4 min read

Temperature change in Czech Republic - part II

2018-12-18 
library(tidyverse)
library(broom)
library(lubridate)
library(scales)
df_prec_raw <- readRDS("../../static/data/chmu_srazky.rds")
df_temp_raw <- readRDS("../../static/data/chmu_teploty.rds")
df_prec_raw
## # A tibble: 9,744 x 5
##    year_month          year  month region              precipitation
##    <dttm>              <chr> <chr> <chr>                       <dbl>
##  1 1961-01-01 00:00:00 1961  Jan   Česká republika                22
##  2 1961-01-01 00:00:00 1961  Jan   Praha a Středočeský            15
##  3 1961-01-01 00:00:00 1961  Jan   Jihočeský                      14
##  4 1961-01-01 00:00:00 1961  Jan   Plzeňský                       22
##  5 1961-01-01 00:00:00 1961  Jan   Karlovarský                    39
##  6 1961-01-01 00:00:00 1961  Jan   Ústecký                        23
##  7 1961-01-01 00:00:00 1961  Jan   Liberecký                      41
##  8 1961-01-01 00:00:00 1961  Jan   Královéhradecký                38
##  9 1961-01-01 00:00:00 1961  Jan   Pardubický                     21
## 10 1961-01-01 00:00:00 1961  Jan   Vysočina                       16
## # ... with 9,734 more rows
df_temp_raw
## # A tibble: 9,744 x 5
##    year_month          year  month region              temperature
##    <dttm>              <chr> <chr> <chr>                     <dbl>
##  1 1961-01-01 00:00:00 1961  Jan   Česká republika            -3.4
##  2 1961-01-01 00:00:00 1961  Jan   Praha a Středočeský        -2.8
##  3 1961-01-01 00:00:00 1961  Jan   Jihočeský                  -3.9
##  4 1961-01-01 00:00:00 1961  Jan   Plzeňský                   -3.2
##  5 1961-01-01 00:00:00 1961  Jan   Karlovarský                -3.4
##  6 1961-01-01 00:00:00 1961  Jan   Ústecký                    -3.4
##  7 1961-01-01 00:00:00 1961  Jan   Liberecký                  -3.7
##  8 1961-01-01 00:00:00 1961  Jan   Královéhradecký            -3.3
##  9 1961-01-01 00:00:00 1961  Jan   Pardubický                 -3.4
## 10 1961-01-01 00:00:00 1961  Jan   Vysočina                   -3.9
## # ... with 9,734 more rows

Merge both dataset together. Check if relevant columns are identical first.

map2(df_prec_raw, df_temp_raw, identical)
## $year_month
## [1] TRUE
## 
## $year
## [1] TRUE
## 
## $month
## [1] TRUE
## 
## $region
## [1] TRUE
## 
## $precipitation
## [1] FALSE
df <- bind_cols(df_prec_raw, df_temp_raw) %>% 
  select(year_month, year, month, region, temperature, precipitation) %>% 
  gather(key, value, -year_month, -year, -month, -region) %>% 
  mutate(season = case_when(
    str_detect(month, "Dec|Jan|Feb") ~ "Winter",
    str_detect(month, "Mar|Apr|May") ~ "Spring",
    str_detect(month, "Jun|Jul|Aug") ~ "Autumn",
    str_detect(month, "Sep|Oct|Nov") ~ "Summer"))
df
## # A tibble: 19,488 x 7
##    year_month          year  month region            key       value season
##    <dttm>              <chr> <chr> <chr>             <chr>     <dbl> <chr> 
##  1 1961-01-01 00:00:00 1961  Jan   Česká republika   temperat~  -3.4 Winter
##  2 1961-01-01 00:00:00 1961  Jan   Praha a Středoče~ temperat~  -2.8 Winter
##  3 1961-01-01 00:00:00 1961  Jan   Jihočeský         temperat~  -3.9 Winter
##  4 1961-01-01 00:00:00 1961  Jan   Plzeňský          temperat~  -3.2 Winter
##  5 1961-01-01 00:00:00 1961  Jan   Karlovarský       temperat~  -3.4 Winter
##  6 1961-01-01 00:00:00 1961  Jan   Ústecký           temperat~  -3.4 Winter
##  7 1961-01-01 00:00:00 1961  Jan   Liberecký         temperat~  -3.7 Winter
##  8 1961-01-01 00:00:00 1961  Jan   Královéhradecký   temperat~  -3.3 Winter
##  9 1961-01-01 00:00:00 1961  Jan   Pardubický        temperat~  -3.4 Winter
## 10 1961-01-01 00:00:00 1961  Jan   Vysočina          temperat~  -3.9 Winter
## # ... with 19,478 more rows

What was the question?

Claim - “The precipitation in Czech Republic decreases over last decades.”

df %>% 
  
  filter(key == "precipitation") %>% 
  filter(region == "Česká republika") %>% 
  
  ggplot(aes(year_month, value)) + 
  geom_col() + 
  geom_smooth(method = "lm") +
  scale_x_datetime(labels = date_format("%Y"), date_breaks = "1 year") + 
  theme(axis.text.x = element_text(angle = 75, hjust = 1)) + 
  labs(y = "precipitation [mm]", x = "year") +
  facet_wrap(~region, scales = "free")

Nope (on average).

Claim - “It rains less and less every summer in Czech Republic.”

df %>% 
  
  filter(key == "precipitation") %>% 
  filter(region == "Česká republika") %>% 
  
  ggplot(aes(year_month, value)) + 
  geom_col() + 
  geom_smooth(method = "lm") +
  # coord_flip() + 
  scale_x_datetime(labels = date_format("%Y"), date_breaks = "1 year") + 
  theme(axis.text.x = element_text(angle = 75, hjust = 1)) + 
  labs(y = "precipitation [mm]", x = "year") +
  facet_wrap(~season, scales = "free")

Well, seems to be quite stable, no drama here.

Claim - “The temperature in Czech Republic increases over last decades.”

df %>% 
  
  filter(key == "temperature") %>% 
  filter(region == "Česká republika") %>% 
  
  ggplot(aes(year_month, value)) + 
  geom_col() + 
  geom_smooth(method = "lm") +
  scale_x_datetime(labels = date_format("%Y"), date_breaks = "1 year") + 
  theme(axis.text.x = element_text(angle = 75, hjust = 1)) + 
  labs(y = "°C", x = "year") + 
  facet_wrap(~region)

Yes, it does increase. On average by:

df %>% 
  
  filter(key == "temperature") %>% 
  filter(region == "Česká republika") %>% 
  
  mutate(year = year %>% as.integer()) %>% 

  group_by(year) %>% 
  summarise(temp_avg_year = mean(value)) %>% 
  
  lm(temp_avg_year~year, data = .) %>%
  tidy() %>%
  pull(estimate) %>% 
  .[[2]] %>% 
  round(3)
## [1] 0.031

..°C per year.