Air Quality Zonal Modeling

We conducted air quality zonal modeling for predicting zonal change in air quality. In general, the air quality zonal modeling represents the overall pollution concentration in any specific region based on several pollutant particles obtained from remotely sensed raster datasets [73]. In other words, the air quality zonal model is a compilation of total air pollution load in a particular zone accumulated by various air pollution particles which is represented through individual grids over a specific time period. This spatial and temporal model were created through different phases using various statistical analyses of GES-DISC time-averaged map and area-averaged time-series datasets obtained from MODIS- Terra, MERRA-2, OMI, and AIRS.

For this study, we conducted various sets of statistical analyses for interpreting remotely sensed data. In the first phase, we divided China's mainland into 145 grids, each grid representing 65,025 km2. After that, we extracted each pollutant's average concentration from an individual grid using the raster calculator in ArcMap. The Moving Mean (MM) of 10 days total column of NO2 of 15 cities was interpreted using equation 1. A specific grid value was ranked using exploratory analysis of ranking given by Alvo and Philip [74] (Equation (2)). The average concentration of each air pollutant particle of 145 grids was categorized based on the maximum and minimum range of datasets which is computed in the second phase (Equation (3)). After that, we compiled total pollution concentration in a specific grid using composite indexing and principal component analysis (PCA); subsequently, we conducted Spearman's model for factor analysis for assessing the maximum and minimum reduction [75] in specific air pollutants which is compiled in the third phase (Equations (2) and (3)). The 60-day average total air pollution load was calculated using Equation (4); here, we excluded zero factors of different raster datasets. The sum of the 60-day average concentration of different air pollutants and their percentage change was calculated using Equation (5). Zonal indexing of specific grids was formulated for extracting the exact concentration of air pollution (Equation (6)) and in the last phase, we calculated area/grid-wise total air pollution load (Equations (7) and (8)) through spatial analysis tools in GIS using air quality zonal modeling.

$$
\lambda m m = \frac{\sum 10}{n} \times 6 \tag{1}
$$

where *mm* = moving mean, ∑ 10 = sum of 10 days average, *n* = number, and 6 = number of raster data.

$$m = \sum\_{j=1}^{d \ge 1} \frac{\left(n\_j v\_j\right) \times i}{n} \tag{2}$$

where *m* = mean rank, *dx* = dimension, *i* = th entry equals, *vj*, *j* = 1 represents all possible rankings of the *dx*, *nj* = frequency of ranking, *vj*(*i*) = rank score, and *n* = number of raster data sets.

$$\mathbf{x} = \begin{bmatrix} n\_1 \ F\_1 + n\_2 \ F\_2 + \end{bmatrix} \dots \begin{bmatrix} + n\_{mm} \ F\_m + s\_m \ F\_m \end{bmatrix} \tag{3}$$

where *x* = variation in maximum and minimum range of air pollutants with zero mean, *n*1, *n*<sup>2</sup> = loading factors of specified air pollutants and *nmm* = moving mean, *F*1, *F*<sup>2</sup> = common factors and *Fm* = moving mean of the common factor with zero mean, and *sm* = specific factor of the individual mean.

$$\mu p = \left[\frac{\sum (\mathbf{x1} + \mathbf{x2} + \mathbf{x3} + \mathbf{x4} + \mathbf{x5} + \mathbf{x6})}{6}\right] \cup [na] \tag{4}$$

where *μp* = 60 days average of pollution load, *x* = sum of 10 days average of pollution load, 6 = number of datasets, and ∪ [*na*] = excluding factor of zero (data not available).

$$p = \frac{\left(\sum 60a - \sum 60b\right)}{\sum 60b} \times 100\tag{5}$$

where *p* = change in percentage, ∑ 60 = sum of 60 days average concentration, *a* = during lockdown, and *b* = before.

$$a = \sqrt{\frac{\sum wi(\chi i - \chi)^2}{\sum wi}}\tag{6}$$

where *a* = area of individual cells, *w* = weights, *i* = an index over all the data points being averaged, and *xi* = individual pollutant variables.

$$t = \sum tapp - [na] \tag{7}$$

where *t* = total pollution counts in individual grids, *taop* = total air pollution load, [*na*]= excluding 'data not available' girds.

$$a = n \times d\mathfrak{x} \tag{8}$$

where *a* = total area of each category, *n* = number of countable grids, and *dx* = dimension of different variables/ grids (255 × 255 km2).
