The mutual relationship among daily averaged PM₁₀, PM_2.5, and NO₂ concentrations in two megacities (Seoul and Busan) connected by the busiest highway in Korea was investigated using an artificial neural network model (ANN)-sigmoid function, for a novel coronavirus (COVID-19) pandemic period from 1 January to 31 December 2020. Daily and weekly mean concentrations of NO₂ in 2020 under neither locked down cities, nor limitation of the activities of vehicles and people by the Korean Government have decreased by about 15%, and 12% in Seoul, and Busan cities, than the ones in 2019, respectively. PM ₁₀ (PM_2.5) concentration has also decreased by 15% (10%), and 12% (10%) in Seoul, and Busan, with a similar decline of NO₂, causing an improvement in air quality in each city. Multilayer perception (MLP), which has a back-propagation training algorithm for a feed-forward artificial neural network technique with a sigmoid activation function was adopted to predict daily averaged PM₁₀, PM_2.5, and NO₂ concentrations in two cities with their interplay. Root mean square error (RMSE) with the coefficient of determination (R²) evaluates the performance of the model between the predicted and measured values of daily mean PM₁₀, PM_2.5, and NO_2, in Seoul were 2.251 with 0.882 (1.909 with 0.896; 1.913 with 0.892), 0.717 with 0.925 (0.955 with 0.930; 0.955 with 0.922), and 3.502 with 0.729 (2.808 with 0.746; 3.481 with 0.734), in 2 (5; 7) nodes in a single hidden layer. Similarly, they in Busan were 2.155 with 0.853 (1.519 with 0.896; 1.649 with 0.869), 0.692 with 0.914 (0.891 with 0.910; 1.211 with 0.883), and 2.747 with 0.667 (2.277 with 0.669; 2.137 with 0.689), respectively. The closeness of the predicted values to the observed ones shows a very high Pearson r correlation coefficient of over 0.932, except for 0.818 of NO₂ in Busan. Modeling performance using IBM SPSS-v27 software on daily averaged PM₁₀, PM_2.5, and NO₂ concentrations in each city were compared by scatter plots and their daily distributions between predicted and observed values.

Multiple statistical prediction modeling of PM₁₀, PM_2.5 and PM₁ at Gangneung city, Korea, was performed in association with local meteorological parameters (air temperature, wind speed and relative humidity) and PM₁₀ and PM_2.5 concentrations of an upwind site in Beijing, China, in the transport route of Chinese yellow dusts which originated from the Gobi Desert and passed through Beijing to the city from 18 March to 27 March 2015. Before and after the dust periods, the PM₁₀, PM_2.5 and PM₁ concentrations showed as being very high at 09:00 LST (the morning rush hour) by the increasing emitted pollutants from vehicles and flying dust from the road and their maxima occurred at 20:00 to 22:00 LST (the evening departure time) from the additional pollutants from resident heating boilers. During the dust period, these peak trends were not found due to the persistent accumulation of dust in the city from the Gobi Desert through Beijing, China, as shown in real-time COMS-AI satellite images. Multiple correlation coefficients among PM₁₀, PM_2.5 and PM₁ at Gangneung were in the range of 0.916 to 0.998. Multiple statistical models were devised to predict each PM concentration, and the significant levels through multi-regression analyses were p < 0.001, showing all the coefficients to be significant. The observed and calculated PM concentrations were compared, and new linear regression models were sequentially suggested to reproduce the original observed PM values with improved correlation coefficients, to some extent.

Multivariate regression models for real-time coastal air quality forecasting were suggested from 18 to 27 March 2015, with a total of 15 kinds of hourly input data (three-hours-earlier data of PM and gas with meteorological parameters from Kangnung (Korea), associated with two-days-earlier data of PM and gas from Beijing (China)). Multiple correlation coefficients between the predicted and measured PM₁₀, PM_2.5, NO₂, SO₂, CO and O₃ concentrations were 0.957, 0.906, 0.886, 0.795, 0.864 and 0.932 before the yellow sand event at Kangnung, 0.936, 0.982, 0.866, 0.917, 0.887 and 0.916 during the event and 0.919, 0.945, 0.902, 0.857, 0.887 and 0.892 after the event. As the significance levels (p) from multi-regression analyses were less than 0.001, all correlation coefficients were very significant. Partial correlation coefficients presenting the contribution of 15 input variables to 6 output variables using the models were presented for the three periods in detail. Scatter plots and their hourly distributions between the predicted and measured values showed the quite good accuracy of the modeling performance for the current time forecasting of six output values and their high applicability.

The intensity change of Typhoon Songda (TY-0418) in the vicinity of the semi-enclosed Sea of Japan (SJ) was numerically investigated using 3D-WRF and UM-KMA models and GOES-IR satellite images on 4 to 8 September 2004. After the typhoon originated in the Western Pacific Ocean in August, it moved to the East China Sea. Following the north-eastward Kuroshio Warm Current, it developed with horizontal and vertical asymmetrical wind and moisture patterns until 5 September. On 7 September, closing to the Kyushu Island, it was divided into three wind fields near the surface due to the increased friction from the surrounding lands and shallower sea depth close to the land, but it still maintained its circular shape over 1 km in height. As it passed by the Korea Strait and entered the SJ, it became a smaller, deformed typhoon due to the SJ’s surrounding mountains, located between the East Korea and Tsushima Warm Currents inside the SJ. Its center matched a high equivalent potential temperature area, releasing significant latent heat through the condensation of water particles over warm currents. The latent heat converted to kinetic energy could be supplied into the typhoon circulation, causing its development. Moist flux and streamline at 1.5 km in height clearly showed the moisture transportation via the mutual interaction of the cyclonic circulation of the typhoon and the anti-cyclonic circulation of the North Pacific High Pressure from the typhoon’s tail toward both the center of the SJ and the Russian Sakhalin Island in the north of Japan, directly causing large clouds in its right quadrant. Simultaneously, the central pressure decrease with time could converge both transported moist air by the typhoon itself and water particles evaporated from the sea, causing them to rise and resulting in the formation of large clouds and the rapid development of the typhoon circulation. The strong downslope winds from the surrounding mountains of the SJ to its center also produced a cyclonic vortex due to the Coriolis force to the right, enhancing the typhoon’s circulation.

The effects of different activation functions (sigmoid and hyperbolic tangent), and node numbers in a hidden layer of artificial neural network (ANN) models on urban air quality forecasting were investigated in a coastal city of Korea. The ANN models of multilayer perceptron (MLP) with a back-propagation training algorithm for error calculation in cases of 13, 15, and 17 nodes in each hidden layer were performed using 15 input independent variables (PM, gas, and meteorological data of Gangneung city (Republic of Korea)), affected by PM and gas of an upwind Beijing city (China). Root mean square error (RMSE) and the coefficient of determination (R²; Pearson R) were evaluated to assess the two models’ forecasting abilities between the predicted and measured values. The values of R by ANN-sig (ANN-tanh) with 13, 15, and 17 hidden neuron numbers were 0.930 (0.950), 0.920 (0.947), 0.926(0.953) on PM₁₀, 0.953 (0.956), 0.927 (0.938), 0.949 (0.960) on PM_2.5, and 0.880 (0.959), 0.917 (0.886), and 0.882 (0.939) on NO₂. Regardless of node numbers and activation functions, the prediction abilities of the two models were excellent, showing the highest values of R in the ANN-tanh model with more neuron numbers in the hidden layer. Unlike previous studies’ insistence that smaller nodes (larger) in the hidden layer produce the overfit (underfit) result in the ANN model, the present study proves that more nodes in its hidden layer than the input layer can yield the best prediction than any other, as shown in their temporal distributions and scatter plots of predicted and measured data. Future-time air quality forecasting at Gangneung city can be calculated sequentially using its current time data and previous time data from Beijing city, using the suggested empirical formulas.