Analysis for story and graphics: In urban heat islands, climate crisis hits harder

Explanation [cq]

Over an 11-day period in July 2019 – July 12 through July 22 – each day during that stretch had a maximum temperature of at least 90 degrees F and maximum heat index values of at least 92 degrees F. This was measured at a National Weather Service monitoring station located in the Inner Harbor.

Supporting code and output [cq]

 dmh %>%
  filter(month == 7,
         year == 2019,
         day >= 12,
         day <= 22) %>%
  group_by(`date`) %>%
  summarise(min_temp = min(avg_hourly_temperature_dmh),
            max_temp = max(avg_hourly_temperature_dmh),
            mean_temp = mean(avg_hourly_temperature_dmh),
            min_heat_index = min(avg_hourly_heat_index_dmh),
            max_heat_index = max(avg_hourly_heat_index_dmh),
            mean_heat_index = mean(avg_hourly_heat_index_dmh)
  ) %>%
  select(date, max_temp, max_heat_index)

Explanation [cq]

Tammy Jackson, a resident of Baltimore’s McElderry Park neighborhood, had a temperature and humidity sensor installed in her house by our reporters. In her home, the highest temperature reading on Saturday, July 20, the day referenced in the story, was 92 degrees (91.6 degrees F).

For outdoor temperature, in most of our analysis, we used hourly snapshot readings taken at a National Weather Service monitoring station in the Inner Harbor. As the table below shows, the highest hourly reading in that data set on July 20 was 99 degrees. But the NWS official daily summary for that monitoring station – which incorporates all readings taken, not just hourly snapshots – indicates it did hit 100 degrees at some point on July 20. The data can be viewed here.

Supporting code and output [cq]

# Max temperature in Jackson's House on July 20

tammy_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  filter(date == "2019-07-20") %>%
  group_by(date) %>%
  summarise(max_indoor_temperature = max(mean_indoor_temperature))

# Max temperature outside on July 20
dmh %>%
 filter(month == 7,
         year == 2019,
        day == 20) %>%
  group_by(`date`) %>%
  summarise(max_temp = max(avg_hourly_temperature_dmh))

Explanation [cq]

The date and time referenced in the graphic was 11:20 a.m. on August 1, 2019. To get the outdoor temperature we used the NWS hourly snapshots at the Inner Harbor monitoring station. In the 11 a.m. hour, it was 87.1 degrees.

Supporting code and output [cq]

dmh %>%
 filter(month == 8,
         year == 2019,
         day == 1,
         hour == 11) %>%
  select(date, avg_hourly_temperature_dmh)

Explanation [cq]

To compute this average annual temperature change over the last century, we pulled historical average annual temperatures between 1895 and 2018, and the amount each annual average departed from the overall mean temperature during that period. The data for Baltimore and the U.S. came from the National Climactic Data Center. We used linear regression to examine the trend and produce a line of best fit through plotted points. We took the slope of the line of best fit and multiplied it by the number of years to arrive at the temperature increase over that period for Baltimore (3.4 degrees F), the U.S. (1.9 degrees F) and the difference between the two. Baltimore increased by 3.4 degrees F, 80 percent (nearly twice as much) as the U.S. (1.9 percent). A note: we got the idea to use this method from the excellent Washington Post interactive “2°C: BEYOND THE LIMIT”, which arrived at the same figures as we did for Baltimore and the U.S.

Supporting code and output [cq]

# Create linear model showing trend of change in annual average temperature change 1895-2018 for the US
us_lm <- lm(anomaly ~ date, data = us_change_temp)

# Get the slope of the line of best fit from the linear model
us_slope <- tidy(us_lm) %>%
  filter(term == "date") %>%
  mutate(location = "us") %>%
  select(location, estimate) %>%
  rename(slope = estimate)

# Get the number of yearly observations from original data
us_years_count <- us_change_temp %>%
  # group_by(date) %>%
  summarise(years_count= n()) %>%
  mutate(location = "us") %>%
  select(location, years_count)

# Calculate degrees change
us_historical_change <- us_years_count %>%
  left_join(us_slope, by = "location") %>%
  mutate(change_degrees = years_count * slope)

# Create linear model showing trend of change in annual average temperature change 1895-2018 for Baltimore
baltimore_lm <- lm(anomaly ~ date, data = baltimore_change_temp)

# Get the slope of the line of best fit from the linear model
baltimore_slope <- tidy(baltimore_lm) %>%
  filter(term == "date") %>%
  mutate(location = "baltimore") %>%
  select(location, estimate) %>%
  rename(slope = estimate)

# Get the number of yearly observations from original data
baltimore_years_count <- baltimore_change_temp %>%
  # group_by(date) %>%
  summarise(years_count= n()) %>%
  mutate(location = "baltimore") %>%
  select(location, years_count)

# Calculate degrees change
baltimore_historical_change <- baltimore_years_count %>%
  left_join(baltimore_slope, by = "location") %>%
  mutate(change_degrees = years_count * slope)

# Bind together US and Baltimore
us_baltimore <- bind_rows(baltimore_historical_change, us_historical_change)

# Show change degrees and diference between two values as percentage
difference <- us_baltimore %>%
  select(location, change_degrees) %>%
  tidyr::spread(location, change_degrees) %>%
  mutate(difference = (baltimore-us)/us)

difference

Explanation [cq]

The Union of Concerned Scientists [Killer Heat in the United States] study released this summer included detailed tables projecting the number of days of 100+ heat index days for most U.S. cities, and projected how many additional days of 100+ heat index days there would be by mid-century. They found there were six days of 100+ heat index days per year historically in Baltimore, and by mid-century there would be 37 days, a 6x increase. Note: our own analysis of historical Baltimore heat index data at a NWS Inner Harbor monitoring station found that the 6 days per year figure was likely a conservative estimate.

Supporting code and output [cq]

heat_index_projections %>%
  select(state, city, historical_100_plus, midcentury_no_action_100_plus) %>%
  filter(state == "MD", city == "Baltimore") %>%
  mutate(increase_x = midcentury_no_action_100_plus/historical_100_plus)

Explanation [cq]

Tammy Jackson, a resident of Baltimore’s McElderry Park neighborhood, had a temperature and humidity sensor installed in her house by our reporters. In her home, at 10 p.m. on Sunday, July 21, the heat index was 93 degrees F, 9 degrees hotter than the outside heat index of 84 degrees.

Supporting code and output [cq]

tammy_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-21") %>%
  filter(hour == 22) %>%
  select(date, mean_indoor_heat_index, mean_outdoor_heat_index, indoor_heat_index_difference)

Explanation [cq]

The researchers took detailed measurements of Baltimore’s urban heat island on August 29, 2018, and found differences across the city. Using the mean afternoon temperature for each neighborhood (excluding the representation of Leakin Park in our data – we found the coolest neighborhood was Dickeyville (91 degrees F) and the hottest was McElderry Park (99.4 degrees F), a difference of 8.4 degrees F.

Supporting code and output [cq]

nsa_tree_temp %>%
  select(nsa_name, temp_mean_aft) %>%
  filter(nsa_name != "gwynns falls/leakin park") %>%
  filter((temp_mean_aft == min(temp_mean_aft)) | (temp_mean_aft == max(temp_mean_aft))) %>%
  arrange(desc(temp_mean_aft)) %>%
  tidyr::spread(nsa_name, temp_mean_aft) %>%
  mutate(difference_coolest_hottest = `mcelderry park`-`dickeyville`)

Explanation [cq]

In Baltimore’s “community statistical areas”, we examined the relationship between heat (mean afternoon temperature in our urban heat island data) and poverty, life expectancy, unemployment rates and violent crime by computing the correlation coefficient (r) for each metric. An r of 1 would indicate a perfect positive linear relationship and an r of -1 would indicate a perfect negative linear relationship and 0 indicating no relationship. There were moderate correlations with poverty (r =.4), violent crime (r=.58), life expectancy (r=-.41), and the unemployment rate (r=.32).

A table with the values used in the graphic is also displayed below. The same table is used in the scatterplot graphics that appear lower in the story.

Supporting code and output [cq]

# Correlation coefficients
csa_tree_temp_demographics %>%
  select_if(is.numeric) %>%
  as.matrix() %>%
  correlate() %>%
  focus(matches("temp_")) %>%
  mutate(variable=rowname) %>%
  select(variable, temp_mean_aft) %>%
  filter(str_detect(variable, "households_living|violent|life|unemployment"))

# Graphic table
csa_tree_temp_demographics %>%
  select(csa2010, temp_mean_aft, matches("households_living|violent|life|unemployment"))

Explanation [cq]

Using the mean afternoon temperature in the urban heat island study, McElderry Park was the city’s hottest neighborhood, with a mean afternoon temperature of 99.4 degrees F.

Supporting code and output [cq]

nsa_tree_temp %>%
  select(nsa_name, temp_mean_aft) %>%
  arrange(desc(temp_mean_aft))

Explanation [cq]

We used condition prevalence rates from inpatient hospital admissions and emergency room visits, which are only available by ZIP code, and the median afternoon temperature of Baltimore ZIP codes, as calculated in the urban heat island data to examine relationships. The correlation coefficient for people diagnosed with asthma during hospital visits (r=.56) and emergency room visits (r=.49), and people diagnosed with COPD during hospital visits (r=.67) and emergency room visits (r=.57), indicated a moderate to strong positive relationship between heat and asthma and copd rates. This is not causal. In hotter areas, rates of asthma and copd are higher, and vice versa.

Supporting code and output [cq]

ip_full_zip_correlation_matrix %>%
    filter(str_detect(rowname, "asthma|copd")) %>%
    select(rowname, temp_median_aft)

op_er_full_zip_correlation_matrix %>%
    filter(str_detect(rowname, "asthma|copd")) %>%
    select(rowname, temp_median_aft)

Explanation [cq]

By merging a dataset of EMS calls with a dataset of hourly Baltimore heat index values, we were able to determine the heat index at the time of each EMS call during summer 2018. We looked at select conditions affected by heat, and compared how call rates changed when it was very hot – over 103 heat index, a level defined by NWS as "dangerous – and when the heat index was under 80. The second and third columns below reflect the number of hours that passed between calls (on average) when the heat index was below 80 degrees or above 103 degrees.

For example, when the heat index was above 103, there was a call for cardiac arrest every 2.94 hours. When the heat index was under 80, the calls happened less frequently, every 5.3 hours. That meant there were 80 percent more calls per day in very hot weather for cardiac arrest. There were 70 percent more calls for congestive heart failure. There were 345x more calls for heat exhaustion in hot weather and 19x more calls for dehydration.

Supporting code and output [cq]

# Select conditions
conditions <- c("Heat Exhaustion/Heat Stroke", "Dehydration","Respiratory Distress", "COPD (Emphysema/Chronic Bronchitis)", "End Stage Renal Disease", "Diabetic Hyperglycemia", "Diabetic Hypoglycemia", "Cardiac Arrest", "CHF (Congestive Heart Failure)")

# Calculate the total number of hours over the course of Summer 2018 that the heat index fell into each heat index level, as defined by the national weather service: not unsafe (under 80), caution (80-89), extreme caution (90-102), danger (103-124).   

heat_index_count_per_nws_five_scale_bucket <- dmh_ems %>%
  select(heat_index_nws_five_scale_bucket) %>%
  group_by(heat_index_nws_five_scale_bucket) %>%
  summarise(heat_index_count_per_nws_five_scale_bucket=n()) %>%
  arrange(heat_index_nws_five_scale_bucket)

# For each target condition, calculate the number of hours between calls at each temperature level.  This metric allows us to account for the fact that simply counting calls in each bucket would be flawed, because it wouldn't adjust for the rarity of very hot temperatures.

EMS_all %>%
  filter(primary_impression_group %in% conditions) %>%
  group_by(primary_impression_group, adjusted_heat_index_nws_five_scale_bucket) %>%
  summarise(condition_calls_count_per_bucket=n()) %>%
  inner_join(heat_index_count_per_nws_five_scale_bucket, by = c("adjusted_heat_index_nws_five_scale_bucket" = "heat_index_nws_five_scale_bucket")) %>%
  mutate(hours_per_call = heat_index_count_per_nws_five_scale_bucket/condition_calls_count_per_bucket) %>%
  select(primary_impression_group, adjusted_heat_index_nws_five_scale_bucket, hours_per_call) %>%
  tidyr::spread(adjusted_heat_index_nws_five_scale_bucket, hours_per_call) %>%
  select(primary_impression_group, `not_unsafe_under_80`,`danger_103_124`) %>%
  mutate(`calls_per_day_under_80` = 24/`not_unsafe_under_80`) %>%
  mutate(`calls_per_day_over_103` = 24/`danger_103_124`) %>%
  mutate(difference_day_percent = ((`calls_per_day_over_103`-`calls_per_day_under_80`)/`calls_per_day_under_80`))

Explanation [cq]

In Baltimore’s “community statistical areas”, we examined the relationship between heat (mean afternoon temperature in our urban heat island data) and poverty. There were moderate correlations between poverty and heat (r =.4).

Supporting code and output [cq]

# Correlation coefficients
csa_tree_temp_demographics %>%
  select_if(is.numeric) %>%
  as.matrix() %>%
  correlate() %>%
  focus(matches("temp_")) %>%
  mutate(variable=rowname) %>%
  select(variable, temp_mean_aft) %>%
  filter(str_detect(variable, "family_households"))

Explanation [cq]

In Baltimore “community statistical areas”, there is a moderate negative correlation between an area’s tree canopy cover and the area’s poverty rate, with a correlation coefficient (r) of -.34. Generally, the poorer the area, the fewer trees it will have, and vice versa.

Supporting code and output [cq]

 csa_tree_temp_demographics %>%
  select_if(is.numeric) %>%
  as.matrix() %>%
  correlate() %>%
  focus(matches("15_lid_mean")) %>%
  mutate(variable=rowname) %>%
  select(variable, `15_lid_mean`) %>%
  filter(variable=="percent_of_family_households_living_below_the_poverty_line")

Explanation [cq]

In Baltimore “community statistical areas”, there is a moderate positive correlation between an area’s violent crime rate and the area’s poverty rate, with a correlation coefficient (r) of .43, where 1 would reference a perfect positive correlation. Generally, the poorer the area, the more violent crime, and vice versa.

Supporting code and output [cq]

 csa_tree_temp_demographics %>%
  select_if(is.numeric) %>%
  as.matrix() %>%
  correlate() %>%
  focus(matches("percent_of_family_households_living_below_the_poverty_line")) %>%
  mutate(variable=rowname) %>%
  select(variable, percent_of_family_households_living_below_the_poverty_line) %>%
  filter(str_detect(variable, "violent")) %>%
  rename(poverty_rate = percent_of_family_households_living_below_the_poverty_line)

Explanation [cq]

In the homes referenced in this story where we placed sensors, the highest heat index reading we captured was 119 degrees, and the highest temperature reading was 96.8 degrees.

Supporting code and output [cq]

# bind together four sensor data sets
all_sensors_day_minute_averages <- bind_rows(michael_day_minute_averages, tammy_day_minute_averages, stephanie_day_minute_averages, audrey_day_minute_averages)

# return the highest temperature
all_sensors_day_minute_averages %>%
  filter(mean_indoor_temperature == max(mean_indoor_temperature)) %>%
  select(date_hour_minute, mean_indoor_temperature)

# return the highest heat index
all_sensors_day_minute_averages %>%
  filter(mean_indoor_heat_index == max(mean_indoor_heat_index)) %>%
  select(date_hour_minute, mean_indoor_heat_index)

Explanation [cq]

The outdoor heat, humidity and heat index were calculated using data from the Inner Harbor NWS monitoring station. Sensors we placed in their homes were used to calculate the indoor temperatures, heat index and humidity.

Supporting code and output [cq]

# temperature and heat index at 4 p.m. Saturday, July 20, 2019
dmh %>%
 filter(month == 7,
         year == 2019,
         day == 20,
         hour == 16) %>%
  select(date, hour, avg_hourly_temperature_dmh, avg_hourly_heat_index_dmh)

# Michael and Alberta
michael_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-20") %>%
  filter(hour == 16) %>%
  select(date, hour, mean_indoor_temperature, mean_indoor_relative_humidity, mean_indoor_heat_index)

# Audrey
audrey_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-20") %>%
  filter(hour == 16) %>%
  select(date, hour, mean_indoor_temperature, mean_indoor_relative_humidity, mean_indoor_heat_index)

# Stephanie
stephanie_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-20") %>%
  filter(hour == 16) %>%
  select(date, hour, mean_indoor_temperature, mean_indoor_relative_humidity, mean_indoor_heat_index)

Explanation [cq]

The following code shows the values presented in the sentence above.

Supporting code and output [cq]

tammy_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-21") %>%
  filter(hour == 22) %>%
  select(date_hour, mean_indoor_heat_index, mean_outdoor_heat_index, indoor_heat_index_difference)

Explanation [cq]

Between July 16 and July 23, the heat index inside of Pingley’s house was higher than the outside for more hours (108) than hours when the opposite was true (60).

Supporting code and output [cq]

stephanie_day_hourly_averages %>%
  filter(date_hour >= "2019-07-16",
         date_hour < "2019-07-23") %>%
  mutate(difference = case_when(
      indoor_heat_index_difference > 0 ~ "hours hotter inside",
      indoor_heat_index_difference <= 0 ~ "hours hotter outside"
  )) %>%
  group_by(difference) %>%
  summarise(count=n())

Explanation [cq]

The average temperature between July 16 and July 22 was 94.2 degrees, hotter than any period of the summer so far.

Supporting code and output [cq]

dmh %>%
  filter(year == 2019) %>%
  filter(date >= "2019-06-21",
         date <= "2019-09-21")

# compute means for the summer
summer_means <- dmh %>%
  filter(year == 2019) %>%
  filter(date >= "2019-06-21",
         date <= "2019-09-21") %>%
  group_by(`date`) %>%
  summarise(min_temp = min(avg_hourly_temperature_dmh),
            max_temp = max(avg_hourly_temperature_dmh),
            mean_temp = mean(avg_hourly_temperature_dmh),
            min_heat_index = min(avg_hourly_heat_index_dmh),
            max_heat_index = max(avg_hourly_heat_index_dmh),
            mean_heat_index = mean(avg_hourly_heat_index_dmh)
  )

# calculate mean temperature for seven days prior in each date in data
running_average <- as_tibble(runMean(summer_means$mean_heat_index, n=7))

# bind moving average back to summer means
summer_means <- bind_cols(summer_means, running_average) %>%
  select(date, value) %>%
  rename(seven_day_prior_mean_temp = value) %>%
  arrange(desc(seven_day_prior_mean_temp))

# display summer means
summer_means

Explanation [cq]

The heat index reached 111 degrees at 4 p.m. July 21.

Supporting code and output [cq]

dmh %>%
  filter(date == "2019-07-21") %>%
  filter(hour == 16) %>%
  select(date, hour, avg_hourly_heat_index_dmh)

Explanation [cq]

The heat index reached 113 degrees at 8 p.m. on July 19. The mean indoor heat index was 98 degrees during that period.

Supporting code and output [cq]

# Stephanie heat index on July 19
stephanie_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-19") %>%
  filter(hour == 20) %>%
  select(date_hour, mean_indoor_heat_index)

# Average indoor heat index during seven day period
stephanie_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  filter(date >= "2019-07-16") %>%
  filter(date <= "2019-07-22") %>%
  summarise(mean_indoor_heat_index = mean(mean_indoor_heat_index))

Explanation [cq]

The lowest average hourly heat index in the bedroom during that period was 88.8 degrees, or 89 rounded.

Supporting code and output [cq]

stephanie_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  filter(date >= "2019-07-16") %>%
  filter(date <= "2019-07-22") %>%
  summarise(min_indoor_heat_index = min(mean_indoor_heat_index))

Explanation [cq]

During that week period, there were 108 hours when the heat index inside was higher than the outside heat index, and only 60 hours when the opposite was true.

Supporting code and output [cq]

stephanie_day_hourly_averages %>%
  filter(date_hour >= "2019-07-16",
         date_hour < "2019-07-23") %>%
  mutate(difference = case_when(
      indoor_heat_index_difference > 0 ~ "hours hotter inside",
      indoor_heat_index_difference <= 0 ~ "hours hotter outside"
  )) %>%
  group_by(difference) %>%
  summarise(count=n())

Explanation [cq]

The heat index in Baltimore hit 102 degrees at 8 p.m. on July 19.

Supporting code and output [cq]

dmh %>%
  filter(date == "2019-07-19") %>%
  filter(hour == 20) %>%
  select(date, hour, avg_hourly_heat_index_dmh)

Explanation [cq]

At 8 p.m. on July 19, the heat index indoors was 112.7 degrees inside, and the outdoor heat index was 102 degrees, with a difference of 10.7 degrees.

Supporting code and output [cq]

stephanie_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-19") %>%
  filter(hour ==  20) %>%
  select(date_hour, mean_indoor_heat_index, mean_outdoor_heat_index,indoor_heat_index_difference)

Explanation [cq]

At 4 a.m. July 20, the heat index in Baltimore was still 89 degrees.

Supporting code and output [cq]

dmh %>%
  filter(date == "2019-07-20") %>%
  filter(hour == 4) %>%
  select(date, hour, avg_hourly_heat_index_dmh)

Explanation [cq]

The heat index in Pingley’s house at 4 a.m. July 20 was 99.3 degrees.

Supporting code and output [cq]

stephanie_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-20") %>%
  filter(hour ==  4) %>%
  select(date_hour, mean_indoor_heat_index)

Explanation [cq]

On Saturday, July 20, the heat index in their apartment hit 111.8 degrees.

Supporting code and output [cq]

michael_day_hourly_averages %>%
  mutate(date = date(date_hour)) %>%
  mutate(hour = hour(date_hour)) %>%
  filter(date == "2019-07-20") %>%
  arrange(desc(mean_indoor_heat_index)) %>%
  select(date_hour, mean_indoor_heat_index)

Fact: Historical heat index days

“Historically, the Baltimore area has averaged about six days a year when the heat index exceeded 100 degrees, according to new research from the Union of Concerned Scientists and the University of Idaho. If no action is taken to reduce carbon emissions, by mid-century that figure will rise to more than 37 days annually, according to the researchers. The study defines mid-century as starting in 17 years. By the end of the century, as a baby born today becomes a senior citizen, there will be 65 days with a heat index of 100 degrees or higher, the researchers projected. That’s about the same number as McAllen, Texas, a city that abuts the Mexican border.”

Explanation [cq]

The Union of Concerned Scientists [Killer Heat in the United States] study released this summer included detailed tables projecting the number of days of 100+ heat index days for most U.S. cities, and projected how many additional days of 100+ heat index days there would be by mid-century. They found there were six days of 100+ heat index days per year historically in Baltimore, and by mid-century there would be 37 days. By end of the century, there would 65, about as many as McAllen, Texas has today (67). Note: our own analysis of historical Baltimore heat index data at a NWS Inner Harbor monitoring station found that the 6 days per year figure was likely a conservative estimate.

Supporting code and output [cq]

heat_index_projections %>%
  select(state, city, historical_100_plus, midcentury_no_action_100_plus, endcentury_no_action_100_plus) %>%
  filter((state == "MD" | state == "TX"), (city == "Baltimore" | city == "McAllen"))

Explanation [cq]

The block in the photo is on North Milton and East Monument at the edge of the McElderry Park neighborhood. The street shown is split between two U.S. Census blocks, with IDs of “245100702004002” and “245100702004002”. They had 7.6 percent and 8.7 percent tree canopy cover in 2015.

Supporting code [cq]

blocks_tree_temp_demographics %>%
  filter(geoid10 == "245100702004002" | geoid10 == "245100702005005") %>%
  select(geoid10, `15_lid_mean`)

Explanation [cq]

The code below shows the percent of tree canopy cover in more than a dozen east Baltimore neighborhoods in 2015. The range is from 4 to 14.

Supporting code and output [cq]

east_baltimore_nsas <- c("Berea", "Broadway East", "Oliver", "Middle East",
                 "Biddle Street","Milton-Montford", "Madison-Eastend",
                 "CARE", "McElderry Park", "Ellwood Park/Monument",
                 "Patterson Place", "Patterson Park Neighborhood",
                 "Baltimore Highlands", "Highlandtown",
                 "Upper Fells Point") %>%
  lapply(tolower)

nsa_tree_temp %>%
  filter(nsa_name %in% east_baltimore_nsas) %>%
  select(nsa_name, `15_lid_mean`) %>%
  arrange(`15_lid_mean`) %>%
  mutate(`15_lid_mean_percentage` = 100*(`15_lid_mean`)) %>%
  select(nsa_name, `15_lid_mean_percentage`)