README.md

In tbradley1013/dwqr: **D**rinking **W**ater **Q**uality **R**

dwqr

Package to handle Drinking Water Quality issues in R

This package is still in development

This package will hold functions that will allow for calculation of different statistics for drinking water quality in drinking water distribution systems. Some of the calculations that this package will hope to address are:

• Nitrification Action Levels
• DBP OEL and LRAA values (currently not implemented)

Installation

To install this package use:

# if you do not already have devtools installed run:
# install.packages("devtools")

devtools::install_github("tbradley1013/dwqr")

library(dwqr)
library(tidyverse)

## -- Attaching packages ----------------------------------------------------------------------------------------------- tidyverse 1.2.1 --

## v ggplot2 3.1.0     v purrr   0.2.5
## v tibble  1.4.2     v dplyr   0.7.7
## v tidyr   0.8.2     v stringr 1.3.1
## v readr   1.1.1     v forcats 0.3.0

## -- Conflicts -------------------------------------------------------------------------------------------------- tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

Nitrification Action Levels

Nitrification is a common problem for many utilites that use chloramines as a secondary disinfectant. As chloramines decay, free ammonia is released. This free ammonia acts as food for ammonia and nitrite oxidizing bacteria (AOB and NOB, respectively). As the population of AOB and NOB increase, the problem is exacerbated causing chloramine to decay faster, releasing more free ammonia. The increased decay of chloramines is a major issue for water utilities as disinfectant is the main prevention of harmful bacteria from entering the water system. As a result, utitlies must establish strategies for dealing with nitrification in their distribution systems. In some states, such a Pennsylvania, regulators require a set action plan to be created in the form of a nitrification control plan. AWWA provides recommendations for dealing with nitrification and creating in their [M56 Manual: Nitrification Prevention and Control in Drinking Water]. One of the major aspects of a nitrification control is determing necessary operational responses to different levels of potential nitrification.

But how should utilities go about deciding what these action levels should be? That is where dwqr comes in. dwqr provides functions in R that will allow for utilites to determine system or site specific action levels based on chloramine residual. Since the decay of chloramine increases as nitrification begins to occur, it is a typical indicator of when nitrification could be occurring. nitrification_al uses chloramine residual values to calculate action levels for nitrification control in one of two ways.

Falling Limb method

This method uses a novel approach to calculate the “falling limb” of the chloramine curve. The falling limb of the curve can be visualized as follows:

Once the falling limb of your data has been isolated, percentiles of the values that are classified as being a part of the falling limb are used as the action levels. By default the percentiles of the falling limb used are 80% (Action Level 1), 50% (Action Level 2), and 10% (Action Level 3). You can calculate them as shown below on an example dataset nitrification (which is included in this package):

nitrification %>% 
  filter(site == "Site 1") %>% 
  nitrification_al(date_col = date, value_col = chlorine, method = "FL", rolling_window = 7)

## # A tibble: 1 x 3
##   `Action Level 1` `Action Level 2` `Action Level 3`
##              <dbl>            <dbl>            <dbl>
## 1             1.24             0.92            0.435

You may notice that the argument rolling_window was set to 7. This argument helps to determine how many data points should be used to calculate the rolling mean of the chloramine residual and of its first and second derivative. This is done to prevent oversensitivity to slight changes in chloramine residual but to better capture trends in the data. By default, this argument is set to 8. To know how to best set rolling_window, you can visualize how the data is classified by using the plot_fl function.

nitrification %>% 
  filter(site == "Site 1") %>% 
  plot_fl(date_col = date, value_col = chlorine, rolling_window = 7,
          ylab = "Total Chlorine Residual (mg/L)") + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))

You can see how the classifications change when the window size is changed

nitrification %>% 
  filter(site == "Site 1") %>% 
  plot_fl(date_col = date, value_col = chlorine, rolling_window = 12, 
          ylab = "Total Chlorine Residual (mg/L)",)  + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))

In general, datasets that have more frequent data points will benefit from larger rolling windows while datasets with less frequent data points will benefit from smaller windows.

You can assess how well these action levels categorize a given dataset by using the plot_al function.

nitrification %>% 
  filter(site == "Site 1") %>% 
  plot_al(date_col = date, value_col = chlorine, 
          ylab = "Total Chlorine Residual (mg/L)", rolling_window = 7) + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))

General Percentiles

The second method that can be used to create action levels is general percentiles. In this case, rather than taking percentiles of a subset of the data (as in the falling limb mehtod), percentiles are taken of the overall data. By default, the percentiles selected are the same as the ones used in the falling limb method (80%, 50%, and 10%).

This option is done by setting method = "P".

nitrification %>% 
  filter(site == "Site 1") %>% 
  nitrification_al(date_col = date, value_col = chlorine, method = "P",
                   ylab = "Total Chlorine Residual (mg/L)",)

## # A tibble: 1 x 4
##   ylab                  `Action Level 1` `Action Level 2` `Action Level 3`
##   <chr>                            <dbl>            <dbl>            <dbl>
## 1 Total Chlorine Resid~             1.35             0.81              0.4

Similar to the falling lumb method, you can plot the classification of data points to see how well these action levels classify the data by using the plot_al function.

nitrification %>% 
  filter(site == "Site 1") %>% 
  plot_al(date_col = date, value_col = chlorine, method = "P",
          ylab = "Total Chlorine Residual (mg/L)") + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))

## Warning: Removed 2 rows containing missing values (geom_point).

Mutliple Sites

dwqr’s nitrification functions allow for multiple sites to be passed to them so that site specific action levels can be generated. This is done by passing all grouping columns as unnmaed arguments after the date_col and value_col columns. In the case of our nitrification dataset, this would be the site column. You may have noticed that I was using a filter statement in the code above to only include data for “Site 1”. Now let’s try the same thing as before but set the site column as a grouping variable.

nitrification %>% 
  nitrification_al(date_col = date, value_col = chlorine, site)

## # A tibble: 2 x 4
##   site   `Action Level 1` `Action Level 2` `Action Level 3`
##   <chr>             <dbl>            <dbl>            <dbl>
## 1 Site 1             1.08             1.05             1.05
## 2 Site 2             1.31             0.9              0.3

As we can see, there is now action levels generated for each of the sites in the data set! If a grouping variable is passed to one of the plotting functions, the groups are separated by facet_wrap in the ggplot2 package.

nitrification %>% 
  plot_fl(date_col = date, value_col = chlorine, site,
          ylab = "Total Chlorine Residual (mg/L)",
          ncol = 1) + 
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1))

Plotting the first derivative and nitrite

The plot_fl function also allowd for users to plot the first derivative of the chlorine trend (a key component in the calculation of the falling limb) and any associated nitrite data. In order to include the first derivative plot, you simply must set include_first = TRUE.

nitrification %>% 
  filter(site == "Site 1") %>% 
  plot_fl(
    date_col = date, value_col = chlorine, 
    rolling_window = 7, include_first = TRUE,
    ylab = "Total Chlorine Residual (mg/L)",
    theme = theme(
      axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1)
    )
  )

## Warning: Removed 6 rows containing missing values (geom_path).

NOTE: In this past plot, the theme argument was used inside of plot_fl. When either the first derivative or nitrite plots are included with your falling limb plot, you must specify all ggplot2 theme elements with the theme argument. This will apply your desired themes to all of the plots. Otherwise, it will only apply it to the last plot.

To plot nitrite data with the falling limb classification, the nitrite data must be in a separate column with corresponding sample points lined up. You can use the nitrification data set as an example.

nitrification

## # A tibble: 536 x 4
##    site   date       chlorine nitrite
##    <chr>  <date>        <dbl>   <dbl>
##  1 Site 1 2013-10-02     0.66   0.195
##  2 Site 1 2013-10-09     0.54   0.198
##  3 Site 1 2013-10-16     0.7    0.189
##  4 Site 1 2013-10-23     0.68   0.211
##  5 Site 1 2013-10-30     0.52   0.181
##  6 Site 1 2013-11-06     0.65   0.133
##  7 Site 1 2013-11-13     0.88   0.105
##  8 Site 1 2013-11-20     0.79   0.102
##  9 Site 1 2013-11-27     1.09   0.064
## 10 Site 1 2013-12-04     0.88   0.058
## # ... with 526 more rows

Plotting the two together would look like this.

nitrification %>% 
  filter(site == "Site 1") %>% 
  plot_fl(
    date_col = date, value_col = chlorine, 
    rolling_window = 7, nitrite_col = nitrite,
    ylab = "Total Chlorine Residual (mg/L)",
    nitrite_ylab = "Nitrite (mg N/L)",
    theme = theme(
      axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1)
    )
  )

## Warning: Removed 1466 rows containing missing values (geom_point).

A quick note of tidy eval

This packages uses tidy eval via the rlang package. This is why the column names of the date_col and value_col are passed to the nitrification functions as unquoted variable names rather than strings. This is also what allows for the grouping variables to be set dynamically by the user. It is important to keep this in mind as you use this package as passing a quoted column name will result in an error.

tbradley1013/dwqr documentation built on Feb. 11, 2024, 10:03 a.m.