Abstract
To address the issues that traditional statistical estimation models have difficulty in representing the spatiotemporal heterogeneity and non-stationarity of haze, as well as the problem of insufficient accuracy, this paper proposes a Temporal and spatial adaptability estimation model (TSAEM). The model introduces the digital elevation model (DEM) and time effect, constructs a four-dimensional space-time kernel function by using three-dimensional space distance and one-dimensional time distance, and calculates four-dimensional space-time weight. Taking the research area of Zhengzhou City, Henan Province, China as an example, based on \(PM_{2.5}\) ground observation data and meteorological data collected from December 2017 to February 2018, \(PM_{2.5}\) was estimated from MODIS MYD-3K AOD data using the GWR, TWR, GTWR and TSAEM models. The results showed that the MAE (mean absolute error) of the TSAEM model decreased by 54.13%, 54.06% and 37.90%, compared to those of the GWR, TWR, and GTWR models, respectively, and that the \(PM_{2.5}\) concentrations predicted by the TSAEM model were closest to the measured values. The \(R^2\) (the correlation coefficient) of the TSAEM model was 0.9496, which was better than those of the GWR ( \(R^2 = 0.7761\) ), TWR ( \(R^2 =0.7763\) ) and GTWR ( \(R^2=0.8811\) ) models. The TSAEM model can reveal the spatial heterogeneity of \(PM_{2.5}\) concentrations and the DEM’s influence on the spatial dimension and improve the precision of \(PM_{2.5}\) estimation.
Graphic Abstract
This study proposes a Spatiotemporal Adaptive Estimation Model (TSAEM) for fine particle concentration, which integrates satellite-derived Aerosol Optical Depth (AOD) with Digital Elevation Model (DEM) and temporal effects. By constructing a four-dimensional spatiotemporal kernel function using three-dimensional spatial distance and one-dimensional temporal distance, TSAEM calculates spatiotemporal weights to reveal the spatial heterogeneity of \(PM_{2.5}\) concentrations and enhance estimation accuracy. Using Zhengzhou City, Henan Province, China, as a case study, \(PM_{2.5}\) ground observation and meteorological data from December 2017 to February 2018 were employed to compare TSAEM with GWR, TWR, and GTWR models. Results show that TSAEM reduces the Mean Absolute Error (MAE) by 54.13%, 54.06%, and 37.90% compared to GWR, TWR, and GTWR, respectively, with a correlation coefficient ( \(R^2\) ) of 0.9496–significantly outperforming other models. This validates DEM’s impact on \(PM_{2.5}\) spatial distribution, demonstrating that TSAEM better captures the four-dimensional spatiotemporal variations of \(PM_{2.5}\) , providing an effective approach for \(PM_{2.5}\) concentration estimation.
