Causality Analysis of Air Quality and Meteorological Parameters for 1 PM 2.5 Characteristics Determination: Evidence from Jakarta 2

Abstract


A C C E P T E D M A N U S C R I P T
8 with both the temperature and wind speed in 2016. As the temperature reaches its minimum value 148 during the morning, aerosol will not be able to rise into the atmosphere. On the contrary, aerosol 149 tend to rise into the atmosphere during the afternoon accompanied by an increase of temperature. 150 Moreover, the mean wind speed with a value less than 1 knot is not strong enough and causes 151 pollutant to be trapped near the Earth surface (Vallero, 2014;Hutauruk et al., 2020). It aligns with 152 our finding that there is a high concentration of PM2.5 in the morning. 153

Data collection 154
The data of PM2.5 can be accessed through http://aqicn.org/ (last accessed: August 4 th , 2022) 155 which contains official information about PM2.5 AQI (air quality index) collected by the US 156 Embassy in Indonesia (https://www.airnow.gov) using several sensors in Central Jakarta (-6.18098, 157 106.83021) and South Jakarta (-6.25521, 106.80699) since December 4 th , 2015. In addition, AQI 158 PM2.5 data were also collected from Indonesia's Meteorological, Climatological, and Geophysical 159 Agency (BMKG). Furthermore, information regarding the air quality index in Jakarta including 160 PM10, NO2, SO2, CO, and O3 since 2010 can be accessed through https://data.jakarta.go.id/dataset/. 161 As shown in Fig. 2, the data were collected from five different locations in Jakarta, such as 162 Bunderan HI (HI) sensors which are located on the side of the roadway in Central Jakarta (-163 6.194699,106.823028), Kelapa Gading (KG) sensors which are located near the industrial and 164 residential areas in North Jakarta (-6.159690,106.905541), Jagakarsa (JK) sensors which are A C C E P T E D M A N U S C R I P T 9 located near residential areas with lots of vegetations in South Jakarta (-6.336216, 106.818082), 166 Lubang Buaya (LB) sensors which are located near residential areas in East Jakarta (-6.290561, 167 106.906839), and Kebon Jeruk (KJ) sensors which are located near the residential areas in West 168 Jakarta (-6.192061, 106.770606). Meteorological parameter data were obtained from the 169 https://power.larc.nasa.gov/data-access-viewer/ website that contains data from the MERRA-2 170 satellite. The AQI data of PM2.5, PM10, NO2, SO2, CO and O3 that we collected ranges from 171 December 25 th , 2015, to February 28 th , 2020. 172

Method 173
As shown in Fig. 3, the Pearson correlation analysis examined the relationship and influence of 174 meteorological and air quality parameters based on seasonal variations in PM2.5 concentrations. 175 The correlation coefficient ( ) was used to determine whether the relationship between the free 176 variables and the bound variable is strongly correlated or not (Pearson et al., 1895). The 177 causality was then analyzed using the CCM method. 178

Convergent cross mapping (CCM) analysis 179
As shown in Fig. S2, in the = ( , ) system, cross mapping means that the point on the 180 manifold is a variable , so the corresponding point can be searched for the same time . 181 According to the Takens theorem, we can calculate the and shadow manifolds that cross 182 the map to the actual manifolds of the system (Sauer, 2006;Huke, 2006 If the value of ( | ) is high, prediction from to is good. In contrast, if ( | ) is low, 192 prediction from to is bad. ↔ means bidirectional causality. If the value of ( | ) and 193 ( | ) are high, predictions from to and to are good. 194 When compared with Granger causality which states that causes , we will be able to predict 195 if the is given (   Jakarta (-0.017, 0.52) and South Jakarta (-0.23, 0.58). In JK, LB, and KJ area, however, a high 278 concentration of CO particles is mainly caused by domestic activities. 279

Causality of PM2.5 concentrations on meteorological parameters 280
As shown in Table 1 Table S2.  Fig. 6 is then made to illustrate the nonlinear coupling patterns of air quality parameters 311 with PM2.5 concentration based on seasonal variation.

A C C E P T E D M A N U S C R I P T
17 As shown in Table 2