25 research outputs found
The estimation of ground-level nitrogen dioxide (NO2) and ozone (O3) concentrations using Real-Time Learning (RTL)-based machine learning approach
No full textDepartment of Urban and Environmental Engineering (Environmental Science and Engineering)Nitrogen dioxide (NO2) and ozone (O3) are the significant components of gaseous air pollutants that have harmful effects on human health. The monitoring and analysis of air pollutant exposure and persistence, and short-term forecasts are necessary for efficient public health management. In this study, the estimation model for the ground-level O3 and NO2 concentrations was developed which are spatially continuous over the land and ocean. The ground-level estimation was developed using the RTL-based machine learning technique with various satellite data and numerical model data as input variables. Three models were tested to build an accurate model using the most available data. 1) the ocean model using only ocean variables that have values for all regions2) the land model using all available data with assigning constant values to ocean variables3) the combined model that combines the results of the ocean model for sea area and the results of the land model for land area. Since NO2 and O3 have a relatively short lifespan, the real-time learning model is effective in estimating accurate ground-level concentrations.ope
Estimating surface nitrogen dioxide and ozone concentrations using satellite-based and numerical model-based data
No full text
Estimating Ground-level Particulate Matter Concentrations Using Satellite Observations and Numerical Model Output
No full text
A novel lead detection approach to Arctic sea ice thickness estimation using CryoSat-2 satellite data
No full textEstimation of Ground-level Nitrogen Dioxide and Ozone Concentrations Using Satellite Data and Numerical Model Output
No full textLong exposure to high concentrations of nitrogen dioxide (NO2) and ozone (O3) at ground level could be harmful to human health. Air pollutant concentrations including NO2 and O3 have been measured at monitoring stations, which has a major limitation that it is difficult to provide spatially continuous air quality information. In this study, machine learning based models were developed to estimate ground-level NO2 and O3 concentrations using satellite-based remote sensing data and numerical model output over East Asia to overcome such a limitation. NO2 and O3 vertical column density products from the Aura Ozone Monitoring Instrument (OMI) play an important role in monitoring of the spatial and temporal patterns of the gases, although one third to one half of the OMI products have been missing due to row anomalies. In this study, missing pixels of OMI products were filled using an interpolation approach to generate spatio-temporally continuous distribution of NO2 and O3 concentrations. In addition to satellite-derived data, model-based meteorological parameters and emission information during 2015-2016 were used to estimate surface air quality concentrations over East Asia. Random forest (RF) was used to develop the estimation models for NO2 and O3 concentrations. Over South Korea, the RF-based models showed good performance resulting in R2 values of 0.78 and 0.73, and RMSEs of 8.88 ppb and 10.50 ppb for NO2 and O3, respectively. The NO2 vertical column density was identified most important variable in both models. The model-based meteorological variables such as max wind speed, planetary boundary layer height (PBLH), frictional velocity, and solar radiation were also considered significant for estimation. Spatial distribution of ground-level NO2 and O3 concentrations were also examined over South Korea. Relatively high concentrations were shown around large cities including Seoul metropolitan area
Estimating ground-level particulate matter concentrations using satellite-based data: a review
No full textParticulate matter (PM) is a widely used indicator of air quality. Satellite-derived aerosol products such as aerosol optical depth (AOD) have been a useful source of data for ground-level PM monitoring. However, satellite-based approaches for PM monitoring have limitations such as impacts of cloud cover. Recently, many studies have documented advances in modeling for monitoring PM over the globe. This review examines recent papers on ground-level PM monitoring for the past 10 years focusing on modeling techniques, sensor types, and areas. Satellite-based retrievals of AOD and commonly used approaches for estimating PM concentrations are also briefly reviewed. Research trends and challenges are discussed based on the review of 130 papers. The limitations and challenges include spatiotemporal scale issues, missing values in satellite-based variables, sparse distribution of ground stations for calibration and validation, unbalanced distribution of PM concentrations, and difficulty in the operational use of satellite-based PM estimation models. The literature review suggests there is room for further investigating: 1) the spatial extension of PM monitoring to global scale; 2) the synergistic use of satellite-derived products and numerical model output to improve PM estimation accuracy, gap-filling, and operational monitoring; 3) the use of more advanced modeling techniques including data assimilations; 4) the improvement of emission data quality; and 5) short-term (hours to days) PM forecasts through combining satellite data and numerical forecast model results
Different Spectral Domain Transformation for Land Cover Classification Using Convolutional Neural Networks with Multi-Temporal Satellite Imagery
No full textThis study compares some different types of spectral domain transformations for convolutional neural network (CNN)-based land cover classification. A novel approach was proposed, which transforms one-dimensional (1-D) spectral vectors into two-dimensional (2-D) features: Polygon graph images (CNN-Polygon) and 2-D matrices (CNN-Matrix). The motivations of this study are that (1) the shape of the converted 2-D images is more intuitive for human eyes to interpret when compared to 1-D spectral input; and (2) CNNs are highly specialized and may be able to similarly utilize this information for land cover classification. Four seasonal Landsat 8 images over three study areas-Lake Tapps, Washington, Concord, New Hampshire, USA, and Gwangju, Korea-were used to evaluate the proposed approach for nine land cover classes compared to several other methods: Random forest (RF), support vector machine (SVM), 1-D CNN, and patch-based CNN. Oversampling and undersampling approaches were conducted to examine the effect of the sample size on the model performance. The CNN-Polygon had better performance than the other methods, with overall accuracies of about 93%-95 % for both Concord and Lake Tapps and 80%-84% for Gwangju. The CNN-Polygon particularly performed well when the training sample size was small, less than 200 per class, while the CNN-Matrix resulted in similar or higher performance as sample sizes became larger. The contributing input variables to the models were carefully analyzed through sensitivity analysis based on occlusion maps and accuracy decreases. Our result showed that a more visually intuitive representation of input features for CNN-based classification models yielded higher performance, especially when the training sample size was small. This implies that the proposed graph-based CNNs would be useful for land cover classification where reference data are limited
