Grokbase Groups R r-help August 2012
I am currently conducting a performance analysis for an urban stormwater
management facility which employs Low Impact Development practices. I have
the Event Mean Concentrations (EMCs) of several water quality constituents
(ex: Total Phosphorus (TP), Total Oxidized Nitrogen (OxN), etc) at inflow
and outflow locations for a particular stormwater management practice. I
wish to use the Effluent Probability Method to create a normal probability
plot comparing the log-transform (most often) data of both inflow and
outflow Event Mean Concentrations (EMCs) as a method to visually compare the
water quality constituent EMCs at the inflow and outflow locations across
all storm events. Due to the censored nature of environmental water quality
data I am employing a probability plot function which uses a robust
regression on ordered statistics (ROS) method to determine the placement of
uncensored data in the probability plot. This robust ROS method takes the
proportion of data which is censored, or below the reporting limit, into
account when determining the position of the uncensored data in the
probability plot.

The plot(cenros(Observation, ObservationCensored)) command creates a
censored lognormal ROS probability plot for the data. Where
Observation=the raw data
Where: TRUE=a censored observation
FALSE=an uncensored observation

I have included the data and commands I used in R at the end of this
message. If wanted, they can be pasted directly into R to see the result (as
long as you have the NADA package installed).

If you have copied the below input into R, then you will see two censored
lognormal ROS probability plots overlaid onto a single graph. However you
will also notice that the x- and y- scales for the axes do not line up
properly. The following list describes how I have attempted to set the x-
and y-limits to the same scale for the two probability plots:
? I tried setting the x- and y-limits using xlim=c() and ylim=c()
respectively in each plot() command however an error occurs because ?formal
argument ?xlim? and ?ylim? matched by multiple actual arguments?.
? I also tried setting the second plot?s axesúLSE within second plot()
command however it did not seem to have an effect.
? I have tried setting usr=c(x1,x2,y1,y2) within the par() command, again it
didn?t have an effect.
? I tried setting xaxs= ?d? in the par() command however ?d? is
unimplemented in R and thus an error occurred. Same for yaxs=?d?.
? I tried setting xaxt= ?n? in the second graph to suppress plotting the
axis however this only suppressed printing the values of the bottom x-axis
and thus didn?t work for what I wanted.
? Long story short, I have been unable to graph two probability plots on the
same plot (with matching x-and y-axis).

I would appreciate any help in how to make the x- and y-axes the same by
setting the xlim and ylim modifiers equal in both probability plots and/or
by using modifiers within the par(new=TRUE) command.

Thank you for your time and consideration,


Below are the data and commands I used to create the two censored lognormal
ROS probability plots overlaid onto a single graph for use in the Effluent
Probability Method. If wanted, they can be pasted directly into R to see the
result (as long as you have the NADA package installed).




Search Discussions

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupr-help @
postedAug 8, '12 at 6:21p
activeAug 8, '12 at 6:21p

1 user in discussion

Plebel: 1 post



site design / logo © 2022 Grokbase