Segmentation of Any Fire Event (SAFE): A Rapid and High-Precision Approach for Burned Area Extraction Using Sentinel-2 Imagery DOI Creative Commons
Shuaijun Liu, Yong Xue, Hui Chen

и другие.

Remote Sensing, Год журнала: 2024, Номер 17(1), С. 54 - 54

Опубликована: Дек. 27, 2024

The timely and accurate monitoring of wildfires other sudden natural disasters is crucial for safeguarding the safety residents their property. Satellite imagery wildfire offers a unique opportunity to obtain near-real-time disaster information through rapid, large-scale remote sensing mapping. However, existing methods are constrained by temporal spatial limitations imagery, preventing comprehensive fulfillment need high resolution in early warning. To address this gap, we propose high-precision extraction method without training—SAFE. SAFE combines generalization capabilities Segmentation Anything Model (SAM) effectiveness hotspot product data such as MODIS VIIRS. employs two-step localization strategy incrementally identify burned areas pixels post-wildfire thereby reducing computational load providing high-resolution impact areas. area generated can subsequently be used train lightweight regional models, establishing detection models applicable various regions, ultimately undetected We validated four test regions representing two typical scenarios—grassland forest. results showed that SAFE’s F1-score was, on average, 9.37% higher than alternative methods. Additionally, application scenarios demonstrated its potential capability detect fine distribution impacts global scale.

Язык: Английский

Modeling of Forest Fire Risk Areas of Amazonas Department, Peru: Comparative Evaluation of Three Machine Learning Methods DOI Open Access
Alex J. Vergara, Sivmny V. Valqui-Reina, Dennis Cieza-Tarrillo

и другие.

Forests, Год журнала: 2025, Номер 16(2), С. 273 - 273

Опубликована: Фев. 5, 2025

Forest fires are the result of poor land management and climate change. Depending on type affected eco-system, they can cause significant biodiversity losses. This study was conducted in Amazonas department Peru. Binary data obtained from MODIS satellite occurrence between 2010 2022 were used to build risk models. To avoid multicollinearity, 12 variables that trigger selected (Pearson ≤ 0.90) grouped into four factors: (i) topographic, (ii) social, (iii) climatic, (iv) biological. The program Rstudio three types machine learning applied: MaxENT, Support Vector Machine (SVM), Random (RF). results show RF model has highest accuracy (AUC = 0.91), followed by MaxENT 0.87) SVM 0.84). In fire map elaborated with model, 38.8% region possesses a very low occurrence, 21.8% represents high-risk level zones. research will allow decision-makers improve forest Amazon prioritize prospective strategies such as installation water reservoirs areas zone. addition, it support awareness-raising actions among inhabitants at greatest so be prepared mitigate control generate solutions event occurring under different scenarios.

Язык: Английский

Процитировано

1
Research on the Simulation Model of Dynamic Shape for Forest Fire Burned Area Based on Grid Paths from Satellite Remote Sensing Images DOI Creative Commons
Xinsheng Ling, Gui Zhang,

Ying Zheng

и другие.

Remote Sensing, Год журнала: 2025, Номер 17(1), С. 140 - 140

Опубликована: Янв. 3, 2025

The formation of forest fire burned area, influenced by a variety factors such as meteorology, topography, vegetation, and human intervention, is dynamic process line burning that develops from the point ignition to boundary area. Accurately simulating predicting this can provide scientific basis for control suppression decisions. In study, five typical fires located in different regions China were used study object. straight path distances grid each on Sentinel-2 imageries target variables. We obtained values 11 independent variables pathway, including wind speed component, Temperature, Relative Humidity, Elevation, Slope, Aspect, Degree Relief, Normalized Difference Vegetation Index, Type, Fire Duration, Gross Domestic Product reflecting intervention capacity fires. value variable its corresponding constituted sample. Four machine learning models, Random Forest (RF), Gradient Boosting Decision Trees (GBDT), Support Vector Machine (SVM), Multilayer Perceptron (MLP), trained using 80% effective samples four fires, 20% verify above models. hyper-parameters model optimized search method. After analyzing validation results models which showed temperature non-significant variable, training was repeated after excluding temperature. show RF optimal with 49.55 m root mean square error (RMSE), 29.19 absolute (MAE) 0.9823 coefficient determination (R2). This construct shape areas lengths all line. dynamically capture development scenes.

Язык: Английский

Процитировано

0
Post-Fire Forest Ecological Quality Recovery Driven by Topographic Variation in Complex Plateau Regions: A 2006–2020 Landsat RSEI Time-Series Analysis DOI Open Access

Jiayue Gao,

Yue Chen, Bo Xu

и другие.

Forests, Год журнала: 2025, Номер 16(3), С. 502 - 502

Опубликована: Март 12, 2025

Forest fires are an important disturbance that affects ecosystem stability and pose a serious threat to the ecosystem. However, recovery process of forest ecological quality (EQ) after fire in plateau mountain areas is not well understood. This study utilizes Google Earth Engine (GEE) Landsat data generate difference indices, including NDVI, NBR, EVI, NDMI, NDWI, SAVI, BSI. After segmentation using Simple Non-Iterative Clustering (SNIC) method, were input into random (RF) model accurately extract burned area. A 2005–2020 remote sensing index (RSEI) time series was constructed, post-fire EQ evaluated through Theil–Sen slope estimation, Mann–Kendall (MK) trend test, analysis, integration with topographic information systems. The shows (1) from 2006 2020, improved year by year, average annual increase rate 0.014/a. exhibited overall “decline initially-fluctuating increase-stabilization”, indicating RSEI can be used evaluate complex mountainous regions. (2) Between forests significant increasing spatially, 84.32% showing notable growth RSEI, while 1.80% regions experienced declining trend. (3) coefficient variation (CV) area 0.16 during period 2006–2020, good recovery. (4) Fire has impact on low-altitude areas, steep slopes, sun-facing slow. offers scientific evidence for monitoring assessing also inform restoration management efforts similar areas.

Язык: Английский

Процитировано

0
Orman Yangın Alanlarında Arazi ve Toprak Örtüsündeki Değişimlerin İzlenmesi DOI
Betül Kırımlıoğlu, Esra Tunç Görmüş

Turkish Journal of Remote Sensing and GIS, Год журнала: 2025, Номер 6(1), С. 96 - 118

Опубликована: Март 26, 2025

Ormanlar, dünyamızın en büyük doğal zenginliklerinden biri olup ekosistemin dengesinde önemli bir rol oynamaktadır. Uzaktan algılama teknolojilerinin gelişmesiyle orman yangının yol açtığı hasar ve buna bağlı olarak meydana gelen zamansal değişimler daha hızlı etkili şekilde izlenebilmektedir. Bu çalışmada 2019 Haziran ile 2020 Mayıs tarihleri arasında gerçekleşen Avustralya yangınından çok etkilenen Sidney şehrinden yanan alan seçilmiştir. Yangın öncesi sonrası Landsat 8 uydu görüntüleri kullanılarak kontrollü sınıflandırma işlemi tespiti yapılmış farklı bantların yangın hasarını belirlemedeki etkinliği eşik yöntemleri (Otsu, Tsai, Kapur, Kittler) incelenmiştir. Bunun yanı sıra Yanmış Alan İndeksi (BAI), arazi örtüsünde (NDVI, NDMI, NDBI, EVI, LAI, SAVI) toprak (BSI, LST, SMI, SSI) topraktaki mineraller (CM, IOR, FM, Fe+3, Fe+2) üzerindeki etkisi de detaylı Sonuç olarak, ciddi zarar verdiğini, bitkilerin yok olmasıyla çıplak örtüsünün ortaya çıktığını yüzey sıcaklığının arttığı gözlenmiştir. durum, nem oranının tuzluluğunun azalmasına sebep olmuştur. Bitkilerin yeniden canlanmasında etken olan demir seviyesinde yangından sonra artış yaşanmıştır. çalışma, etkilerini doğanın kendini yenileme sürecinin uzaktan başarılı izlenebileceğini göstermektedir.

Процитировано

0
Forest Fire Risk Prediction in South Korea Using Google Earth Engine: Comparison of Machine Learning Models DOI Creative Commons
Jang-Soo Choi, Young Jo Yun,

Heemun Chae

и другие.

Land, Год журнала: 2025, Номер 14(6), С. 1155 - 1155

Опубликована: Май 27, 2025

Forest fires pose significant threats to ecosystems, economies, and human lives. However, existing forest fire risk assessments are over-reliant on field data expert-derived indices. Here, we assessed the nationwide in South Korea using a dataset of 2289 4578 non-fire events between 2020 2023. Twelve remote sensing-based environmental variables were exclusively derived from Google Earth Engine, including climate, vegetation, topographic, socio-environmental factors. After removing snow equivalent variable owing high collinearity, trained three machine learning models: random forest, XGBoost, artificial neural network, evaluated their ability predict risks. XGBoost showed best performance (F1 = 0.511; AUC 0.76), followed by 0.496) network 0.468). DEM, NDVI, population density consistently ranked as most influential predictors. Spatial prediction maps each model revealed consistent high-risk areas with some local differences. These findings demonstrate potential integrating cloud-based sensing for large-scale, high-resolution modeling have implications early warning systems effective management vulnerable regions. Future predictions can be improved incorporating seasonal, real-time meteorological, activity data.

Язык: Английский

Процитировано

0
Segmentation of Any Fire Event (SAFE): A Rapid and High-Precision Approach for Burned Area Extraction Using Sentinel-2 Imagery DOI Creative Commons
Shuaijun Liu, Yong Xue, Hui Chen

и другие.

Remote Sensing, Год журнала: 2024, Номер 17(1), С. 54 - 54

Опубликована: Дек. 27, 2024

The timely and accurate monitoring of wildfires other sudden natural disasters is crucial for safeguarding the safety residents their property. Satellite imagery wildfire offers a unique opportunity to obtain near-real-time disaster information through rapid, large-scale remote sensing mapping. However, existing methods are constrained by temporal spatial limitations imagery, preventing comprehensive fulfillment need high resolution in early warning. To address this gap, we propose high-precision extraction method without training—SAFE. SAFE combines generalization capabilities Segmentation Anything Model (SAM) effectiveness hotspot product data such as MODIS VIIRS. employs two-step localization strategy incrementally identify burned areas pixels post-wildfire thereby reducing computational load providing high-resolution impact areas. area generated can subsequently be used train lightweight regional models, establishing detection models applicable various regions, ultimately undetected We validated four test regions representing two typical scenarios—grassland forest. results showed that SAFE’s F1-score was, on average, 9.37% higher than alternative methods. Additionally, application scenarios demonstrated its potential capability detect fine distribution impacts global scale.

Язык: Английский

Процитировано

0