Analytical Method for Evaluating Chlorophyll Criteria Exceedances

An Analytical Method for Evaluating Chlorophyll Criteria Exceedances

Dr. Paul Jacobson has suggested to the Tidal Monitoring Network Redesign Workgroup ameaningliil and robust way of analyzing chlorophyll data
for criteria exceedances (Figure 1 - 4). The analytical method is best applied to 2D images of chlorophyll concentration (such as the data collected
with remote sensing data from satellite or overflights), but can be used with in vivo fluorescence data collected with towed arrays and point data
collected at fixed stations.

Figure 1: Data Processing

Figure 2: Data Evaluation

Figure 3: Exceedance Layers

Figure 4: CFD Plots

SEGMENT

Station

A segment is divided into a grid
of surface cells. Satellite or
aircraft overflight images are
processed into gridded
chlorophyll layers for individual
dates. Gridded salinity lasers for
each monitoring cruise are
generated from station point data
with the Chesapeake Bay
Program's 3D Interpolator model.
Individual chlorophyll layers are
paired with the salinity layer
closest in time. Note: gridded
chlorophyll layers can also be
generated from in-vivo
fluorescence data collected with a
towed sensor and pmcessed in the
3D Interpolator Model.

SEGMENT

~ Meets Criteria
¦ Exceeds Criteria

Matching cells from the paired
salinity and chlorophyll layers
(Figure 1) are analyzed to
determine if they meet or exceed
the chlorophyll criteria,

e.g. "if salzone = F and
season = Spring and Chi
>25 ug/liter, then
exceedance =yes; or
"if salzone = M and
season = Spring and Chi
>30 ug/liter, then
exceedance = yes; or....
else, exceedance = no."
A new, "exceedance" layer is
generated which identifies grid
cells that exceeded the season-
salinity specific chlorophyll
criteria.

e.g. SPRING of YYYY

Area of Exceedance
sorted from high to low

For each exceedance layer, the %
of the segment area exceeding the
chlorophyll criteria is calculated by
dividing the number of cells in
exceedance by segment's total
number of cells. Multiple
exceedance layers are generate for
each Spring and Summer from the
monitoring data. The values of %
Area of Exceedance within a
season are sorted from high to low
and a cumulative frequency
distribution (CFD) curve is
generated. Note: if the attainment
period is three years, CFD's would
be generated from Spring and/or
Summer % Area of Exceedance
pooled over three years.

Frequency and Extent of Chlorophyll Exceedances

Severe Impairment

10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
% Area in Exceedance of Chlorophyll Criterion

J Impairment | | Non-Impairment

A segment's CFD curve is plotted
against % Area in Exceedance and
compared to the reference curve. If
the segment's CFD curve falls
completely above the reference
curve (gray area in graph), the
segment is impaired. If the CFD
curve falls completely below the
reference curve (white area in
graph), the segment is not
impaired. If the CFD curve is close
to the reference curve and doesn't
completely exceed it, the segment
is marginally impaired. It may
exhibit infrequent but extensive
exceedances (blue line), or frequent
but local exceedances (green line).

-------
Evaluating CBP Fixed-Station Point Data

The analytical method can be adapted to evaluate
point data currently collected at fixed-stations by
the Chesapeake Bay Program. Point data collected
from one or two stations within a segment are
grouped by Season-Year and used as if they were a
"layer." The % exceedance in the Season-Year's
pooled samples are analogous to the geographical
"% Area of Exceedance." In the real-data example
in Table 1 (Potomac, 1995 - 1999), 4 of 5 (80%)
mesohaline fixed-station samples in the Summer of
1996 exceeded the chlorophyll criterion as
determined by the Water Quality Binning Method.
The % Exceedance in the segment, therefore, is
estimated to be 80% for the Summer 1996 "layer."
None of the 1998 Spring and 1999 Summer
mesohaline samples exceeded the chlorophyll
criteria. Therefore, the % Exceedance in the
segment for these "layers" is estimated to be 0%.
And so on. When % Exceedance has been
calculated for each "layer," a cumulative frequency
distribution curve is determined for the segment.
The CFD is compared to a reference curve to
determine if the segment is impaired (Figure 5, 6).

In Figure 5, the Potomac mesohaline CFD's for
Spring and Summer are compared to the Water
Quality Reference Curve since the chlorophyll
criteria used to evaluate the data were those
determined by the Water Quality Binning Method.
When the chlorophyll criteria used to determine
exceedance are based on the Food Quality Method,
the resulting CFD's are compared to the draft Food
Quality Reference Curve, as in Figure 6. The
examples indicate the mesohaline Potomac is
impaired by pbytoplankton (as chlorophyll) from
both water and food quality perspectives.

Table 1: Example of Evaluated Point Data

MESOHALINE CHL CRITERIA (Water Quality Burning Method)

Spring: Chlorophyll < 27 uafliter Summer: Chlorophyll <16 uafliter

Year Season

Exceedance

Samples

Exceedance

Cum

Cum Freq

per Segmt

in Segment

Freq ($

(%)

1996 SUMMER

80.00%

20.00%

1997 SUMMER

20.00%

1998 SUMMER

20.00%

60.00%

1995 SUMMER

16.67/c

80.00%

1999 SUMMER

0.00%

100.00%

1999 SPRING

60.00%

20.00%

1997 SPRING

50.00%

40.01]%

1995 SPRING

20.00%

1996 SPRING

20.00%

80.00%

1998 SPRING

0.00%

100.00%

Figure 5: Point Data CFD Curves vs Water
Quality Reference Curve

Potomac Mesohaline Example

Spring & Summer, 1995- 1999

100%

90% -

~ 70%

§ 60%

CT
CD

£ 50%
1 40% A

30%
20%
10%
0%

Impairment

Water Qualify j^

» Spring^ Summer

Keterence Curve '

0% 10% 20% 30% 40% 50% 60% 70% 30% 90% 100%
% Exceedance in Point Samples Pooled by Season-YYYY

Figured: Point Data CFD Curve vs Food Quality
Reference Curve

100%
90%
g 80%
& 70%

a
-------
Reference Curves

A Water Quality Reference Curve can be
determined from the subset of Chesapeake Bay
chlorophyll data associated with the Water Quality
Binning Method's "Better" and "Best," and sometimes
"Mixed_BetterLight," water quality categories. These
categories were the basis for characterizing the
Phytoplankton Reference Community. Chlorophyll
criteria derived with this method were the 95th
percentiles of chlorophyll values occurring in these
categories (Table 2). CFDs of data points exceeding
the 95th percentiles are considered exceedances that
occur naturally under reference conditions (i.e. growth-
limiting nitrogen and phosphorous concentrations and
relatively high light levels). Chlorophyll exceedances in
the eight season-salinity reference communities fall on
very similar CFDs, so a single CFD curve can be used
to evaluate all Spring and Summer data (Figure 7).

A Food Quality Reference Curve can be determined
from the subset of Chesapeake Bay chlorophyll data
associated with phytoplankton that are not affected by
detrimental taxa concentrations.1 Chlorophyll criteria
derived empirically with the Food Quality Method
(Table 2) are applied to this subset of data. CFDs of
data points exceeding these criteria are considered
exceedances that occur naturally in phytoplankton
communities not affected by detrimental algal taxa
concentrations. Figure 8 illustrates the draft Summer
mesohaline food quality reference curve.

The Water Quality and Food Quality reference curves
are compared for Summer mesohaline in Figure 9. The
Water Quality curve is more conservative than the
Food Quality curve because the criteria derived from
the Water Quality Binning Method are lower. Also
illustrated in Figure 9 is an arbitrarily selected reference
"curve" of 10%. This curve "forgives" infrequent
exceedances that cover large areas and frequent
exceedances that occur locally.

Table 2. Chlorophyll Criteria (ug/liter)

derived from two methods.

Water Quality
Binning Method
(Phytoplankton
Reference
Community)

Spring

Tidal Fresh 12.1
Oligohaline 22.5
Mesohaline 26.8
Polyhaline 7

Summer

Tidal Fresh 16

Oligohaline 22.6
Mesohaline 16.2
Polyhaline 8.8

Food Quality

Method
(Detrimental
Algal Taxa
Concentrations)

-25

-30
-20

-40
-55
-20
-15

Figure 7. Chlorophyll Exceedances in the
Phytoplankton Reference Community (Water
Quality Binning Method)

20% 40% 60% 80%
%Area in Exceedance

100%

Tidal Fresh Oligohaline Mesohaline Polyhaline

~ Summer
oSpring

~ Summer

~ Spring

ASummer
ASpring

O Summer
o Spring

1 Detrimental taxa concentrations are still being determined. This involves identifying thresholds above which
mesozooplankton parameters are suppressed by the major phytoplankton taxa.

-------
Figure 8. Chlorophyll Exceedances in Data
Not Impacted by Detrimental Taxa
Concentrations (DRAFT).

%Area in Exceedance

Figure 9. Possible Reference Curves (Summer
Mesohaline Example)

Claire Buchanan,

with help from Paul Jacobson, Elgin Perry, and Marcia Olson
February 4, 2002

-------
-------