An Atmospheric Visibility Grading Method Based on Ensemble Learning and Stochastic Weight Average

In order to adequately characterize the visual characteristics of atmospheric visibility and overcome the disadvantages of the traditional atmospheric visibility measurement method with significant dependence on preset reference objects, high cost, and complicated steps, this paper proposed an ensemble learning method for atmospheric visibility grading based on deep neural network and stochastic weight averaging. An experiment was conducted using the scene of an expressway, and three visibility levels were set, i.e., Level 1, Level 2, and Level 3. Firstly, the EfficientNet was transferred to extract the abstract features of the images. Then, training and grading were performed on the feature sets through the SoftMax regression model. Subsequently, the feature sets were ensembled using the method of stochastic weight averaging to obtain the atmospheric visibility grading model. The obtained datasets were input into the grading model and tested. The grading model classified the results into three categories, with the grading accuracy being 95.00%, 89.45%, and 90.91%, respectively, and the average accuracy of 91.79%. The results obtained by the proposed method were compared with those obtained by the existing methods, and the proposed method showed better performance than those of other methods. This method can be used to classify the atmospheric visibility of traffic and reduce the incidence of traffic accidents caused by atmospheric visibility.

A Method for Mining Granger Causality Relationship on Atmospheric Visibility

Atmospheric visibility is an indicator of atmospheric transparency and its range directly reflects the quality of the atmospheric environment. With the acceleration of industrialization and urbanization, the natural environment has suffered some damages. In recent decades, the level of atmospheric visibility shows an overall downward trend. A decrease in atmospheric visibility will lead to a higher frequency of haze, which will seriously affect people's normal life, and also have a significant negative economic impact. The causal relationship mining of atmospheric visibility can reveal the potential relation between visibility and other influencing factors, which is very important in environmental management, air pollution control and haze control. However, causality mining based on statistical methods and traditional machine learning techniques usually achieve qualitative results that are hard to measure the degree of causality accurately. This article proposed the seq2seq-LSTM Granger causality analysis method for mining the causality relationship between atmospheric visibility and its influencing factors. In the experimental part, by comparing with methods such as linear regression, random forest, gradient boosting decision tree, light gradient boosting machine, and extreme gradient boosting, it turns out that the visibility prediction accuracy based on the seq2seq-LSTM model is about 10% higher than traditional machine learning methods. Therefore, the causal relationship mining based on this method can deeply reveal the implicit relationship between them and provide theoretical support for air pollution control.

Predicting the Number of Days With Visibility in a Specific Range in Warsaw (Poland) Based on Meteorological and Air Quality Data

Atmospheric visibility is an important parameter of the environment which is dependent on meteorological and air quality conditions. Forecasting of visibility is a complex task due to the multitude of parameters and nonlinear relations between these parameters. In this study, meteorological, air quality, and atmospheric visibility data were analyzed together to demonstrate the capabilities of the multidimensional logistic regression model for visibility prediction. This approach allowed determining independent variables and their significance to the value of the atmospheric visibility in four ranges (i.e., 0–10, 10–20, 20–30, and ≥ 30 km). We proved that the Iman–Conover (IC) method can be used to simulate a time series of meteorological and air quality parameters. The visibility in Warsaw (Poland) is dependent mainly on air temperature and humidity, precipitation, and ambient concentration of PM10. Three logistic models of visibility allowed us to determine precisely the number of days in a month with visibility in a specific range. The sensitivity of the models was between 75.53 and 90.21%, and the specificity 78.51 and 96.65%. The comparison of the theoretical (modeled) with empirical (measured) distribution with the Kolmogorov–Smirnov test yielded p-values always above 0.27 and, in half of the cases, above 0.52.

Multifractal analysis of extinction coefficient and its consequences in characterizing atmospheric visibility

<p>Extinction coefficient (σ<sub>e</sub>) is a measure of light attenuation in the atmosphere, due to absorption and scattering properties of constituent gases and aerosols. In meteorological context, σ<sub>e </sub>is used to understand transparency of the atmosphere, by estimating visibility or meteorological observable range (MOR). An accurate representation of visibility is required for safe functioning of various domains such as transport sectors, free optic communication, etc., and for understanding regional variations in air quality and climate. As the measurement of visibility is subjective and dependent on the instrument and range of measurement, here we attempt to characterize the same using extinction coefficient. σ<sub>e </sub>was investigated under the framework of universal multifractals (UM), which is widely used to analyze and characterize geophysical fields that exhibit extreme variability over measurement scales.</p><p>For this study, σ<sub>e </sub>was extracted from forward scattering visibility data by disdrometer (Campbell Scientific PWS100) located in the Paris area (France), operated by Hydrology, Meteorology, and Complexity laboratory of École des Ponts ParisTech (HM&Co, ENPC). As governing nonlinear equations of the atmosphere such as Navier-Stokes possess scale invariance, it was assumed here that the behavior of light attenuating particles should inherit similar scaling properties and hence be treated as multifractal fields. σ<sub>e </sub>extracted from MOR measured at Paris-Charles de Gaulle airport was also subjected to multifractal analysis during the same time period for comparison. With direct analysis and simulations, it was found that σ<sub>e </sub>exhibits multifratcal properties but are influenced by upper limit of visibility range in the instrument used for measurement. From the study, we suggest usage of extinction coefficient (σ<sub>e</sub>) for characterizing atmospheric visibility as the former is a more physically relevant quantity which is objectively measured by instruments and directly related to particles in the atmosphere; while emphasizing the need to consider biases from instrumental range.</p>

Error Analysis and Improved Design of Target Board Reflection for Visibility Measurement by the Image Method

The reflection of colors and surfaces of common targets lead to errors in the measurement of visibility by the image method. This study aims to investigate the problem of inaccurate visibility detection. Through analysis of the error of visibility measurement caused by the reflection of the blackboard surface of an artificial target, the design method of improving the structure of the target board is proposed, so as to improve the accuracy of atmospheric visibility measurement by the image method. The experimental results show that the new target board designed by this method can greatly improve the measurement accuracy of the intrinsic apparent brightness ratio, which can increase 18.4% in the fairing environment and closer to −1 in the side light environment. Therefore, when the side light is selected for the image method visibility measurement, more accurate visibility results can be obtained.