Capítulo 15: Identificación de suelo desnudo utilizando Random Forest en la zona norte y centro del departamento de Sucre - Colombia DOI
Luis Alberto Ávila Lorduy

Опубликована: Дек. 31, 2024

La degradación del suelo en la región norte y centro departamento de Sucre, Colombia, se ha intensificado debido a deforestación al uso inadecuado suelo, afectando gravemente su sostenibilidad ambiental. Este estudio tuvo como objetivo identificar áreas desnudo mediante imágenes Landsat 8 clasificación supervisada usando el modelo Random Forest. El análisis abarcó 5,123.58 km² empleó OLI/TIRS año 2020. algoritmo Forest combinó con técnica validación cruzada RepeatedStratifiedKFold, 10 pliegues 3 repeticiones, utilizando 2,571 puntos 912 otras coberturas. alcanzó una precisión promedio 99%, exactitud 0.985, valores medios recall F1-score 0.99, un AUC 1.00 coeficiente Kappa 0.96. Los resultados subrayaron relevancia las bandas SWIR2, Red Blue para identificación desnudo, lo cual respaldó investigaciones anteriores. En conclusión, esta metodología demostró ser eficaz apoyar estrategias restauración manejo sostenible zonas afectadas por erosión Sucre.

Near Real-Time Flood Monitoring Using Multi-Sensor Optical Imagery and Machine Learning by GEE: An Automatic Feature-Based Multi-Class Classification Approach DOI Creative Commons
Hadi Farhadi, Hamid Ebadi, Abbas Kiani

и другие.

Remote Sensing, Год журнала: 2024, Номер 16(23), С. 4454 - 4454

Опубликована: Ноя. 27, 2024

Flooding is one of the most severe natural hazards, causing widespread environmental, economic, and social disruption. If not managed properly, it can lead to human losses, property damage, destruction livelihoods. The ability rapidly assess such damages crucial for emergency management. Near Real-Time (NRT) spatial information on flood-affected areas, obtained via remote sensing, essential disaster response, relief, urban industrial reconstruction, insurance services, damage assessment. Numerous flood mapping methods have been proposed, each with distinct strengths limitations. Among widely used are machine learning algorithms spectral indices, though these often face challenges, particularly in threshold selection indices sampling process supervised classification. This study aims develop an NRT approach using classification based features. method automatically generates training samples through masks derived from indices. More specifically, this uses FWEI, NDVI, NDBI, BSI extract water/flood, vegetation, built-up soil, respectively. Otsu thresholding technique applied create masks. Land cover then performed Random Forest algorithm generated samples. final map by subtracting pre-flood water class post-flood image. proposed implemented optical satellite images Sentinel-2, Landsat-8, Landsat-9. method’s accuracy rigorously evaluated compared those techniques. suggested achieves highest overall (OA) 90.57% a Kappa Coefficient (KC) 0.89, surpassing SVM (OA: 90.04%, KC: 0.88), Decision Trees 88.64%, 0.87), like AWEI 84.12%, 0.82), FWEI 88.23%, 0.86), NDWI 85.78%, 0.84), MNDWI 87.67%, 0.85). These results underscore superior effectiveness detection monitoring multi-sensor imagery.

Язык: Английский

Процитировано

8
Application of Remote Sensing for Identifying Soil Erosion Processes on a Regional Scale: An Innovative Approach to Enhance the Erosion Potential Model DOI Creative Commons
Siniša Polovina, Boris Radić, Ratko Ristić

и другие.

Remote Sensing, Год журнала: 2024, Номер 16(13), С. 2390 - 2390

Опубликована: Июнь 28, 2024

Soil erosion represents a complex ecological issue that is present on global level, with negative consequences for environmental quality, the conservation and availability of natural resources, population safety, material security, both in rural urban areas. To mitigate harmful effects soil erosion, map can be created. Broadly applied Balkan Peninsula region (Serbia, Bosnia Herzegovina, Croatia, Slovenia, Montenegro, North Macedonia, Romania, Bulgaria, Greece), Erosion Potential Method (EPM) an empirical model widely process creating maps. In this study, innovation identification mapping processes was made, coefficient types extent slumps (φ), representing one most sensitive parameters EPM. The (φ) consisted applying remote sensing methods satellite images from Landsat mission. research area which were obtained thematic maps (coefficient φ) created Federation Herzegovina Brčko District (situated Herzegovina). Google Earth Engine (GEE) platform employed to retrieve 7 Enhanced Thematic Mapper plus (ETM+) 8 Operational Land Imager Thermal Infrared Sensor (OLI/TIRS) imagery over period ten years (from 1 January 2010 31 December 2020). performed based Bare Index (BSI) by equation fractional bare cover. spatial–temporal distribution cover enabled definition values field. An accuracy assessment conducted 190 reference samples field using confusion matrix, overall (OA), user (UA), producer (PA), Kappa statistic. Using OA 85.79% obtained, while UA ranged 33% 100%, PA 50% 100%. Applying statistic, 0.82 indicating high level accuracy. time series multispectral each month crucial element monitoring occurrence various (surface, mixed, deep) Additionally, it contributes significantly decision-making, strategies, plans domain control work, development identifying erosion-prone areas, defense against torrential floods, creation at local, regional, national levels.

Язык: Английский

Процитировано

5
The ensemble learning combined with the pruning model reveals the spectral response mechanism of tidal flat mapping in China DOI Creative Commons

Jiapeng Dong,

Kai Jia, Chongyang Wang

и другие.

Ecological Informatics, Год журнала: 2025, Номер unknown, С. 103104 - 103104

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0
Validation of sentinel 2 based machine learning models for Czech National Forest Inventory DOI Creative Commons
Richard Kovárník, Jitka Janová

Ecological Informatics, Год журнала: 2025, Номер unknown, С. 103133 - 103133

Опубликована: Апрель 1, 2025

Язык: Английский

Процитировано

0
A detection method for multi-type earth's surface anomalies based on multi-dimensional feature space DOI Creative Commons
Haishuo Wei, Kun Jia, Qiao Wang

и другие.

International Journal of Digital Earth, Год журнала: 2024, Номер 17(1)

Опубликована: Авг. 30, 2024

Язык: Английский

Процитировано

1
Multicriteria evaluation and remote sensing approach to identifying degraded soil areas in northwest Peru DOI Creative Commons
Marielita Arce-Inga,

Nilton Atalaya-Marin,

Elgar Barboza

и другие.

Geocarto International, Год журнала: 2024, Номер 40(1)

Опубликована: Дек. 23, 2024

Soil is a vital nonrenewable resource characterized by rapid degradation and slow regeneration processes. In this study, soil in Jaén San Ignacio was assessed via multicriteria evaluation approach combined with remote sensing (RS) data. Nine factors were analyzed classified three categories: environmental, topographic, edaphological factors. The results revealed that the slope (59.07%) main influencing factor, followed land use cover (LULC) (56.36%). map 83.48% of area exhibited moderate degradation, 14.49% low 1.56% high degradation. districts Pomahuaca José de Lourdes demonstrated largest areas accounting for 13.71% 22.54%, respectively. Bellavista Huarango very 0.27% 0.08%, (AHP) method RS data employed to assess highlighting need sustainable restoration conservation strategies.

Язык: Английский

Процитировано

0
Capítulo 15: Identificación de suelo desnudo utilizando Random Forest en la zona norte y centro del departamento de Sucre - Colombia DOI
Luis Alberto Ávila Lorduy

Опубликована: Дек. 31, 2024

La degradación del suelo en la región norte y centro departamento de Sucre, Colombia, se ha intensificado debido a deforestación al uso inadecuado suelo, afectando gravemente su sostenibilidad ambiental. Este estudio tuvo como objetivo identificar áreas desnudo mediante imágenes Landsat 8 clasificación supervisada usando el modelo Random Forest. El análisis abarcó 5,123.58 km² empleó OLI/TIRS año 2020. algoritmo Forest combinó con técnica validación cruzada RepeatedStratifiedKFold, 10 pliegues 3 repeticiones, utilizando 2,571 puntos 912 otras coberturas. alcanzó una precisión promedio 99%, exactitud 0.985, valores medios recall F1-score 0.99, un AUC 1.00 coeficiente Kappa 0.96. Los resultados subrayaron relevancia las bandas SWIR2, Red Blue para identificación desnudo, lo cual respaldó investigaciones anteriores. En conclusión, esta metodología demostró ser eficaz apoyar estrategias restauración manejo sostenible zonas afectadas por erosión Sucre.

Процитировано

0