Sentinel-1 Polarization Comparison for Flood Segmentation Using Deep Learning

Опубликована: Фев. 22, 2023

Flood is one of the most damaging natural hazards, and timely detection it very important to save human lives assess level damage. Floods usually occur in certain weather conditions such as excessive rainfall, which makes presence clouds sky region likely. For this reason, radar-based sensors are suitable choice for real-time flood mapping. In present study, ETCI 2021 event competition dataset, organized by NASA Advanced Concepts Implementation Team collaboration with IEEE GRSS Geoscience Informatics Technical Committee, has been used. Moreover, we have utilized U-Net architecture a segmentation model map flooded regions. This study aims identify areas from radar images area two different polarizations. By examining comparing obtained results, was observed that network designed VV polarization made better predictions Intersection Over Union (IOU) score improved 64.46 67.35 compared VH polarization.

Язык: Английский

---

A QUICK SEASONAL DETECTION AND ASSESSMENT OF INTERNATIONAL SHADEGAN WETLAND WATER BODY EXTENT USING GOOGLE EARTH ENGINE CLOUD PLATFORM

ISPRS annals of the photogrammetry, remote sensing and spatial information sciences, Год журнала: 2023, Номер X-4/W1-2022, С. 699 - 706

Опубликована: Янв. 14, 2023

Abstract. Understanding the variation of Water Extent (WE) can provide insights into Wetland conservation and management. In this study, and-inter inner-annual variations WE were analyzed during 2019–2021 to understand spatiotemporal changes International Shadegan Wetland, Iran. We utilized a thresholding process on Modified Normalized Difference Index (MNDWI) extract quickly accurately using Google Earth Engine (GEE) platform. The water surface analysis showed that: (1) had downward trend from 2019 2021, with overall average being 1405.23 km2; (2) area reached its peak due supply through Jarahi River upstream reservoirs at end beginning 2020, largest body appeared in Winter 2019, reaching 1953.31 km2. contrast, smallest Autumn 563.56 (3) wetland predictable seasonal characteristics. was largest, an value 1829.1 km2, while it Summer, 1100.3 (4) 1490.5 km2 whereas 2020 2021 decreased by 9% 25%, respectively, 968.6 811.9 Finally, evaluate proposed model, results compared Random Forest (RF) classification results. Accordingly, Histogram Analysis (HA) achieved 94.6% accuracy Kappa coefficient 0.93, but RF method obtained 95.38% 0.94.

Язык: Английский

---

Ultra-short-term PV power prediction based on Informer with multi-head probability sparse self-attentiveness mechanism

Frontiers in Energy Research, Год журнала: 2023, Номер 11

Опубликована: Ноя. 16, 2023

As a clean energy source, solar power plays an important role in reducing the high carbon emissions of China’s electricity system. However, intermittent nature system limits effective use photovoltaic generation. This paper addresses problem low accuracy ultra-short-term prediction distributed PV power, compares various deep learning models, and innovatively selects Informer model with multi-head probability sparse self-attention mechanism for prediction. The results show that CEEMDAN-Informer proposed this has better accuracy, error index is improved by 30.88% on average compared single model; superior to other models LSTM RNN medium series prediction, its significantly than two. study improves proves feasibility superiority Meanwhile, can provide some reference renewable sources, such as wind power.

Язык: Английский

---

Futuristic flood risks assessment, in the Upper Vellar Basin, integrating AHP and bivariate analysis

Advances in Space Research, Год журнала: 2024, Номер 74(11), С. 5395 - 5416

Опубликована: Авг. 15, 2024

Язык: Английский

---

Water Budget Input Linked to Atmospheric Rivers in British Columbia's Nechako River Basin

Hydrological Processes, Год журнала: 2024, Номер 38(10)

Опубликована: Окт. 1, 2024

ABSTRACT This study explores the contribution of atmospheric rivers (ARs) to water budget input Nechako River Basin (NRB) in British Columbia (BC), western Canada. The quantifies fraction precipitation, rainfall, snowfall, and snow equivalent (SWE) associated with ARs at multiple scales tests for trends using Mann–Kendall (MK) test. AR‐related totals 1950–2021 were created by linking AR events variables ERA5‐Land reanalysis product on a daily scale. Associations different phases El Niño‐Southern Oscillation (ENSO) climate pattern contributions NRB are also investigated. Results indicate an increasing fractional rain landfalling last two decades (2000–2019). Moreover, 21% total annual precipitation is ARs, decreasing from west east. October has higher than other months, while March, May June least affected. contribute disproportionately more mid‐ high‐intensity totals, provide up 45% 24% seasonal rainfall respectively. SWE relatively autumn due increased frequency intensity resulting greater snowpack compared winter. influence accumulation during fall (18%) winter (13%) but increase risk natural hazards. MK test scale identified no significant trends. However, snowfall shows NRB, specifically Upper Nechako, Lower Stellako sub‐basins summer. Over period, consistently one‐fifth NRB's budget. provides first quantitative assessment trend analyses reservoir‐regulated watershed north‐central BC, yielding valuable information hydropower production, ecological flows, irrigation, domestic industrial use.

Язык: Английский

---

EFFECT OF TRANSFERRING PRE-TRAINED WEIGHTS ON A SIAMESE CHANGE DETECTION NETWORK

ISPRS annals of the photogrammetry, remote sensing and spatial information sciences, Год журнала: 2023, Номер X-4/W1-2022, С. 19 - 24

Опубликована: Янв. 13, 2023

Abstract. Change Detection (CD) is one of the most crucial applications in remote sensing which identifies meaningful changes from bitemporal images taken same location. Enhancing temporal efficiency and accuracy this task great importance way to achieve through transfer learning. In study, we investigate influence transferring pre-trained weights on performance a Siamese CD network using benchmark dataset. For purpose, an autoencoder with encoder architecture as model trained whole Then, are transferred two modes. first mode, frozen only decoder section models while second mode trains without freezing any part model. Moreover, also set basis for comparisons. The results indicate that relatively lower but offers considerable amount training phase. On other hand, after weight acquires best result improvement 12.43% Intersection over Union (IoU) metric.

Язык: Английский

---

Cloud Detection Method Based on Spatial–Spectral Features and Encoder–Decoder Feature Fusion

IEEE Transactions on Geoscience and Remote Sensing, Год журнала: 2023, Номер 61, С. 1 - 15

Опубликована: Янв. 1, 2023

Cloud obscuration in remote sensing images affects Earth observation tasks by causing blurred and incomplete surface information. Regarding this, cloud detection is crucial the processing of images. However, existing methods present some challenges, such as missed thin areas false caused confusing clouds with highlighted snow ice. To address these problems, this paper, we proposed a network that incorporates spectral feature enhancement spatial-spectral fusion. Based on difference reflectivity ground objects atmosphere, short-wave infrared index (SWIR-Index) designed feature-guided module to incorporate into guide training enhance network’s ability learn differential features snow, ice, clouds. fully utilize band information spatial images, developed fusion extracts at different scales performs inter-spectral bands. Furthermore, encoder-decoder automatically calculates pixel weights using weight extraction block. The ablation study proves our method can improve ability, reduce leakage misdetection, accuracy. Experimental results Sentinel-2A demonstrate superior performance method, reaching 98.65(%) OA WHUS2-CD dataset, 97.50(%) S2-CMC 92.36(%) CloudSEN12 which outperforms other algorithms.

Язык: Английский

---

Enhancing River Flood Prediction in Early Warning Systems Using Fuzzy Logic-Based Learning

International Journal of Engineering and Technology Innovation, Год журнала: 2024, Номер 14(4), С. 434 - 450

Опубликована: Сен. 6, 2024

Previous studies show that the fuzzy-based approach predicts incoming floods better than machine learning (ML). However, with numerous observation points, difficulties in manually determining fuzzy rules and membership values increase. This research proposes a novel logic-based (FLBL) embeds missing data imputations rule optimization strategy to enhance ML performance while still benefiting from theory. The simple moving average handles sensors’ data. logical mapping is used for fuzzification automation generation. join function between Szymkiewicz–Simpson coefficient similarity max applied optimize model. case study uses three rivers traversing districts Semarang City. As result, FLBL achieves 97.87% accuracy predicting flood, outperforming decision tree (96%) neural network (73.07%). work significant as part of preventive flood-related disaster plans.

Язык: Английский

---

Earthquake-induced building damage detection using the fusion of optical and radar data in intelligent systems

Earth Science Informatics, Год журнала: 2024, Номер 18(1)

Опубликована: Дек. 17, 2024

Язык: Английский

---

Simulation of Flood Hazard Risk for Naban Reservoir Safety Management: A Comprehensive Assessment

Acadlore Transactions on Geosciences, Год журнала: 2023, Номер 2(3), С. 132 - 144

Опубликована: Авг. 2, 2023

Safety of reservoir dams remains pivotal for societal stability, underscoring the significance efficient emergency management strategies. This investigation focuses on Naban Reservoir, where BREACH model was employed to simulate potential dam failures. By integrating one-dimensional and two-dimensional modeling approaches, a mathematical representation developed scrutinize flood progression in adjacent region. Correlation coefficients devised ranged from 0.945 0.986, with relative errors -13.72%, -0.23%, -17.41%, -15.44%. Comparisons indicated that observed flow rates align closely simulated rates. Notably, significant land slippages surrounding were not detected, implying an enhanced downstream surge due upstream collapse is unlikely. Nevertheless, breach main could result catastrophic outcomes zones, particularly affecting infrastructure communities along Shangsi Zaimiao Basins. Critical observation such as Siyang Town Shangshi County, Nakan Ningming identified, emphasizing need precautionary measures safeguard human lives, property, stability. research has paved way novel early warning system tailored ensuring timely predictions alerts. Such advancements augment disaster prevention capacity, offering valuable insights mitigating risks small medium-sized reservoirs.

Язык: Английский

---