Estimating high-spatial-resolution daily PM2.5 mass concentration from satellite top-of-atmosphere reflectance based on an improved random forest model

Tang, Yuming; Deng, Ruru; Liang, Yeheng; Zhang, Ruihao; Cao, Bin; Liu, Yongming; Hua, Zhenqun; Yu, Jie

doi:10.1016/j.atmosenv.2023.119724

Cited by 7 publications

(3 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thirdly, while preserving its precision, the model performs at a swift pace, and it performs better processing high-dimensional and big data rather than low-dimensional and small data. Additionally, the model offers a weight learning mechanism, which shields against the problem of overfitting in attribute evaluations of large complex nonlinear systems, and evaluates variable importance with a potent ranking function [43]. The technique flowchart for the random forest model is shown in Figure 4.…”

Section: Random Forest Modelmentioning

confidence: 99%

Spatial Characteristics and Influencing Factors of Multi-Scale Urban Living Space (ULS) Carbon Emissions in Tianjin, China

Zhang,

Liao,

Yin

et al. 2023

Buildings

View full text Add to dashboard Cite

Urban living space (ULS) is known to be a significant contributor to carbon emissions. However, there is a lack of studies that have considered the impact of spatial organization indexes (SOIs) of various scales on urban living space carbon emissions (ULSCE), and so far, no definitive conclusions have been reached. To address this gap, taking Tianjin as an example, the measurement methods of ULSCE and SOI at different scales were proposed, and a random forest model was constructed to explore the effects of SOI on ULSCE. The results indicated that on the district scale, Beichen had the highest carbon emissions and absorption in 2021, with carbon emissions reaching 1.43 × 108 t and carbon absorption at 7.29 × 105 kg. In terms of area scale, the comprehensive service area had the highest carbon emissions at 3.57 × 108 t, accounting for 47.70%, while the green leisure area had the highest carbon absorption at 5.76 × 105 kg, accounting for 32.33%. At the block scale, the industrial block had the highest carbon emissions at 1.82 × 108 t, accounting for 54.02%, while the forest block had the highest carbon absorption at 1.25 × 106 kg, accounting for 91.33%. Each SOI had varying impacts, with the industrial land ratio (ILR) having the highest order of importance at the area scale, followed by road network density (RND), residential land ratio (RLR), bus station density (BSD), public service facilities land ratio (PLR), land mixing degree (LMD), open space ratio (OSR), and commercial land ratio (CLR). ILR, RND, and RLR were particularly important, each exceeding 10%, with importance values of 50.66%, 17.79%, and 13.17%, respectively. At the block scale, building area (BA) had the highest importance, followed by building density (BD), building height (BH), land area (LA), and floor area ratio (FAR). BA and BD were particularly important, with values of 27.31% and 21.73%, respectively. This study could serve as both theoretical and practical guidance for urban planning to aid the government in developing differentiated carbon emissions reduction strategies that can mitigate the heat island effect and promote low-carbon healthy urban planning.

show abstract

Section: Random Forest Modelmentioning

confidence: 99%

Spatial Characteristics and Influencing Factors of Multi-Scale Urban Living Space (ULS) Carbon Emissions in Tianjin, China

Zhang,

Liao,

Yin

et al. 2023

Buildings

View full text Add to dashboard Cite

show abstract

“…The first category involves Aerosol optical depth (AOD) derived from Top-of-atmosphere (TOA) reflectance calculations [3,4]. However, the AOD data requires strict surface assumptions during the calculation process, leading to a large amount of missing data in the final data, which directly affects the accuracy of the PM2.5 inversion results [5]. To address this issue, some researchers suggested utilizing TOA reflectance to directly estimate PM2.5 [6,7], but the data they used originated from polar orbiting satellite, which could not reflect the hourly changes in PM2.5 concentration.…”

Section: Introductionmentioning

confidence: 99%

Spatiotemporal continuous PM2.5 concentrations inversion based on multisource data and hybrid model

Wang,

Xu,

et al. 2024

Fifth International Conference on Geology, Mapping, and Remote Sensing (ICGMRS 2024)

View full text Add to dashboard Cite

Air pollution affects public health, destroys ecological environment and even aggravates climate change, and PM2.5 is an important factor causing pollution. Therefore, accurate and continuous spatiotemporal PM2.5 concentration inversion is of great significance for air pollution control. Nevertheless, current high temporal resolution inversions are predominantly conducted using Top-of-atmosphere reflectance, nighttime concentration inversion cannot be realized in the visible band. Moreover, existing hybrid models often rely on linear integrated, increasing the risk of overfitting. Therefore, we propose an improved stacking model that contains XGBoost and CatBoost model, stacked by elastic net regression. By fusing features from multi-source data and accounting for spatiotemporal heterogeneity, the model enables the simultaneous inversion of daytime and nighttime PM2.5 concentrations. Compared to using Top-of-atmosphere reflectance alone, the daytime inversion results demonstrate a 1.08% increase in R 2 and a reduction of 1.34µg/m 3 in RMSE. Meanwhile, the nighttime inversion results show a 2.04% improvement in R 2 and reduce the MAE to 9.54µg/m 3 when spatiotemporal features are used. Applying the model to the Beijing-Tianjin-Hebei region enables spatiotemporal continuous inversion of PM2.5 concentrations in the region.

show abstract

“…In recent research, the application of TOA data for PM 2.5 estimation has become more feasible and practical [5][6][7]. Many researchers have used various TOA products, such as the MODIS TOA and Himawari-8 TOA, to estimate PM 2.5 concentrations [8][9][10][11]. The existing models for TOA-based PM 2.5 estimation can generally be categorized into three types: statistical models, machine learning models, and deep learning models.…”

Section: Introductionmentioning

confidence: 99%

High-Resolution PM2.5 Concentrations Estimation Based on Stacked Ensemble Learning Model Using Multi-Source Satellite TOA Data

Fu,

Guo,

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Nepal has experienced severe fine particulate matter (PM2.5) pollution in recent years. However, few studies have focused on the distribution of PM2.5 and its variations in Nepal. Although many researchers have developed PM2.5 estimation models, these models have mainly focused on the kilometer scale, which cannot provide accurate spatial distribution of PM2.5 pollution. Based on Gaofen-1/6 and Landsat-8/9 satellite data, we developed a stacked ensemble learning model (named XGBLL) combined with meteorological data, ground PM2.5 concentrations, ground elevation, and population data. The model includes two layers: a XGBoost and Light GBM model in the first layer, and a linear regression model in the second layer. The accuracy of XGBLL model is better than that of a single model, and the fusion of multi-source satellite remote sensing data effectively improves the spatial coverage of PM2.5 concentrations. Besides, the spatial distribution of the daily mean PM2.5 concentrations in the Kathmandu region under different air conditions was analyzed. The validation results showed that the monthly averaged dataset was accurate (R2 = 0.80 and root mean square error = 7.07). In addition, compared to previous satellite PM2.5 datasets in Nepal, the dataset produced in this study achieved superior accuracy and spatial resolution.

show abstract

Estimating high-spatial-resolution daily PM2.5 mass concentration from satellite top-of-atmosphere reflectance based on an improved random forest model

Cited by 7 publications

References 49 publications

Spatial Characteristics and Influencing Factors of Multi-Scale Urban Living Space (ULS) Carbon Emissions in Tianjin, China

Spatial Characteristics and Influencing Factors of Multi-Scale Urban Living Space (ULS) Carbon Emissions in Tianjin, China

Spatiotemporal continuous PM2.5 concentrations inversion based on multisource data and hybrid model

High-Resolution PM2.5 Concentrations Estimation Based on Stacked Ensemble Learning Model Using Multi-Source Satellite TOA Data

Contact Info

Product

Resources

About