A deep learning model integrating a wind direction-based dynamic graph network for ozone prediction

Ozone pollution is an important environmental issue in many countries. Accurate forecasting of ozone concentration enables relevant authorities to enact timely policies to mitigate adverse impacts. This study develops a novel hybrid deep learning model, named wind direction-based dynamic spatio-temporal graph network (WDDSTG-Net), for hourly ozone concentration prediction. The model uses a dynamic directed graph structure based on hourly changing wind direction data to capture evolving spatial relationships between air quality monitoring stations. It applied the graph attention mechanism to compute dynamic weights between connected stations, thereby aggregating neighborhood information adaptively. For temporal modeling, it utilized a sequence-to-sequence model with attention mechanism to extract long-range temporal dependencies. Additionally, it integrated meteorological predictions to guide the ozone forecasting. The model achieves a mean absolute error of 6.69 μg/m3 and 18.63 μg/m3 for 1-h prediction and 24-h prediction, outperforming several classic models. The model's IAQI accuracy predictions at all stations are above 75 %, with a maximum of 81.74 %. It also exhibits strong capabilities in predicting severe ozone pollution events, with a 24-h true positive rate of 0.77. Compared to traditional static graph models, WDDSTG-Net demonstrates the importance of incorporating short-term wind fluctuations and transport dynamics for data-driven air quality modeling. In principle, it may serve as an effective data-driven approach for the concentration prediction of other airborne pollutants.

Climate Change Risk of Urban Growth and Land Use/Land Cover Conversion: An In-Depth Review of the Recent Research in Iran

This research is the first literature review of the past three decades’ studies on the effects of urban development and land use/land cover (LULC) change on Iran’s climate change. For this purpose, 67 articles were found, evaluated, and classified according to the spatial and temporal scale, case study, period, data type, climatic factor, methodology, and meteorological data. Moreover, the reviewed literature methodologies were classified according to the purpose, method, and data source. According to the spatial-scale results, national- and city-level studies had the lowest and highest numbers, respectively. Tehran was the most case studies because Tehran is Iran’s capital and the largest metropolitan city. In terms of the temporal scale, studies predicting future changes (urban development and climate change) included 5% of the total literature. Satellite images were the most applied data in the reviewed literature (58%). Overall, 79% of the studies used temperature-related factors to explain the climatic impacts of urban growth and LULC conversion. Spatial modeling with 52% publications was the most used method, while numerical modeling with 12% studies was the least used method. This review showed broad study gaps in applying numerical models, neighborhood scales, urban micro-scale parameters, and long-term projections forecasts due to rapid urban development in Iran compared to the rest of the world. Therefore, our synthesis will assist researchers in facilitating better design for future studies in Iran and similar countries.

Future Climate Change Impact on Urban Heat Island in Two Mediterranean Cities Based on High-Resolution Regional Climate Simulations

The Mediterranean is recognized among the most responsive regions to climate change, with annual temperatures projected to increase by 1–5 °C until 2100. Large cities may experience an additional stress discomfort due to the Urban Heat Island (UHI) effect. In the present study, the WRF-ARW numerical weather prediction model was used to investigate the climate change impact on UHI for two Mediterranean cities, Rome and Thessaloniki. For this purpose, three 5-year time-slice simulations were conducted (2006–2010, 2046–2050, 2096–2100) under the Representative Concentration Pathway (RCP) 8.5 emission scenario, with a spatial resolution of 2 km. In order to comprehensively investigate the urban microclimate, we analyze future simulation data across sections crossing urban/non-urban areas, and after grouping them into three classes depending on the location of the grid cells. The urban areas of both cities present increased average minimum temperature (Tmin) in winter/summer compared to other rural areas, with an UHI of ~+1.5–3 °C on average at night/early morning. Considering UHI under future climate change, we found no significant variations (~±0.2 °C). Finally, we found that the numbers of days with Tmin ≥ 20 °C will mostly increase in urban coastal areas until 2100, while the largest increase of minimum Discomfort Index (DImin) is expected in urban low-ground areas.

High Resolution Air Quality Forecasting over Prague within the URBI PRAGENSI Project: Model Performance during the Winter Period and the Effect of Urban Parameterization on PM

The overall impact of urban environments on the atmosphere is the result of many different nonlinear processes, and their reproduction requires complex modeling approaches. The parameterization of these processes in the models can have large impacts on the model outputs. In this study, the evaluation of a WRF/Comprehensive Air Quality Model with Extensions (CAMx) forecast modeling system set up for Prague, the Czech Republic, within the project URBI PRAGENSI is presented. To assess the impacts of urban parameterization in WRF, in this case with the BEP+BEM (Building Environment Parameterization linked to Building Energy Model) urban canopy scheme, on Particulate Matter (PM) simulations, a simulation was performed for a winter pollution episode and compared to a non-urbanized run with BULK treatment. The urbanized scheme led to an average increase in temperature at 2 m by 2 ∘ C, a decrease in wind speed by 0.5 m s − 1 , a decrease in relative humidity by 5%, and an increase in planetary boundary layer height by 100 m. Based on the evaluation against observations, the overall model error was reduced. These impacts were propagated to the modeled PM concentrations, reducing them on average by 15–30 μ g m − 3 and 10–15 μ g m − 3 for PM 10 and PM 2.5 , respectively. In general, the urban parameterization led to a larger underestimation of the PM values, but yielded a better representation of the diurnal variations.

Using a deep convolutional neural network to predict 2017 ozone concentrations, 24 hours in advance

In this study, we use a deep convolutional neural network (CNN) to develop a model that predicts ozone concentrations 24 h in advance. We have evaluated the model for 21 continuous ambient monitoring stations (CAMS) across Texas. The inputs for the CNN model consist of meteorology (e.g., wind field, temperature) and air pollution concentrations (NO _x and ozone) from the previous day. The model is trained for predicting next-day, 24-hour ozone concentrations. We acquired meteorological and air pollution data from 2014 to 2017 from the Texas Commission on Environmental Quality (TCEQ). For 19 of the 21 stations in the study, results show that the yearly index of agreement (IOA) is above 0.85, confirming the acceptable accuracy of the CNN model. The results also show the model performed well, even for stations with varying monthly trends of ozone concentrations (specifically CAMS-012, located in El-Paso, and CAMS-013, located in Fort Worth, both with IOA=0.89). In addition, to ensure that the model was robust, we tested it on stations where fewer meteorological variables are monitored. Although these stations have fewer input features, their performance is similar to that of other stations. However, despite its success at capturing daily trends, the model mostly underpredicts the daily maximum ozone, which provides a direction for future study and improvement. As this model predicts ozone concentrations 24 h in advance with greater accuracy and computationally fewer resources, it can serve as an early warning system for individuals susceptible to ozone and those engaging in outdoor activities.

Setting-up a Real-Time Air Quality Forecasting system for Serbia: a WRF-Chem feasibility study with different horizontal resolutions and emission inventories

In this paper, the influence of the horizontal model grid size and anthropogenic gridded emissions on the air quality forecast in Serbia was analyzed using the online-coupled Weather Research and Forecasting model with Chemistry (WRF-Chem). For that purpose, six simulations were performed. The model horizontal grid size was 20 × 20 km, 10 × 10 km, and 5 × 5 km. Two anthropogenic gridded emission inventories with different grid sizes were used, the global RETRO (REanalysis of the TROpospheric chemical composition) and the EMEP (The European Monitoring and Evaluation Program) for each model horizontal grid size. The modeled O₃, NO₂, and PM₁₀ concentrations in all six simulations were compared with the measured hourly data at the Serbian Environmental Protection Agency (SEPA) stations and an EMEP station during August 2016. The analysis shows that the influence of the model grid size is larger on PM₁₀ than on the O₃ and NO₂ concentration. The concentration of O₃ and PM₁₀ has a similar dependence on the emissions and the model grid size, while NO₂ has a larger dependence on the emission than on the model grid size. The simulation with the 5 × 5 km grid size and the EMEP anthropogenic emissions has optimal performance compared with the measured concentration. In this optimal simulation, the modeled O₃ concentrations overestimated the measured values at 3 stations and underestimated the measured values at 2 stations. At most stations, the modeled NO₂ concentrations underestimated the measured values. The modeled PM₁₀ concentrations highly underestimated the measured values at all stations.

Chemical weather forecasting: a new concept of integrated modelling

Abstract. During the last decade a new field of atmospheric modelling – the chemical weather forecasting (CWF) – is quickly developing and growing. However, in the most of the current studies and publications, this field is considered in a simplified concept of the off-line running chemical transport models with operational numerical weather prediction (NWP) data as a driver. A new concept and methodology considering the chemical weather as two-way interacting meteorological weather and chemical composition of the atmosphere is suggested and discussed. The on-line integration of mesometeorological models and atmospheric aerosol and chemical transport models gives a possibility to utilize all meteorological 3-D fields in the chemical transport model at each time step and to consider feedbacks of air pollution (e.g. urban aerosols) on meteorological processes/climate forcing and then on the atmospheric chemical composition. This very promising way for future atmospheric simulation systems (as a part of and a step to Earth System Modelling) will lead to a new generation of models for meteorological, environmental and chemical weather forecasting. The methodology how to realise the suggested integrated CWF concept is demonstrated on the example of the European Enviro-HIRLAM integrated system. The importance of different feedback mechanisms for CWF is also discussed in the paper.

Multi-scale atmospheric environment modelling for urban areas

Abstract. Modern supercomputers allow realising multi-scale systems for assessment and forecasting of urban meteorology, air pollution and emergency preparedness and considering nesting with obstacle-resolved models. A multi-scale modelling system with downscaling from regional to city-scale with the Environment – HIgh Resolution Limited Area Model (Enviro-HIRLAM) and to micro-scale with the obstacle-resolved Micro-scale Model for Urban Environment (M2UE) is suggested and demonstrated. The M2UE validation results versus the Mock Urban Setting Trial (MUST) experiment indicate satisfactory quality of the model. Necessary conditions for the choice of nested models, building descriptions, areas and resolutions of nested models are analysed. Two-way nesting (up- and down-scaling), when scale effects both directions (from the meso-scale on the micro-scale and from the micro-scale on the meso-scale), is also discussed.