Research on Machine Learning-based Nowcasting Method for Low Visibility Weather in Urumqi Airport

Li, Jing; Wang, Fan; Xue, Kang; Hao, Yusong

doi:10.1109/ctisc54888.2022.9849817

Cited by 3 publications

(3 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MSE is the L2-errors between the ground truth and dehazed image, and PSNR indicates the ratio between the maximal response and MSE. Note that PSNR measures how the strength of the signal contributes to the noise so that it can assess the ability of the from 𝐼 Calculate 𝑃 according to (24) 27) and ( 28) Thresholding 𝑡 ̃ * Calculate 𝐽 * * according to (16) and 𝑡 ̃ * noise control. However, according to the definition, MSE and PSNR are not sensitive to artifacts, structural similarity, and color deviation [50]; therefore, we use SSIM and CIEDE2000 to evaluate the structural similarity and color deviations, respectively.…”

Section: A Goals and Setup Of Experimentsmentioning

confidence: 99%

“…As a significant trend, more and more researchers evaluated the dehazing algorithm using non-reference benchmarks [17]- [19], which assessed the naturalness, contrast, and brightness of images. As the preprocessing algorithms, they played important roles in various fields, such as semantic segmentation [20], object recognition [21] [22], security sensor applications [23], airport control [24], and auto-pilot [20] [25]. However, the performance of dehazing algorithms was quite limited while dealing with various haze conditions.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Enhance Low Visibility Image Using Haze-Removal Framework

Liu

2023

IEEE Access

View full text Add to dashboard Cite

We proposed a novel image enhancement framework to raise the visibility of the image's content. Our primary concern is eliminating haze-like effects and simultaneously increasing images' brightness. Dehazing and luminance enhancement algorithms are considered standard techniques to overcome these issues. However, natural environments usually involve several unfavorable conditions simultaneously, such as insufficient illumination, blur caused by the haze, and color cast resulting from the sun or scattering; this makes dehazing algorithms challenging to overcome environmental issues. Besides, dehazing algorithms sometimes result in artifacts. The proposed framework solves these issues simultaneously by implementing a double-side-enhancement in contrast and brightness based on a new dehazing algorithm. We compare the new dehazing algorithm with others using full-reference benchmarks to ensure performance stability. Afterward, to show the advantage of using the new dehazing algorithm, we evaluate the compatibility between the proposed framework and all dehazing algorithms using non-reference benchmarks. At last, we pair dehazing and luminance enhancement algorithms and compare the combinations with the proposed framework. Eventually, experimental results prove that the new dehazing algorithm outperforms others and is better compatible with the proposed framework. Meanwhile, the proposed framework is superior in contrast and brightness enhancements and outperforms whether single dehazing algorithm or the combinations.

show abstract

Section: A Goals and Setup Of Experimentsmentioning

confidence: 99%

mentioning

confidence: 99%

Enhance Low Visibility Image Using Haze-Removal Framework

Liu

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…While these studies have consistently utilized traditional models, there is a notorious gap in the state-ofthe-art regarding wind speed and direction nowcasting using ML-based techniques. These methods have primarily been applied to study extreme weather events (Castro et al, 2022;Chkeir et al, 2023) and low visibility conditions (Bari et al, 2023;Bartok et al, 1684;Li et al, 2022) with some investigation covering extreme wind speeds (Chkeir et al, 2023). Nevertheless, there appears to be an absence of studies focusing objectively on wind speed and direction nowcasting using ML, and none have been centered on GCTS, indicating a specific research gap to be addressed.…”

Section: Introductionmentioning

confidence: 99%

Low Tropospheric Wind Forecasts in Aviation: The Potential of Deep Learning for Terminal Aerodrome Forecast Bulletins

Alves,

Mendonça,

Mostafa

et al. 2024

Pure Appl. Geophys.

View full text Add to dashboard Cite

In aviation, accurate wind prediction is crucial, especially during takeoff and landing at complex sites like Gran Canaria Airport. This study evaluated five Deep Learning models: Long Short-Term Memory (LSTM), Vanilla Recurrent Neural Network (vRNN), One-Dimensional Convolutional Neural Network (1dCNN), Convolutional Neural Network Long Short-Term Memory (CNN-LSTM), and Gated Recurrent Unit (GRU) for forecasting wind speed and direction. The LSTM model demonstrated the highest precision, particularly for extended forecasting periods, achieving a mean absolute error (MAE) of 1.23 m/s and a circular MAE (cMAE) of 15.80° for wind speed and direction, respectively, aligning with World Meteorological Organization standards for Terminal Aerodrome Forecasts (TAF). While the GRU and CNN-LSTM also showed promising results, and the 1dCNN excelled in wind direction forecasting over shorter intervals, the vRNN lagged in performance. Additionally, the autoregressive integrated moving average model underperformed relative to the DL models, underscoring the potential of DL, particularly LSTM, in enhancing TAF accuracy at airports with intricate wind patterns. This study not only confirms the superiority of DL over traditional methods but also highlights the promise of integrating artificial intelligence into TAF automation.

show abstract

Predicting Airport Visibility using Seq2Seq Models

Wang,

Wang

2023

2023 11th International Conference on Information Systems and Computing Technology (ISCTech)

View full text Add to dashboard Cite

Research on Machine Learning-based Nowcasting Method for Low Visibility Weather in Urumqi Airport

Cited by 3 publications

References 4 publications

Enhance Low Visibility Image Using Haze-Removal Framework

Enhance Low Visibility Image Using Haze-Removal Framework

Low Tropospheric Wind Forecasts in Aviation: The Potential of Deep Learning for Terminal Aerodrome Forecast Bulletins

Predicting Airport Visibility using Seq2Seq Models

Contact Info

Product

Resources

About