Journal of Atmospheric and Solar-Terrestrial Physics最新文献

Predicting rainfall is essential for many applications, including agriculture, hydrology, and disaster management. In this work, we undertake a comparison examination of various machine learning models to forecast rainfall based on meteorological data. The target variable in this study is rainfall, and the dataset used includes characteristics like temperature, relative humidity, wind speed, and wind direction. The following seven machine learning models were assessed: Support Vector Regression (SVR), Multivariate adaptive regression splines (MARS), Random Forest Regression, and Deep Neural Network with Historical Data (DWFH), Haar Wavelet Function, Decision Tree and Discrete wavelet Transform (DWT). Data preprocessing, which includes standardisation and lagging to capture temporal dependencies, comes first in the analysis phase. A wavelet transformation is also used to capture complex patterns in the data. Each model is tested on a different test set after being trained on a subset of the dataset. The results are assessed using the Root Mean Squared Error (RMSE) and Mean Squared Error (MSE), focusing on the RMSE and MSE values for better comparison across models. Our findings reveal that the DWFH model achieved an RMSE of 0.0138807 mm and MSE of 0.000193 mm², demonstrating their effectiveness in predicting rainfall. The Random Forest and SVR models also provided competitive results. This study highlights the importance of selecting an appropriate machine learning model for rainfall prediction and the significance of preprocessing techniques in improving model performance. These insights can aid decision-makers in choosing the most suitable model for their specific application, contributing to more accurate rainfall predictions and enhanced decision support systems.

Astronomical sites have to be selected according to many factors whereas the geographic location of the site and the quality of the atmosphere above the site play an important role in the decision process. The following factors were chosen to create layers 1907 northern and 235 southern observatories: CC (cloud coverage), PWV (precipitable water vapor), AOD (aerosol optical depth), VWV (vertical wind velocity), and HWV (horizontal wind velocity). To estimate the astronomical importance of the sites, DEM (digital elevation model) and LAT (latitude of observatory location) layers were also included. In addition to the variations or trends, a complete statistical analysis was carried out for all factors to investigate the potential correlations between the factors. There is a clear difference between the northern and southern hemispheres. The exchange of meteorological seasons between hemispheres is also compliant with factors. The geographical locations of most of the observatories were found to be “not suitable”. There seem to be no apparent long-term variations and/or patterns in all factors.

Sea surface temperature (SST) is a crucial geophysical parameter in assessing heat exchange between the air and sea surface. Changes in SST and its accurate prediction play a pivotal role in explaining the global heat balance, determining atmospheric circulations, and constructing global climate models. This work aims to reveal a model for one-month-ahead forecasting of SST time series data along the Türkiye coasts, encompassing the Mediterranean, Aegean, Marmara, and Black Seas, and their long-term future forecast. A long short-term memory (LSTM) neural network and seasonal autoregressive integrated moving average (SARIMA) models are used for this purpose. The ECMWF ERA5 (0.5^ox0.5°) monthly SST dataset spanning the years 1970–2023 is used for model development. The results obtained from the LSTM and SARIMA models show that there will be an increasing trend in SSTs along these seacoasts until 2050. The SST measurements of 23.4 °C, 20.2 °C, 17.0 °C, and 16.6 °C recorded along the Mediterranean, Aegean, Marmara, and Black Seas in 2023 are expected to rise to 25.1 °C, 21.9 °C, 18.1 °C, and 18.8 °C, respectively, by 2050. These figures indicate an increase of 7.3%, 8.4%, 6.5%, and 13.3% in the SST values across these coastal seas over the next quarter century.

Air pollution significantly impacts human health, causing numerous premature deaths, particularly with the rise in PM_2.5 concentrations. Therefore, comparing different machine learning (ML) models for predicting PM_2.5 concentration is crucial. This research focuses on six ML models: Linear Regression (LR), Regression Tree (RT), Support Vector Machine (SVM), Ensemble Regression (ERT), Gaussian Process Regression (GPR), and Artificial Neural Networks (ANN). Trained on six years of data (July 2015–December 2021) with optimized hyperparameters, the models consider eight meteorological and chemical indicators as PM_2.5 predictors, including temperature, relative humidity, air pressure, O₃, SO₂, NO₂, dew point, and wind speed. Model efficiency is assessed using Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Correlation Coefficient (R), and Coefficient of Determination (R²) values. The models achieve R² and RMSE values as follows: LR (0.72, 13.52), RT (0.8, 12.156), SVM (0.82, 10.28), ERT (0.81, 11.87), GPR (0.94, 7.65), and ANN (0.99, 2.36). These metrics indicate the superior performance of ANN, with its R² value approaching 1 and the lowest RMSE compared to other models. The results highlight the effectiveness of ANN, particularly the model with three hidden layers, in predicting PM_2.5 concentration. Utilizing ML models for this purpose is crucial for understanding and mitigating the impacts on human health and the environment, with ANN emerging as a promising tool for various investigations.

To address the global warming problem, one of the space-based geoengineering solutions suggests the construction of an occluding disc that can work as a solar curtain to mitigate solar irradiation penetration to the earth atmosphere. A widely discussed concept needs the construction of a large-scale sunshade system near the Sun–Earth L₁ equilibrium point in order to control the average global temperature. However, to improve the accuracy of theoretical estimations, more consistent modeling of the Sun-Curtain-Earth system and solar irradiance reduction rate are required. This study revisits the mathematical modeling of the solar irradiance reduction system and considers the fundamentals of shading physics. Simplified mathematical modeling of solar irradiance reduction rate is derived based on the solar flux density. For the climate control, controllability of the reduction rate by using some physical parameters (e.g., flux reflection rate and angle of the curtain) is discussed. Based on the results of this model, the technical challenges and feasibility of constructing a sunshade system at L₁ Lagrange point are evaluated. Some technologically feasible, near-future options for the warming problem are discussed briefly.

Since air pollution caused by PM 2.5 (particulate matter with an aerodynamic diameter of ≤2.5 μm) is a serious threat to human health, the accurate forecasting of PM 2.5 concentration in metropolitan areas is one of the prior conditions to reduce and eliminate the harmful impacts on human beings produced by PM2.5. In this study, we analyzed the spatiotemporal correlations between target and observation parameters relevant to air pollution forecasting and proposed a convolutional neural network (CNN) and long short-term memory (LSTM) model (also called PM predictor) for next day's daily average PM 2.5 concentration forecasting in Beijing. The proposed spatiotemporal correlations were analyzed for efficient estimation of mutual information, not only if the degrees of variations between the two spaces under consideration are similar, but also if the degrees of variations are significantly different, thereby generating a spatiotemporal feature vector. CNN provided an efficient extraction of inherent features for latent air quality and meteorological input data relevant to PM 2.5, and LSTM delivered the historical information in the time series data, thus a novel PM predictor with remarkably improved performance was constructed, compared with multi-layer perceptron (MLP) and LSTM model in overall forecasting. The air quality and meteorological data from the monitoring stations in Beijing and four surrounding cities from January 1, 2015 to December 31, 2017 were adopted as dataset. The forecasting results suggest that the proposed PM predictor is superior to other models in overall forecasting, while LSTM is better than PM predictor with slight difference in seasonal forecasting.

This study presents the daily and seasonal variation of the atmospheric mixing layer height (MLH) over Gadanki, India (13.45°N, 79.18°E), a tropical rural location based on polarization lidar observations. The observations spanned the years 2009–2014, encompassing 303 instances, and coinciding with radiosonde and surface weather station measurements. The MLH was determined through the analysis of aerosol profiles and confirmed with the MLH values derived from radiosonde data. The lidar depolarization ratio was employed to characterize aerosol shape. This study aims to establish a connection between aerosol backscatter and its shape through lidar observations, considering diurnal and seasonal variations, while also identifying the influencing factors. This study illustrates four distinct case studies conducted during different seasons to depict aerosol behavior in both convectively active and non-active periods. These case studies unveil the influence of aerosol shape on water intake and subsequent residual layer and cloud formation. The observed fluctuations in MLH and aerosol shape suggest a dynamic relationship between local meteorology and long-range aerosol transport.

This study analyzes the raindrop size distribution (RSD) characteristics over New Delhi by dividing the year into three seasons: PreM (March–May), monsoon (June–September), and PostM (October–February). Data from a Joss-Waldvogel Disdrometer, installed at IITM New Delhi, Rajendra Nagar, was used for three years (2021–2023). The observed raindrop spectra were fitted with three-parameter Gamma functions to obtain the RSD. ERA-5 and satellite data were also employed to establish atmospheric and cloud properties for the three seasons. The RSD for the monsoon season shows the highest concentration of midsize (1–3 mm diameter) drops and the highest mean rain rate. PostM has the least concentration of midsize and large (diameter >3 mm) drops. General statistics of rain integral parameters reveal high variability in rain rate (R) and mass-weighted mean diameter (D_m) values during the monsoon season. The mu-lambda scatter plots show considerable differences among the three seasons, indicating slightly distinct rainfall mechanisms in the three seasons. Z-R relations of the form Z = aR^b were derived, with the highest coefficient (a) values observed for the PreM precipitation. The exponent (b) is found to be greater than unity in all three seasons. Rainfall was stratified based on rain rate. RSD gets broader with increasing R. Large drops are not found appreciably in the spectrum for R < 20 mm/h. A notable disparity between convective and stratiform RSD is evident. The values of rain integral parameters show considerable differences between the convective and stratiform regimes. A higher fraction of large drops is found for the stratiform rainfall in the PreM season compared to the other two seasons. CAPE, water vapor, surface temperature, and surface winds were higher during PreM and monsoon months compared to PostM. The distribution of differential temperature (δT) indicates that clouds with significant depth are found in PreM and monsoon seasons but are often lacking during PostM.

The Galactic Cosmic Rays (GCR) flux can contribute to the formation of condensation nuclei (CN), radionuclides, and other particles, which in turn influence the formation of rain and extreme weather events. The aim of this analysis was to investigate the possible influence of GCR flux on the extreme rainfall events that occurred in Greece and Libya in September 2023. We used time series data for GCR, rainfall estimates from ERA5, and Sea Surface Temperature (SST) for the period between September 1, 2023, and September 11, 2023. The results revealed a negative correlation between GCR and SST of −0.807 (Greece) and −0.828 (Libya), and a positive correlation between precipitation and SST of +0.972 (Greece) and +0.998 (Libya). The GCR flux and SST accounted for approximately 60.52% and 34.53% of the extreme event in Greece, and 33.71% and 65.96% in Libya, respectively. These statistical results indicate that GCR flux contributed to the formation of the extreme precipitation event that caused significant destruction in Greece and Libya in September 2023.

The processes of ion formation in humid tropospheric air under the action of cosmic rays are considered. In this case, positive and negative cluster ions appear. For this analysis, a kinetic model was developed that includes 55 components and 161 reactions. The calculation was carried out using the KINET software package. It is shown that the ionization of air by cosmic rays at altitudes of 5–35 km leads to the formation of plasma consisting mainly of $N{H}_4^{+}\cdot N{H}_3\cdot H_{2}O$, $H^{+}\cdot {\left(H_{2}O\right)}_4$ and $O_2^{-}\cdot {\left(H_{2}O\right)}_2$ ions. The maximum concentrations of ions under conditions of minimum magnetic rigidity are observed at altitudes from 10 to 18 km. These results differ sharply from the calculation results obtained for the dry air model.