In shaughnessyar/driftR: Drift Correcting Water Quality Data

Overview

A number of functions in driftR, including dr_factor(), dr_replace(), and dr_drop, rely on date and time information supplied by the instrument. driftR anticipates that these two pieces of information will be stored in character vectors in predictable formats that can be easily parsed. Under the hood, we use the lubridate package's parse_date_time() function to automatically detect and properly format date and time inputs supplied by the user and their instrument.

Working with Dates in driftR

We currently support two date formats in our functions - the MDY (or "month day year") format and the YMD (or "year month day") format.

The following are examples of MDY that lubridate can parse:

Jan 1, 2018
January 1, 2018
1/1/18
01/01/18
1/1/2018
01/01/2018
1-1-18
01-01-18
1-1-2018
01-01-2018

The following are examples of YMD that lubridate can parse:

2018/1/1
2018/01/01
2018-1-1
2018-01-01

To prevent confusion, we do not support the DMY format. If you have data that is DMY formatted, as in 15/01/2018, you can convert it to YMD formatted character using the following sample syntax:

x <- "15/01/2018"
y <- as.character(lubridate::dmy(x))

Once you have a properly formatted character vector, you can use it with any of the driftR functions that require a date!

Working with Times in driftR

Times in Your Data

We currently support times that include the hours, minutes, and seconds of a particular measurement. These should be formatted using a 24 hour clock (e.g. 00:05:05 for five minutes after midnight). If your data are formatted ysing a 12 hour clock (e.g. 12:05:05 AM), the following code can be used to convert them to a 24 hour clock

library(dplyr)

data <- data %>%
    mutate(time = format(strptime(time, "%I:%M %p"), format="%H:%M:%S"))

Times in `dr_replace` and `dr_drop`

Both dr_replace() and dr_drop() allow you to specify ranges of times to replace or drop values for. We recommend using a 24 hour clock here as well. If you are not used to using a 24 hour clock, you can find a conversion table here. There is one key difference with these two functions that is worth noting - when you specify a time for each, you do not need to specify seconds but may optionally choose to do so. For example, both 12:05 and 12:05:00 are valid inputs for both dr_replace() and dr_drop().

Time Zones

The parse_date_time() function from lubridate relies on timezone information. By default, driftR functions use your computer's timezone setting. You can see this by using the following base function in your console:

>  Sys.timezone()
[1] "America/Chicago"

What is critical about this setting is that the timezone must match the timezone where your data were collected. If you are using data collected in a timezone that is different from the timezone your computer is set to, you need to specify the correct timezone for your data in dr_factor(), dr_replace(), and dr_drop using the tz argument. Include the string name of the appropriate timezone as the input for this argument. You can get a full list of acceptable timezone strings from the OlsonNames() function, which has 592 possible strings:

> OlsonNames()
  [1] "Africa/Abidjan"  "Africa/Accra"  "Africa/Addis_Ababa"   "Africa/Algiers"  
  [5] "Africa/Asmara"  "Africa/Asmera"  "Africa/Bamako"  "Africa/Bangui"  
# output truncated

Copy the appropriate string into your driftR function from this output.

Splitting a Combined Date and Time Variable

If your data arrive with the date and time combined, you can use the following functions to (1) convert it from a string to a properly formatted date and time value, and then (2) split it into separate date and time values:

library(lubridate)

input <- "1/12/18 12:05:00"

parsed <- parse_date_time(inpout, orders = "mdy HMS")

date <- as.character(as_date(parsed))
time <- format(parsed, "%H:%M:%S")