Correlation: weather parameters and PM2.5 in Hanoi and HCMC, Vietnam

Binh Nguyen, Independent Researcher, Hanoi, Vietnam

Abstract
What is in this post?
  • Lots of graphs with big pictures on PM2.5 and wind patterns in Hanoi and Ho Chi Minh City from 2016-2019
  • Only analyzing correlation weather parameters to the transport stage of PM2.5 in which wind and precipitation drive the process
  • Correlation of wind speed and precipitation rate to the observed PM2.5 concentration
  • A closer look on change of wind speed or change of PM2.5 and association of such in Hanoi
  • What this post cannot provide?
  • Causation between PM2.5 to weather patterns or any mathematical-derived relations
  • Any comprehensive conceptualization of mass-balance on PM2.5
  • Critical questions such as source, solution for PM2.5 conundrum
  • 1. Introduction

    Fine particulate matters or particles with a diameter smaller than 2.5µm (called PM2.5 hereafter) are the primary group of air pollutants in developing countries. Effects of PM2.5 on human health range from damages to the respiratory tract, lung cancers, heart strokes to premature deaths.

    PM2.5 created from incomplete burnings such as from internal combustion engines, coal-fired facilities to burning agricultural residues. Some PM2.5 created from other precursors such as the chemical reaction of ammonium with nitrous oxides or sulfur dioxide and forms ammonia salts. In the end, compoents of PM2.5 are included carbonous origins, sulfate, nitrate, crustal origins, water and few unknown quantities. The ratios are reported to be different between urban and rural sites, dry and raining reasons or different locations on the globe.

    A concentration in the ambient air is a temporal state of a mass balance. In this post, I provide correlations of weather parameters to the PM2.5 concentration. The key parameters include wind speed and direction, precipitation, relative humidity and temperature. Climate data was extracted from MERRA-2 products published by NASA. The ground PM2.5 data was downloaded from AirNow measured by FEM stations placed on the Department of States' sites over the globe. This post covers the sites of Hanoi and Ho Chi Minh City, Vietnam.

    The author only evaluated relations between each weather input with PM2.5 concentration with a given timestamp. This approach aims at the correlation and not able to build mechanics for the fate-transport mechanism.

    2. Methods and Materials

    Weather data was extracted from MERRA-2 by the subsetting approach. The MERRA-2 are stored in netCDF format with each file is for a timestamp of analysis with the grid covers the entire globe. To extract limited numbers of sites, subsetting via OpeNDAP is affordable with limited storage space, bandwidth and computing powers.

    The PM2.5 data is stored in the .CSV files accessible via the AirNow.gov website. The files are ready for analysis with minimal efforts to clean up.

    All extracting, cleaning and analyzing data are carried out with Python and its libraries such as Pandas, Matplotlib. All of data and tools used in this report are free and open access.

    3. Results and Discussion
    3.1 Big pictures

    The horizontal transport of PM2.5 is influenced mostly by wind speed and direction. Figs. 1-2 presented windroses and wind speed distribution for Hanoi and HCMC. Dominant winds in Hanoi is from the North-East and South-East directions, and HCMC is South-East and South-West. The long-tail distribution in Fig. 2 indicated a stronger wind in HCMC located in the South of Vietnam and with a flatter surrounding terrain. Hanoi is located in the North in Red River delta surrounded by tall mountain ranges (over 1000m) on the North West direction.

    Background
    Fig. 1: Windroses for Hanoi and Ho Chi Minh city (2016-2019).
    Background
    Fig. 2: Wind speed distribution.

    PM2.5 data in CSV files are tagged with four categories as shown in Figure 3. The distribution provides a reference for the performance of a FEM station. For all analysis followed, only a "valid" tag of PM2.5 hourly concentration was used.

    Background
    Fig. 3: Quality Control of each point (for each hour) are tagged with one of the four types

    Figs. 4-5 gives a perspective of PM2.5 daily-concentration with WHO guidelines, US's NAAQS, and Vietnam National Technical Regulation (QCVN 05:2013).

    Seasonal patterns are pronounced in sites Hanoi and HCMC with a higher concentration during the winter seasons. The PM2.5 in Hanoi has more days above the QCVN05:2013, which is the most lenient in these three levels. In the total of 1427 days during the period of 1/2016 to 1/2020, HCMC has 66 days above the QCVN05 level and 709 days above the WHO's guideline. The number for Hanoi is 414 and 937 out of the total 1459 hours measured during this period.

    For the distribution as shown in Fig. 6, the mean concentration for Hanoi is 46 µg/m3 and HCMC is 27.4 µg/m3. The mean presented for 50% of the hours that have the concentration less than 46 in Hanoi, and 27.4 in HCMC.

    Background
    Fig. 4: Daily-averaged PM2.5 in Hanoi (1/2016-1/2020)
    Background
    Fig. 5: Daily-averaged PM2.5 in Hanoi (2016-1/2020)
    Background
    Fig. 6: Distribution of PM2.5 in Hanoi and HCMC during 2016-1/2020
    3.2 Correlation with observed values
    Wind

    The following figures (figs. 7-8) presented the correlation between PM2.5 with wind speed with four directions. The shade of each point was marked for the total precipitation. A heavy rain (0.4-0.5 kg/m2) correlated with a lower PM2.5 which is inlined with wet removal. A strong wind (>6m/s) also correlated with a lower PM2.5 in a range of 50µg/m3 or less. The high concentration of PM2.5 in Hanoi observed with light to now wind (<4m/s) and almost no precipitation.

    For Hanoi, the East-South wind correlated with a dozen of points with PM2.5 above 250 µg/m3. West-North and North-East contributed a few other high concentration with a light wind. Site HCMC exhibits a long tail of PM2.5 with a low concentration with a strong wind. Heavy rains such as with South-West wind also associated with a lower PM2.5 as in the lower-left graph on Fig. 8.

    Background
    Fig. 7: Correlation of PM2.5 in Hanoi with wind and precipitation
    Background
    Fig. 8: Correlation of PM2.5 in HCMC with wind and precipitation
    Time

    Alternatively, we can examine the correlation of PM2.5 with time such as by months and hours (Figs. 9-10). The range of observed concentration is display as the white background inside the black edge. In both sites, the distribution exihibits a long tail toward the high concentration. This suggests multiple suddent measurements of high PM2.5 concentration.

    Distribution by month refects the variations shown in Figs. 4-5 which mostly driven by seasonal weather patterns with an presumption that no significant fluctuation of emission rates of PM2.5 year around.

    Interesting differences can be observed in Fig. 10 in which Hanoi's PM2.5 was lower in afternoons and highers during midnights to early mornings. This pattern is inlined with frequent temperature inversions in Hanoi observed during the late Fall to early Spring seasons. The temperature inversion occurs at night when the air layer near the ground was cooling faster than the above layer air. The influx of hotter air from South-West with the remain lower temperature air during seasonal transition also created occasion temperature inversion. When this pattern occurs, the vertical mixing is limited and the PM2.5 is kept the near ground at a much higher concentration.

    Background
    Fig. 9: Distribution of PM2.5 by month
    Background
    Fig. 10: Distribution of PM2.5 by hour in day
    Temperature and relative humidity

    The temperature is the primary indicator of the weather conditions. Correlation of the temperature with PM2.5 as shown in Figs. 11 is less obvious. The higher concentration of PM2.5 occurs with a mid-range of the temperature which could be the artifact of more points in the middle. For both sides, lower and higher tails, the PM2.5 concentration is lower. However, when extreme weather occurs, the PM2.5 becomes insignificant.

    Background
    Fig. 11: Daily-averaged PM2.5 in Hanoi (1/2016-1/2020)

    The relative humidity (RH) is an important weather condition that effects the growth of PM2.5 particles. With RH > 70%, that growth becomes significant and a major concern for analyzing the composition and speciation of PM2.5. To reduce the effect of RH on PM2.5 measurement, PM2.5 monitors approved by US EPA (FEM) equipped with a smart heater. The PM2.5 used in this analysis is from FEM monitor such as MetOne BAM-1020 that widely used by US EPA to continuously monitor PM2.5 in ambient air. The underlying assumption here is with proper operation, the effect of RH in ambient air is insignificant to the reliability of PM2.5 reading by such FEM monitors. The standard RH for weight filter in accordance with 40 CRF Part 50, Appendix L is 40%. Other technical references are here (PM2.5 Federal Reference Method, EQPM-0308-170, and Method 201A. A collection of files for PM2.5 continuous monitoring is here.

    Background
    Fig. 12: Daily-averaged PM2.5 in Hanoi (2016-1/2020)

    The RH could be a confusing factor with a high of PM2.5. The confusion stems from the vision reduction possibly associated with a high RH or "mist" condition or with a high PM2.5 concentration. Fig. 13 (left panel) shows the RH with PM2.5 above 100µg/m3. The RH ranges from 50% to over 100% with a more point in close to the condensed level (80-100%). The right panel of Fig. 13 shows with RH >= 90% and the corresponding PM2.5 values. The mean PM2.5 with RH>0.9 is 44µg/m3 which means 50% of sample points with high RH is less than the mean concentration. The mean RH for high PM2.5 is 0.83 or 50% of the sampling point that have 100 or more µg/m3 when RH > 0.83 in the ambient air. The author emphasized that the RH inside the FEM monitor should close to 40% by EPA's requirement. Using both RH and PM2.5 from FEM monitors, we can ascertain if the low vision contributed by high RH in forms of mist or fog or fine particles or both.

    Background
    Fig. 13: Association high PM2.5 (left) and high RH (right) in Hanoi
    3.3 Correlation with changes in observed values in Hanoi

    The above analysis dealt with the observed values. In the following session, we will analyze the change of the observed values.

    With four directions of wind, the largest change either reduction or addition occurs with the wind speed around 2m/s. Strong wind or heavy precipitation seems not lead to signficant changes in PM2.5 concentration. This is logical with the air was already low in PM2.5 as show in Figs. 7-8.

    Background
    Fig. 14: PM2.5 observed values with the change of wind speed

    Figure 15 shows the association of high PM2.5 with the change of wind speed. A high PM2.5 occurs with stagnant air shown in a pine corn shape. Change of wind speed in any direction partially leads to a lower observed PM2.5. Existing strong wind (>6m/s) also leads to a lower PM2.5.

    Background
    Fig. 15: Change of wind speed with change of PM2.5

    Figure 16 provides details of the changes of wind speed and of PM2.5. Stagnant air lead to large fluctuation of observed values which happened mostly during the early mornings and a high observed PM2.5 as shown in Fig. 17.

    Background
    Fig. 16: Change of PM2.5 with change of wind speed
    Background
    Fig. 17: Association large change of PM2.5 with time a day and observed value
    4. Conclusion

    Using available data from the AirNow.gov website and the MERRA-2 reanalaysis product, I analyzed weather patterns to correlate PM2.5 concentration. A stagnant air associated with a high concentration of PM2.5. A stronger wind (>6m/s) or heavy precipitation lowers the PM2.5 concentration around 50µg/m3.

    ▣ ▣ ▣