Degree Days

Computing Degree Days from the Daily Min & Max Temp

1) How do you mathematically model temperature over the course of the day?

Researchers have explored several options:

2) How you do you find the area under the curve, that falls within the range of usuable temperatures for a given species?

Phenology Tables

Degree days are not very useful by themselves. You need to use them with a phenology table that predicts when events will take place based on accumulated degree days.

Phenology tables also tell you when to start counting degree days (i.e. biofix).

The biofix could be a calendar date, or the date of an observation such as when you see eggs in your traps.

Chill Hours & Chill Portions

R-tricks: Date Parts

Analyzing multi-year data often requires using date parts (e.g., month, year, Julian day), for filtering and grouping.

The lubridate package has functions you can use to define date parts. Examples:

(x <- as.Date("2022-09-05"))

## [1] "2022-09-05"

lubridate::day(x)

lubridate::month(x)

lubridate::year(x)

lubridate::yday(x)  ## 'Julian' date

## [1] 5
## [1] 9
## [1] 2022
## [1] 248

R-tricks: Converting Units

Weather data often requires you to convert units.

The units package comes with a database of conversion constants so you don’t have to look them up.

Many R packages (including API clients) are returning values as ‘units’ objects (numeric values with units attached).

Example:

library(units)
boiling_int <- 212                            ## plain number
boiling_f <- set_units(boiling_int, degF)     ## a 'units' object
boiling_f

## 212 [degF]

boiling_c <- set_units(boiling_f, degC)       ## convert to degrees Celsius by setting different units
boiling_c

## 100 [°C]

R-tricks: Grouping Rows of Data

For multi-year or multi-site data, you often need to work with groups of rows.

To work with groups in dplyr, we add functions that:

1. Define groups

group_by()

The argument(s) to group_by() should be column(s) or expression(s) that defines the row groups.

Each unique value will become a group of rows.

Example:

storms |>  
  group_by(year) |>    ## group the rows by year
  head()

## # A tibble: 6 × 13
## # Groups:   year [1]
##   name   year month   day  hour   lat  long status categ…¹  wind press…² tropi…³
##   <chr> <dbl> <dbl> <int> <dbl> <dbl> <dbl> <chr>  <ord>   <int>   <int>   <int>
## 1 Amy    1975     6    27     0  27.5 -79   tropi… -1         25    1013      NA
## 2 Amy    1975     6    27     6  28.5 -79   tropi… -1         25    1013      NA
## 3 Amy    1975     6    27    12  29.5 -79   tropi… -1         25    1013      NA
## 4 Amy    1975     6    27    18  30.5 -79   tropi… -1         25    1013      NA
## 5 Amy    1975     6    28     0  31.5 -78.8 tropi… -1         25    1012      NA
## 6 Amy    1975     6    28     6  32.4 -78.7 tropi… -1         25    1012      NA
## # … with 1 more variable: hurricane_force_diameter <int>, and abbreviated
## #   variable names ¹​category, ²​pressure, ³​tropicalstorm_force_diameter

Note - nothing looks different! This is because the groups are defined invisibly.

To do something productive with the groups, we have to add something else…

2a. Compute Metrics for Each Group Separately

group_by() + mutate()

When working with cumulative metrics (i.e., chill portions) for multiple years of data, you may want to calculate the metric for groups of rows. You can compute metrics for groups of rows by following group_by() with mutate().

Example:

We make a fake time series dataset consisting of a random number 0-10 each day for a year:

val_tbl <- tibble(dt = seq(from = as.Date("2010-01-01"), 
                               to = as.Date("2010-06-30"),
                               by = 1)) |>
  mutate(val = runif(n(), min = 0, max = 10))

val_tbl |> head()

## # A tibble: 6 × 2
##   dt           val
##   <date>     <dbl>
## 1 2010-01-01  1.15
## 2 2010-01-02  4.58
## 3 2010-01-03  4.46
## 4 2010-01-04  5.44
## 5 2010-01-05  1.10
## 6 2010-01-06  6.11

Next we compute the cumulative sum of val for each month (note how month is computed on-the-fly by group_by()):

val_cumsum_bymonth_tbl <- val_tbl |> 
  group_by(month = lubridate::month(dt)) |> 
  mutate(cumsum_val = cumsum(val))

val_cumsum_bymonth_tbl |> head()

## # A tibble: 6 × 4
## # Groups:   month [1]
##   dt           val month cumsum_val
##   <date>     <dbl> <dbl>      <dbl>
## 1 2010-01-01  1.15     1       1.15
## 2 2010-01-02  4.58     1       5.73
## 3 2010-01-03  4.46     1      10.2 
## 4 2010-01-04  5.44     1      15.6 
## 5 2010-01-05  1.10     1      16.7 
## 6 2010-01-06  6.11     1      22.8

Plot it to see the pattern:

ggplot(val_cumsum_bymonth_tbl, aes(x = dt, y = cumsum_val)) + geom_line()

2b. Compute Summaries for each Group

group_by() + summarize()

Commonly used in analysis,summarize() will create column(s) of summary statistics for each group of rows.

To define the summaries, you can use any aggregate function that takes a vector of values and returns a single value. Examples:

Example: How many category 4 and higher storms were there each month?

library(dplyr)

storms |>  
  select(name, month, category, hurricane_force_diameter) |>         ## select the columns we need
  filter(category >= 4) |>                                           ## filter those rows where category >= 4
  group_by(month) |>                                                 ## group the rows by month
  summarize(num_hurricanes = n(),                                    ## for each group, report the count
            avg_diam = mean(hurricane_force_diameter, na.rm=TRUE))

## # A tibble: 5 × 3
##   month num_hurricanes avg_diam
##   <dbl>          <int>    <dbl>
## 1     7             12     63.6
## 2     8             95     92.9
## 3     9            275     86.7
## 4    10             91     78.2
## 5    11             24     51.9

Notebook 2: Cummulative metrics and multi-year summaries

Or see the Completed Notebook #2.

End