Al Leydecker, November 18, 2008
100,000
phosphate (mg/L)
above the WWTP
below detection limit
Foster Park
10,000
0.2
1,000
100
0.1
10
r-0
8
-0
7
Ap
O
ct
r-0
7
-0
6
Ap
O
ct
r-0
6
-0
5
Ap
r-0
5
O
ct
-0
4
Ap
r-0
4
O
ct
-0
3
Ap
O
ct
r-0
3
-0
2
Ap
O
ct
r-0
2
-0
1
Ap
O
ct
Ap
O
ct
r-0
1
1
-0
0
0.0
flow at Foster Park (cfs)
0.3
I’ve prepared graphs for phosphate in the same format as I did for nitrate: in files named
Phosphate.VR.I, Phosphate.VR.II and Phosphate.VR.III. I’m not going to bother showing them.
Instead, I want to concentrate on comparing data from the Ojai Treatment Plant (OVSD) with the
Channelkeeper dataset. Unfortunately, the sanitary district only began analyzing samples for
phosphate beginning in August 2003, and this date, also unfortunately, marks the beginning of a
reduced sampling program; sampling only 3 locations: above the WWTP (R-3), just below the
outfall (R-4) and just above the Canada Larga confluence (R-5). As a result only a limited
amount of OVSD phosphate data is available. The above graph shows data from their above the
WWTP sampling point (R-3). These results should be compatible with Channelkeeper’s Foster
Park data. Unfortunately they are not; phosphate concentrations above the plant appear to be
much higher above the WWTP than they are even further above at Foster Park. The expectation
is that differences between these two sampling locations, if any, should show a decrease in
phosphate, not an increase.
There are (I need to quit saying unfortunately) a number of “non-detects” in the OVSD dataset.
I’ve assigned these samples a value of 0.009 mg/L – this being half the detection limit given in
the most recent results. Detection limits were not given for earlier results and I’ve assumed that
this value is also applicable to them. This is probably not accurate and to compensate, all nondetect samples are shown in a different color on the graph. I have also decided not to apply the
same rule to Channelkeeper results (as Emerson said “a foolish consistency being the hobgoblin
of little minds”). The LTER lab gives the phosphate and total dissolved phosphorus (TDP)
detection limit as 0.3 µM, i.e., 0.0093 mg/L. I probably should have begun assigning half this
value, 0.005 mg/L, to all phosphorus samples below 0.3 µM back in 2001. Since I didn’t, and
never have, I’ve also left the values in these graphs unchanged. However the dashed blue line
marks the < 0.01 mg/L TP limit for “good quality” waters (from the 305b SWAMP Report) (the
red line marks the > 0.1 mg/L limit defining “poor quality”) and results below this line are usually
below the stated LTER detection limit.
100,000
phosphate (mg/L)
3
Shell Road.
Stanley Drain
10,000
2
1,000
100
1
10
8
g0
Au
b08
Fe
7
g0
Au
b07
Fe
6
g0
Au
b06
Fe
5
g0
Au
b05
Fe
4
g0
Au
b04
1
Fe
Au
g0
3
0
flow at Foster Park (cfs)
C.Larga confluence
below detection limit
LTER
This graph shows phosphate data from the site just above the Canada Larga confluence (R-5).
Again, I’ve indicated any samples below the detection limit (only one since this location, below the
plant, is almost always high in phosphate). I’ve also included, from the early summer of 2008 on,
LTER analyzed samples collected at approximately the same location. All are shown against a
background of Foster Park flow, and Shell Road and Stanley Drain monthly phosphate. The Canada
Larga confluence is approximately 0.6 km below the WWTP, Shell Road 2.4 km further downriver,
and Stanley Drain another 1.5 km below. The expected pattern between the confluence and Shell
Road should be similar to that between Shell and Stanley Drain: very little difference during winter
and periods of high flow, but a noticeable decrease during low-flow, dry-season uptake.
The actual pattern, unlike that seen in the first graph, shows the expected similarity during the most
recent years, but higher than expected OVSD concentrations during the earlier period. Part of the
similarity, of course, is due to scale differences between the two graphs (roughly 10:1), but it is
reassuring that the patterns here do resemble one another. What is a little upsetting is that the
OVSD concentrations are noticeably higher – especially the earlier ones. LTER (Channelkeeper)
phosphorus concentrations show a “step” between pre- and post-2005 data (visible in Foster Park
data shown on the first graph), due, most probably, to some kind of analysis or calibration problem.
My general feeling has been that the post-2005 data is the more accurate, i.e., that pre-2005
concentrations are too high. That the OVSD concentrations are higher still is not reassuring.
However, this is not to say that part of the problem might not lie with OVSD data. Although I don’t
know for sure, it seems like at least 3 labs have been used over the years for analysis (2 since
August 2003) and these earlier samples may have been analyzed by a different lab, with different
detection limits, than the later. As a final note, as of September 2008 Channelkeeper has abandoned
sampling at Shell Road and Stanley Drain, and, instead, will sample at the confluence. So in the
future we will have a better chance of comparing phosphorus concentrations. Keep in mind,
however, that, because of variations in WWTP output, different sampling dates and times may make
exact comparisons problematic.
100,000
WWTP
Shell Road.
Foster Park
TDP (mg/L)
5
10,000
4
1,000
3
100
2
10
1
r-0
8
-0
7
Ap
O
ct
r-0
7
-0
6
Ap
r-0
6
O
ct
-0
5
Ap
O
ct
r-0
5
-0
4
Ap
O
ct
r-0
4
-0
3
Ap
r-0
3
O
ct
-0
2
Ap
O
ct
r-0
2
-0
1
Ap
O
ct
Ap
O
ct
r-0
1
1
-0
0
0
flow at Foster Park (cfs)
6
On to total dissolved phosphorus (TDP): This graph shows monthly total phosphorus (TP)
concentrations in WWTP effluent along with Channelkeeper TDP concentrations at Shell Road and
Foster Park. While WWTP output is quite variable, river flow (i.e., the amount of dilution) is the
principal determinant of downriver concentrations. I need to make two points. The first is that
there is typically very little difference between TP and TDP concentrations in river and streamflow.
The main exception occurs during, and shortly after, storms when the river carries an appreciable
sediment load – normally sediment concentrations can be considered negligible (turbidity readings
usually meet drinking water standards).
The second is that Channelkeeper TDP concentrations are what I’ve termed in the past “modified
TDP,” which I often express as TDP*. There has been, and still is, a major problem with the LTER
TDP analysis: TDP concentrations are often lower than those for phosphate in the same sample.
Obviously, this cannot be true: TDP may equal phosphate, but can never be lower. But this is not a
uncommon occurrence if phosphorus concentrations are low and if dissolved organic phosphorus,
DOP, makes up a very small percentage of the phosphorus total (the LTER doesn’t actually
measure phosphate, or what is often called ortho-phosphate, it measures soluble reactive
phosphorus, SRP, which may include some small amount of low molecular-weight organic
phosphorus; either way, phosphate + DOP, or SRP + DOP equals TDP). Unfortunately, this is not
an occasional problem with the LTER analysis; it occurs in 42 % of the samples included in the
Channelkeeper dataset. Modified TDP partially corrects for this problem by substituting phosphate
concentrations for TDP whenever a TDP result is lower than phosphate. All Channelkeeper data in
the above graph, and in all the graphs that follow, show modified TDP or TDP*. While I believe
this doesn’t introduce appreciable error (the majority of “modified” samples containing low
concentrations of DOP) I have no way of proving it, so when looking at Channelkeeper TDP
concentrations the words “at a minimum” should be kept in the back of your mind.
phosphate (mg/L)
5
PO4 = 0.9*(TP)
R2 = 0.87
4
3
2
1
0
0
1
2
3
4
5
total phosphorus (mg/L)
The discussion on the previous page lends itself to the question of how much of the phosphorus
coming out of the WWTP is phosphate? Fortunately, since August 2003 effluent analysis has
included both TP and phosphate. The above graph shows these results. There is a consistent
correlation between the two; phosphate typically makes up 90 % of the total phosphorus found in
effluent (r-square = 0.87; the plain circles represent data considered outliers and were left out of
the regression).
above the WWTP
below detection limit
Foster Park
TDP (mg/L)
0.2
10,000
1,000
0.1
100
10
r-0
8
-0
7
Ap
O
ct
r-0
7
-0
6
Ap
O
ct
r-0
6
-0
5
Ap
O
ct
r-0
5
-0
4
Ap
r-0
4
O
ct
Ap
-0
3
O
ct
r-0
3
-0
2
Ap
O
ct
r-0
2
-0
1
Ap
O
ct
Ap
O
ct
r-0
1
1
-0
0
0.0
flow at Foster Park (cfs)
100,000
In this graph OVSD TP data from samples collected above the treatment plant (R-3) are plotted along
with flow and TDP from Foster Park. Again, there is no joy. Most of the OVSD samples were below
the detection limit (< 1 mg/L up to 2005, < 0.018 from late 2006, and “your guess is as good as
mine” for the period in-between). Although the “non-detects” look like they fit the Channelkeeper
pattern, this is pure coincidence based on my arbitrary choice on assigning concentrations to these
samples. Again, as for phosphate, the light-blue circles show concentrations appreciably higher than
the Channelkeeper Foster Park results.
100,000
C.Larga confluence
below detection limit
LTER
Shell Road.
TDP (mg/L)
3
10,000
1,000
2
100
1
10
r-0
8
-0
7
Ap
O
ct
r-0
7
-0
6
Ap
O
ct
r-0
6
-0
5
Ap
O
ct
r-0
5
-0
4
Ap
O
ct
r-0
4
-0
3
Ap
O
ct
r-0
3
-0
2
Ap
O
ct
r-0
2
Ap
r-0
1
O
ct
Ap
O
ct
-0
1
1
-0
0
0
flow at Foster Park (cfs)
4
Looking at OVSD TP concentrations just above the Canada Larga confluence (R-5) (and a few
taken this summer, marked LTER) compared with Channelkeeper Shell Road TDP results. The
patterns are similar although the actual concentrations are noticeably higher (with a few exceptions).
100,000
TDP (mg/L)
above S.Antonio confluence
below detection limit
LTER
Foster Park
10,000
0.2
1,000
100
0.1
10
r-0
8
Ap
-0
7
O
ct
r-0
7
Ap
-0
6
O
ct
r-0
6
Ap
-0
5
O
ct
r-0
5
Ap
-0
4
O
ct
r-0
4
Ap
-0
3
O
ct
r-0
3
Ap
-0
2
O
ct
r-0
2
Ap
-0
1
O
ct
Ap
O
ct
r-0
1
1
-0
0
0.0
flow at Foster Park (cfs)
0.3
TP concentrations measured just above the San Antonio confluence in earlier years (R-1, when the
OVSD had an expanded sampling program) turn out to be relatively useless, as all but one fell
below the detection limit and that one was collected prior to the Channelkeeper program. Recent
samples collected at this location (LTER) show a great similarity to those from Foster Park (3.4 km
downstream)(a result we might profitably spend some effort contemplating). The one high value is
most likely in error, having anomalously high DOP.
100,000
Shell Road (OVSD)
below detection limit
just below outfall
below detection limit
Shell Road.
TDP (mg/L)
3
10,000
1,000
2
100
1
10
r-0
7
O
ct
-0
7
Ap
r-0
8
6
Ap
O
ct
-0
r-0
6
5
Ap
O
ct
-0
r-0
5
4
Ap
O
ct
-0
r-0
4
3
Ap
O
ct
-0
r-0
3
2
Ap
O
ct
-0
1
r-0
2
Ap
O
ct
-0
Ap
O
ct
-0
r-0
1
1
0
0
flow at Foster Park (cfs)
4
Finally, here’s a plot of OVSD TP concentrations from samples collected just below the WWTP
outfall (R-4) – about 0.5 km above the Canada Larga confluence. Also shown are Shell Road TP
concentrations from the early series of OVSD samples (R-6) and Channelkeeper Shell Road TDP
results. I’ve used a pale-yellow color to indicate non-detect results for both OVSD series. This
graph probably displays the most encouraging results: the TP patterns are similar and the actual
concentrations not far removed from each other. Shell Road values are usually lower than those
right below the outfall – as they should be. I’ve assigned a value of 0.009 mg/L to the long series of
2005-06 (R-4) non-detects (based on the minimum detection limit of later samples); they could just
as easily be given a value of 0.05 mg/L, in which case they would fit right in with the
Channelkeeper results. High flows kept TP concentrations low at all points below the WWTP
during this extended period, reinforcing my point that it’s river flow and not treatment plant output
that determines lower reach phosphorus concentrations.
As to my overall thoughts? As long as we are dealing with relatively high phosphorus
concentrations all the datasets are compatible and in rough agreement with each other, and I’m
confident that they express reality. However, low concentrations are a problem, and one that is
currently irresolvable.
1
TDP* (mg/L)
2001-2007, 2008 means
mean (2001-2007)
2008 mean
0.1
0.01
le
y
an
St
M
ai
n
St
re
et
D
ra
Sh
in
el
lR
C
oa
an
d.
ad
a
U
La
pp
rg
er
a
C
.L
ar
Fo
ga
st
er
lo
w
Pa
er
rk
S.
An
t
on
Li
on
io
C
an
yo
Pi
n
rie
up
C
pe
r
r S eek
.A
Sa
nt
nt
on
a
io
An
a
Bl
vd
M
.
H
at
w
y.
ilij
a
15
be
0
lo
w
N
d
.F
am
or
M
k
at
M
ilij
at
a
ilij
ab
a
ov
e
da
m
0.001
1
TDP* (mg/L)
2005-2007, 2008 means
mean (2005-2007)
2008 mean
0.1
0.01
D
ra
Sh
in
el
lR
C
oa
an
d.
ad
a
U
La
pp
rg
er
a
C
.L
ar
Fo
ga
st
er
lo
w
Pa
er
rk
S.
An
to
Li
ni
on
o
C
an
yo
Pi
n
rie
up
C
pe
r
r S eek
.
A
Sa
nt
nt
on
a
io
An
a
Bl
vd
.
M
H
w
at
y
ilij
.
a
15
be
0
lo
w
N
da
.F
m
or
M
k
at
M
ilij
a
a
til
ija
ab
ov
e
da
m
y
le
an
St
M
ai
n
St
re
et
0.001
Because of the “step” down in all LTER analyzed phosphate and TDP values after 2004 I’ve
decided to show 2008 data in two different ways: first in comparison with averages of 2001
through 2007 concentrations be they for individual months or for all years; and then in comparison
with averages using only 2005 through 2007 data – thus avoiding problems with the phosphorus
dichotomy and presenting a more balanced picture of what happened in 2008. These graphs,
showing mean monthly TDP concentrations, are an example of how including, or excluding, the
first 4 years of data changes the comparison. The TDP files, which show both of these alternatives
along with TDP and phosphate comparisons, are especially complicated. There are 3 of them: TDP
& phosphate.VR.I, TDP & phosphate.VR.II, and TDP & phosphate.VR.III. The third is organized
similarly to the “III” files for nitrate and phosphate, but the first two are somewhat different.