• EPA completes hourly particulate matter characterization and quantification, including heavy metals and harmful compounds.
• Statistical analysis of these data sets allows us to make inferences about the most prominent pollution in Phoenix.
• Over 52 different parameters were measured in this data set, so I chose to focus on metals and semi metal compounds.
• This data set analyzes particulate matter in the size cut of 2.5 microns. These particles can enter the lungs and diffuse into our blood, making them dangerous for public health.

• Phoenix data collected at the JLG Supersite near Uptown Phoenix.

colnames(data)

##  [1] "State.Code"          "County.Code"         "Site.Num"           
##  [4] "Parameter.Code"      "POC"                 "Latitude"           
##  [7] "Longitude"           "Datum"               "Parameter.Name"     
## [10] "Sample.Duration"     "Pollutant.Standard"  "Date.Local"         
## [13] "Units.of.Measure"    "Event.Type"          "Observation.Count"  
## [16] "Observation.Percent" "Arithmetic.Mean"     "X1st.Max.Value"     
## [19] "X1st.Max.Hour"       "AQI"                 "Method.Code"        
## [22] "Method.Name"         "Local.Site.Name"     "Address"            
## [25] "State.Name"          "County.Name"         "City.Name"          
## [28] "CBSA.Name"           "Date.of.Last.Change"

-Limit data set to metals and ions, remove NA, and average the concentration over the entire year

names = unique(data$Parameter.Name)[12:45]
metals_and_ions <-
  filter(data, Parameter.Name %in% names & City.Name == "Phoenix")
averages <- c()
for (name in names) {
  parameter_data <-
    filter(metals_and_ions,
           Parameter.Name == name & !is.na(Parameter.Name))
  averages <- append(averages, mean(parameter_data$Arithmetic.Mean))
}
df <- data.frame(Names = names, Average.Value = averages)

• I chose to use Plotly to create graphs and RColorBrewer to create a ramp of colors.
• The concentrations are measured in micrograms per cubic meter.

bar_df <- filter(df, df$Average.Value < 0.1, df$Average.Value > 0)
colors <- colorRampPalette(c('red', 'blue'))
p_average_yearly <- plot_ly(
  x = bar_df$Names,
  y = bar_df$Average.Value,
  type = 'bar',
  marker = list(color = colors(nrow(bar_df)))
) %>%
  layout(
    title = "Yearly Average Concentrations of Metal/Ionic PM2.5 in Phoenix",
    xaxis = list(title = "Analytes"),
    yaxis = list(title = "Concentration (ug/cubic meter)")
  ) 

• Research question: Which metal concentrations were highest on average?
• The three highest average concentrations of heavy metals over the entire year in Phoenix were Magnesium, Barium, and Copper.
• However, the average values are not especially significant. In terms of public health, the highest values are the most important.
• Next, we will graph the 95th percentile of each category.

• For each parameter, calculate the 95th percentile and append it to a new array

percentile_95 <- c()
for (name in names) {
  parameter_data <-
    filter(metals_and_ions,
           Parameter.Name == name & !is.na(Parameter.Name))
  percentile_95 <-
    append(percentile_95,
           quantile(parameter_data$Arithmetic.Mean, 0.95))
}
df <- data.frame(Names = names, Percentile = percentile_95)

• I limited the axes here to not include abundant ions like Calcium or Chloride simply because they aren’t as harmful as heavy metals, which was more interesting to me.

percentile_95_df <-
  filter(df, df$Percentile < 0.1 & df$Percentile > 0)
p_percentile_95 <- plot_ly(
  x = percentile_95_df$Names,
  y = percentile_95_df$Percentile,
  type = 'bar',
  marker = list(color = colors(nrow(percentile_95_df)))
) %>% layout(
  title = "95th percentile concentrations of Heavy Metal PM2.5 in Phoenix",
  xaxis = list(title = "Analytes"),
  yaxis = list(title = "Concentration (ug/cubic meter)")
)

• Research Question: Which metals are highest in concentration in the 95th percentile?
• The highest concentrations this time were found for Titanium, Cerium, Barium, and Zinc.
• Now these metals will be plotted over the year in order to spot when high values occur.

• Define the names of the parameters and remove problematic dates (their values were NA and caused column length errors)

top_six_names <- c(
  "Zinc PM2.5 LC", "Barium PM2.5 LC", "Titanium PM2.5 LC",
  "Cerium PM2.5 LC", "Magnesium PM2.5 LC", "Copper PM2.5 LC")

dates_removed <-
  filter(
    metals_and_ions,
    metals_and_ions$Date.Local != "2021-02-03" &
      metals_and_ions$Date.Local != "2021-03-17"
  )

• Use group_by and summarize functions to group the data by day and parameter name, then create a new data frame with average daily concentrations for each parameter.

summarized_df <- dates_removed %>%
  group_by(Date.Local, Parameter.Name) %>%
  summarize(average = mean(Arithmetic.Mean))

line_plot_df <- data.frame(Date = unique(summarized_df$Date.Local))


for (name in top_six_names) {
  column <- (filter(summarized_df, Parameter.Name == name))$average
  line_plot_df[word(name, 1)] <-
    c(column, rep(NA, nrow(line_plot_df) - length(column)))
}

fig <- plot_ly(
  line_plot_df,
  x =  ~ Date,
  y =  ~ Zinc,
  name = "Zinc",
  type = "scatter",
  mode = "lines"
) %>% layout(
  title = "Concentrations of Five Metals during the Year",
  yaxis = list(range = list(0, 0.20),
               title = "Concentration (ug/cubic meter)")
)
fig <- fig %>% add_trace(y = ~ Titanium, name = "Titanium")
fig <- fig %>% add_trace(y = ~ Cerium, name = "Cerium")
fig <- fig %>% add_trace(y = ~ Magnesium, name = "Magnesium")
fig <- fig %>% add_trace(y = ~ Copper, name = "Copper")

• One of the most interesting findings is the huge spike near New Years and near July 4th. I would attribute this to the use of fireworks.

• At several points, the trends for several metals match, i.e. spiking at the same time. This could be due to meteorological effects like haboobs kicking up dust and rocks into the air.

• From what I could find, none of these heavy metals violate the EPA standards for PM2.5 air quality.

• Concentrations for Magnesium and Titanium seem to be highest across several days during the summer, while Zinc is highest in spring. Cerium seems fairly unstable, reaching negative values frequently during the year. Copper is stable throughout the year.