Язык: Английский

Процитировано

0
An Efficient Task Implementation Modeling Framework with Multi-Stage Feature Selection and AutoML: A Case Study in Forest Fire Risk Prediction DOI Creative Commons
Ye Su, Longlong Zhao, Hongzhong Li

и другие.

Remote Sensing, Год журнала: 2024, Номер 16(17), С. 3190 - 3190

Опубликована: Авг. 29, 2024

As data science advances, automated machine learning (AutoML) gains attention for lowering barriers, saving time, and enhancing efficiency. However, with increasing dimensionality, AutoML struggles large-scale feature sets. Effective selection is crucial efficient in multi-task applications. This study proposes an modeling framework combining a multi-stage (MSFS) algorithm AutoSklearn, robust framework, to address high-dimensional challenges. The MSFS includes three stages: mutual information gain (MIG), recursive elimination cross-validation (RFECV), voting aggregation mechanism, ensuring comprehensive consideration of correlation, importance, stability. Based on multi-source time series remote sensing data, this pioneers the application AutoSklearn forest fire risk prediction. Using case study, we compare five other (FS) algorithms, including single FS algorithms two hybrid algorithms. Results show that selects half original features (12/24), effectively handling collinearity (eliminating 11 out 13 collinear groups) AutoSklearn’s success rate by 15%, outperforming same number 7% 5%. Among six non-FS, demonstrates highest prediction performance stability minimal variance (0.09%) across evaluation metrics. efficiently filters redundant features, operational efficiency generalization ability tasks. MSFS–AutoSklearn significantly improves AutoML’s production accuracy, facilitating implementation various real-world tasks wider AutoML.

Язык: Английский

Процитировано

2
Impact of Momentum Perturbation on Convective Boundary Layer Turbulence DOI Creative Commons
Mukesh Kumar, Alexandra Jonko, William Lassman

и другие.

Journal of Advances in Modeling Earth Systems, Год журнала: 2024, Номер 16(2)

Опубликована: Фев. 1, 2024

Abstract Mesoscale‐to‐microscale coupling is an important tool for conducting turbulence‐resolving multiscale simulations of realistic atmospheric flows, which are crucial applications ranging from wind energy to wildfire spread studies. Different techniques used facilitate the development turbulence in large‐eddy simulation (LES) domain while minimizing computational cost. Here, we explore impact a simple and computationally efficient Stochastic Cell Perturbation method using momentum perturbation (SCPM‐M) accelerate generation boundary‐coupled LES Weather Research Forecasting model. We simulate convective boundary layer (CBL) characterize production dissipation turbulent kinetic (TKE) variation TKE budget terms. Furthermore, evaluate applying perturbations three magnitudes below, up to, above CBL on Momentum greatly reduce fetch associated with generation. When applied half vertical extent layer, produce adequate amount turbulence. However, when CBL, additional structures generated at top near inversion layer. The budgets produced by SCPM‐M varying heights different amplitudes always higher surface than those No‐SCPM, as their contributions TKE. This study provides better understanding how reduces costs terms contribute simulation.

Язык: Английский

Процитировано

1
Effects of Dust Storm and Wildfire Events on Phytoplankton Growth and Carbon Sequestration in the Tasman Sea, Southeast Australia DOI Creative Commons
Hiep Nguyen Duc, John Leys, Matthew Riley

и другие.

Atmosphere, Год журнала: 2024, Номер 15(3), С. 337 - 337

Опубликована: Март 8, 2024

Dust storms and wildfires occur frequently in south-eastern Australia. Their effects on the ecology, environment population exposure have been focus of many studies recently. do not emit ground-sequestered carbon, but significant quantities carbon into atmosphere. However, both natural events promote phytoplankton growth water bodies because other trace elements such as iron, deposit surface oceans. Carbon dioxide is reabsorbed by via photosynthesis. The balance cycle due to dust well known. Recent emission 2019–2020 summer eastern Australia indicated that this megafire event emitted approximately 715 million tonnes CO2 (195 Tg C) atmosphere from burned forest areas. This study focusses association southeastern with Tasman Sea February 2019 storm Black Summer wildfires. Central western New South Wales were sources (11 16 2019), occurred along coast Victoria (from early November January 2020). WRF-Chem model used for simulation AFWA (Air Force Weather Agency US) version GOCART model, wildfire FINN (Fire Emission Inventory NCAR) data. results show similarities differences deposition particulate matter, reabsorption patterns these events. A higher rate PM2.5 ocean corresponds a growth. Using during 5-day 2019, ~1230 tons total was predicted deposited Sea, while ~132,000 PM10 stage 1 8 2019.

Язык: Английский

Процитировано

1
Interpolation of Temperature in a Mountainous Region Using Heterogeneous Observation Networks DOI Creative Commons

Soorok Ryu,

Joon Jin Song, GyuWon Lee

и другие.

Atmosphere, Год журнала: 2024, Номер 15(8), С. 1018 - 1018

Опубликована: Авг. 22, 2024

Accurately generating high-resolution surface grid datasets often involves merging multiple weather observation networks and addressing the challenge of network heterogeneity. This study aims to tackle problem accurately interpolating temperature data in regions with a complex topography. To achieve this, we introduce deterministic interpolation method that incorporates elevation enhance accuracy datasets. is particularly valuable for areas intricate terrains. Our robust methodology integrates harmonization radial basis function (RBF) topographical regions. The was tested on 10 min average from Jeju Island, South Korea, over 2 years had spatial resolution 100 m. results show significant reduction 5.5% error rates, an 0.73 °C 0.69 °C, by incorporating all adjusted data. Integrating parameterized nonlinear profile further enhances accuracy, yielding 4.4% compared linear model. method, based regression-based functions, demonstrates 6.7% improvement kriging same profile. research offers approach precise interpolation, especially

Язык: Английский

Процитировано

1
Wildfire Towers Drive Firebrand Lofting: Insights from Coupled Fire-Atmosphere Model Simulations DOI Creative Commons

Mukesh Kumar,

Alexander Jon Josephson, Eunmo Koo

и другие.

Research Square (Research Square), Год журнала: 2024, Номер unknown

Опубликована: Июнь 18, 2024

Abstract Previous studies have highlighted the complexity of wildfire behavior, emphasizing significance firebrand dynamics in contributing to spread and severity wildfires. While these provide foundational knowledge, specific role wildland fires' towers troughs lofting has never been addressed. This work aims illustrate intricate relationship between fire tower trough phenomena lofting. Through physics-based simulations, we show presence drives spatial distribution generated firebrands as well vertical trajectory lofted firebrands. We found that majority (78.85 %) get from which are regions updrafts while remaining enter into during process severally limits height distance they travel. The results this study helpful for foresters land managers planning researchers advancing existing model capabilities can save communities enhance safety firefighters wind-driven fires where there higher risk spot fires.

Язык: Английский

Процитировано

0
How well are hazards associated with derechos reproduced in regional climate simulations? DOI Creative Commons
Tristan J. Shepherd, Frederick Letson, R. J. Barthelmie

и другие.

Natural hazards and earth system sciences, Год журнала: 2024, Номер 24(12), С. 4473 - 4505

Опубликована: Дек. 10, 2024

Abstract. A 15-member ensemble of convection-permitting regional simulations the fast-moving and destructive derecho 29–30 June 2012 that impacted northeastern urban corridor USA is presented. This event generated 1100 reports damaging winds, significant wind gusts over an extensive area up to 500 000 km2, caused several fatalities, resulted in widespread loss electrical power. Extreme events such as this are increasingly being used within pseudo-global-warming experiments examine sensitivity historical, societally important global climate non-stationarity how they may evolve a result changing thermodynamic dynamic contexts. As it fidelity with which described hindcast experiments. The presented herein performed using Weather Research Forecasting (WRF) model. resulting explore simulation relative observations for gust magnitudes, spatial scales convection (as manifest high composite reflectivity, cREF), both rainfall hail production function model configuration (microphysics parameterization, lateral boundary conditions (LBCs), start date, use nudging, compiler choice, damping, number vertical levels). We also degree each member differs respect key mesoscale drivers convective systems (e.g., available potential energy shear) critical manifestations deep convection, e.g., velocities, cold-pool generation, those properties relate correct characterization associated atmospheric hazards (wind hail). Use double-moment, seven-class scheme concentrations all species (including graupel) results greatest model-simulated structure event. All members, however, fail capture intensity terms extent near-surface gusts. further show very LBCs employed specifically higher nested ERA-Interim compared ERA5. Excess (CAPE) members after passage leads excess cells, gusts, cREF > 40 dBZ, precipitation during frontal on subsequent day. proved challenging forecast real time reproduce here. Future work could if other initial can achieve greater fidelity.

Язык: Английский

Процитировано

0
Assessing the Turbulence Kinetic Energy Budget in the Boundary Layer Using WRF-LES: Impact of Momentum Perturbation DOI
Mukesh Kumar, Tirtha Banerjee, Alexandra Jonko

и другие.

Опубликована: Март 4, 2021

<p>Mesoscale-to-Large Eddy Simulation (LES) grid nesting is an important tool for many atmospheric model applications, ranging from wind energy to wildfire spread studies. Different techniques are used in such applications accelerate the development of turbulence LES domain. Here, we explore impact a simple and computationally efficient Stochastic Cell Perturbation method (SCPM) generation Weather Research Forecasting (WRF) on Turbulence Kinetic Energy (TKE) budget. In convective boundary layer, study variation TKE budget terms under initial conditions Scaled Wind Farm Technology (SWiFT) facility located West Texas. this study, WRF with horizontal resolution 12 m, one-way nested within idealized mesoscale It crucial understand how forced perturbation shifts balance between quantify shear production, buoyant production unstable case. Since additional introduced SCPM method, investigate dissipation term TKE. addition, also turbulent transport. Generally, it integrates over height null planar homogeneous case without subsidence, indicating positive some heights negative other heights. Furthermore, transport after extending random up certain height. The findings will provide better understanding contribution different simulation.</p>

Язык: Английский

Процитировано

1