e. minerogenic particulate phosphorus model · e. minerogenic particulate phosphorus model 1...

- the unavailable component (PPm/u)

- development, testing, and application

E. Minerogenic Particulate

Phosphorus Model

10/22/2015 Upstate Freshwater Institute 1

Outline


1. background, supporting elements, and an empirical PPm/u

model

2. conceptual mechanistic PPm/u model, input considerations

and development

3. model performance

4. model analyses

5. applicability, advancements, and summary

Outline



model


and development


4. model analyses


Unavailable minerogenic particulate phosphorus

(PPm/u), other P fractions, and bioavailability - (Prestigiacomo et al., 2015)

total P P fraction bioavailability (assays)

total

P

total dissolved

(TDP)

(<0.45 µm)

particulate

SRP

SUP

PPo

PPm

complete

mostly

mostly

limited

soluble reactive

soluble unreactive

organic particulate

minerogenic particulate PP

partitioning PPm = PPm/u + PPm/a

PPm/u

PPm/a

none

complete

minerogenic unavailable

particulate

minerogenic available

particulate, small fraction

here


PPm/a ÷ PPm ≈ fBAP for PP

Implications for Water Quality Model Structure PP needs to be partitioned

according to PPo and PPm

PAVm or a proxy is a potentially valuable state variable to represent minerogenic particles

external loads of PAVm or a proxy would be necessary

longitudinal segmentation needed to differentiate near-shore versus pelagic waters conditions


NYSDEC/TAC meeting, Jan. 15, 2014

based on: Effler et al. 2014. Inland Waters 4:179-192

consistency with implemented tool

Independent estimates of PPm/u and the ratio PPm/u:PAVm

from an empirical model: A test for the dynamic

mechanistic model

development and testing for the lake documented (Effler et al. 2014)

stoichiometry – based model for lake PP, PPo, and PPm/u

supports the partitioning of PP

PP = PPo + PPm/u

PP + = (PPo:Chl)·Chl (PPm/u:PAVm)·PAVm

PP + = PPo PPm/u

stoichiometric

ratios

- from optimization analysis

of lake observations, 1999-2006

1.53 7.1 mg/m3


Chl – chlorophyll a temporal uniformity

in stoichiometric

ratios invoked

Implications of PPm/u concentrations in lakes and

related challenges for its modeling

noteworthy contributions of PPm/u compromise TP concentration as a metric of trophic state because of the unavailability of PPm/u to support plant (e.g., algae) growth

differences in time and space within individual lakes, and between lakes, are to be expected, associated with coupled differences in responsible minerogenic particles (PAVm) – runoff event effects important

runoff events expected to promote higher PAVm and thereby PPm/u

challenges to model PPm/u in lakes, quantification of: 1. delivery, transport, and fate of minerogenic

particles (PAVm), and 2. P associated with these particles (PPm) and its

bioavailability (PPm/u vs PPm/a)


pelagic shelf

TP

PPm/u

Outline



model


and development


4. model analyses


Extensive technical program support for

mechanistic PPm/u model for Cayuga Lake

lake monitoring – LSC program 1999-2006, plus 2013 P fractions, Chl-a, PAVm (size classes and type)

primary tributaries P fractions, sediment (ISPM, SPM), PAVm (size

classes)

bioavailability assays (NYSDEC recommended)

submodels (1) 2-D transport, (2) PAVm (4 size classes)

loading estimates – PAVm and PPm/u

runoff event sampling focus (NYSDEC recommended), Inlet (extra, UFI and Cornell)

sediment trap (extra, UFI and Cornell)


A mechanistic model for unavailable minerogenic

particulate P (PPm/u) for Cayuga Lake: Background


P associated with minerogenic particles (PPm/u) delivered by the watershed interferes with the use of TP concentration as a trophic state metric in lacustrine systems because of its limited bioavailability

a mechanistic mass balance model for PPm/u has been developed and tested, as described here

supporting components long-term and intensive 2013

monitoring

LSC monitoring (1999-2006)

bioavailability assessments

loading estimates, PAVm and PPm/u

transport and PAVm submodels

pelagic shelf

TP

PPm/u

Conceptual model for dynamic mechanistic PPm/u model


model features

PPm/u:PAVm of delivered particles subject to variation

in-lake behavior of PPm/u is assumed to be that of PAVm with which it was associated upon entry

in-lake dynamics of PPm/u reflect external load of PAVm/n, the dynamics of the PPm/u:PAVm ratio in the tributaries and the progression of in-lake PAVm loss processes

PPm/u:PAVm

PAVm/n loads PPm/u loads

= f(Q,PP and PAVm observations);

i.e., varies

transport submodel

Gelda et al. 2015a

PAVm submodel

Gelda et al. 2015b

PPm/u

PAVm/n particles

lake

sink processes for PAVm

PAVm submodel

transport

submodel

PPm/u apportioned to the PAVm size classes PAVm/n

according to their contributions to total PAVm

tributary

inputs

Specifications for modeling PPm/u

well-mixed upper waters targeted (epilimnion)

linear dependency of PPm on PAVm

described by Effler et al. (2014), conceptual consistency

partitioning of PPm according to size class contributions

PPm from tributary inputs of PP according to: PPm = PP·(ISPM:SPM); ISPM:SPM generally > 0.9

PPm subject to temporal variability in tributary inputs according to variations in the PPm:PAVm ratio tributary loads of PPm from PAVm load estimates

PPm load = PAVm load·(PPm:PAVm)

PPm/u loads incorporate tributary-specific bioavailability results PPm/u load = (1-fBAP)·PPm load

supported by rapidity of the transformation

in-lake behavior of PPm/u equivalent to that of the PAVm with which it is associated

Biossay Progression (d)

0 5 10 15 20

f BA

P,t:f

BA

P

0

25

50

75

100


• fBAP = fraction bioavailable

for completed bioassay

• fBAP,t = fraction bioavailable

through progression of

bioassay

f BA

P,t ÷

fB

AP

Considerations for modeling PPm/u: Tributary

conditions and loads of minerogenic particles

positive dependencies of particle

concentration (ISPM), and the

minerogenic component on stream

flow (Q)

implications of runoff events

dominance of minerogenic

particles – glacial lacustrine stream

deposits (Nagle et al. 2007)

origins of variance in dependencies

e.g., character of stream banks

ISP

M (

mg/L

)

0.1

1

10

100

1000 ISP

M:S

PM

0.70

0.75

0.80

0.85

0.90

0.95

1.00

ISPM

ISPM:SPM

Q (m3/s)

0.1 1 10 100

PA

Vm (

m-1

)

0.1

1

10

100


Fall Creek examples

Consideration for modeling PPm/u: Tributary

conditions, dependence of PP on PAVm


strong positive dependence of PP in streams on PAVm

most of PP is as PPm/u

the P content of clay mineral particles delivered

during a runoff event large quantities of PAVm and associated P (PPm) are delivered

associated with PAVm

0 100 200 300

0

200

400

600

800

0 50 100 150

0

100

200

300

400

500

0 100 200 300 400 500

0

1000

2000

3000

Six Mile

linear fit44

1: Fall Creek

2: Cayuga Inlet

3: Salmon Creek

4: Six Mile Creek

1

2

PVVm (ppm)

IS

PM

(g/m

3) y = 2.61x 9.23

R2 = 0.96

(int. not sig.)

3

(a)

(b)

4

Fall Creek

linear fit1

1

32

PAVm (m1)

Tn (

NT

U)

y = 4.03x 3.70

R2 = 0.96

(int. not sig.)

Salmon Creek

linear fit3

(c)

PAVm (m1)

PP

(

g/l

)

y = 6.62x 38.06

R2 = 0.99 (int. sig.)

3

2

4

1

Fig. 4c from

Peng and Effler 2015

Specifications for modeling PPm/u:

Tributary conditions and loads temporal patterns

runoff events dominate sediment loading to the lake

Fall Cr. Q time series 2013, prominent runoff events

cumulative format for PAVm and PPm/u loads

early July and August runoff events dominate


A M J J A S O

Q (

m3/s

)

0

20

40

60

80

100

PA

Vm load

(10

6 m

2)

0

500

1000

1500FC PAV

m

Inlet PAVm

2013

Apr May Jun Jul Aug Sep Oct

PP

m/u (

kg)

0

2000

4000

6000

8000FC PPm/u

Inlet PPm/u

FC PPm


Tributary conditions and loads

positive dependencies on stream flow

(Q)-PP, PP:TP

Fall Cr. examples

PPm:PAVm ratio supports estiamtes of

PPm (and PPm/u) loads from PAVm loads

negative, but variable dependency on

Q

PP

(m

g/L

)

1

10

100

1000

PP

:TP

0.5

0.6

0.7

0.8

0.9

1.0

PP

PP:TP

Q (m3/s)

0.1 1 10 100

PP

:PA

Vm (

mg/m

2)

1

10


PP

m/u:P

AV

m

(mg

/m2)

0

10

20

30

40

Q (m

3/s)

1

10

100PP

m/u:PAV

m

Q

PP

m/u (

kg

)

0

2000

4000

6000

8000FC PPm/u

Inlet PPm/u

FC PPm

2013


PP

m/u:P

AV

m

(mg

/m2)

0

4

8

FC

Inlet


Tributary conditions and loads temporal patterns

example for Fall Cr, PPm/u:PAVm and Q

cumulative formats for PPm/u load and the ratio

PPm/u:PAVm highly variable, negative dependency on Q

dominance of runoff events for PPm/u loads

PPm loads just slightly higher, consistent with low fBAP

cumulative change for ratio will drive changes in-lake relationship


Outline



model


and development


4. model analyses


Mechanistic dynamic PPm/u model performance:

Targeted attributes for testing


tests of consistency

extent of closure with prediction of independently tested empirical PP model in-lake PPm/u:PAVm ratio

in-lake PPm/u

PPm/u with historic TP and PP shelf observations, following runoff events

predicted PPm/u deposition on shelf from runoff event inputs with sediment trap observations


Lake PPm/u:PAVm ratio vs. empirical model value

predictions of in-lake PPm/u:PAVm ratio

only minor shelf vs. pelagic differences

averages (shelf – 8.0 mg/m2; pelagic – 7.4 mg/m2) closed well with

single empirical model value (7.1 mg/m2; Effler et al. 2014)

in-lake variations (cv=0.22) driven by tributary variations (cv=0.46)

tributary dynamics linked to variations in Q

2013


PP

m/u:P

AV

m (m

g/m

2)

0

5

10

15

20shelf

pelagic

empirical model



Comparison of PPm/u predictions, mechanistic vs

empirical models

comparisons for both the shelf

and pelagic waters

predictions for 2013, with Fall

Cr. Q for runoff event context

empirical model predictions

limited to sampling occurrences

reasonably strong relationships

shelf vs pelagic different scaling

magnitudes consistent overall Q

(m

3/s

)

0

20

40

60

80

100(a)

pelagic

shelf

0

5

10

15

2013


PP

m/u

(µ

g/L

)

0

1

2

313 59


Mechanistic PPm/u model performance: Consistency with

historic TP and PP observations following runoff events


identifying runoff events

from LSC monitoring

record

same runoff events

considered for PAVm

(sub)model

Table 2 (Gelda et al. 2015b)

signatures of increased

PPm expected on shelf

following runoff events

The complications of turbidity plumes on the shelf

following major runoff events


issue for PPm/u model

evaluation as with the

PAVm model

Mechanistic PPm/u model performance: Consistency with

historic TP and PP observations following runoff events

variations is predicted PPm/u

for long-term (1999-2006)

simulations performed

reasonably well in

explaining observed

differences for historic

events

performance improves

somewhat with

representation of the effects

of PPo (phytoplankton

biomass) variation also

PPm/u

(µg/L)

1 10 100

PP

obs (µ

g/L

)

1

10

100

1 2 3

4

5 6 7

8

9 10 11 12

13

14

15

I = 0.83S = 0.33

r ² = 0.38

TP

obs (µ

g/L

)

1

10

100

1

2 3 4

5 6

7 8

9 10 11 12

13

14

15

I = 0.77S = 0.48

r ² = 0.46

PPm/u

+ PPo (µg/L)

1 10 100

PP

obs (µ

g/L

)

1

10

100

1 2 3

4

5 6 7

8

9 10 11 12

13

14

15

I = 57S = 0.49

r ² = 0.53


numbers correspond to

the various historic

runoff events listed in

previous table (Table 2,

Gelda et al. 2015b)

recall this is an imperfect

match of P fractions

PP = (PPo:Chl)·Chl

1.53

Effler et al. (2014)

PPm/u

(µg/L)

1 10 100

PP

obs (µ

g/L

)

1

10

100

1 2 3

4

5 6 7

8

9 10 11 12

13

14

15

I = 0.83S = 0.33

r ² = 0.38

TP

obs (µ

g/L

)

1

10

100

1

2 3 4

5 6

7 8

9 10 11 12

13

14

15

I = 0.77S = 0.48

r ² = 0.46

PPm/u

+ PPo (µg/L)

1 10 100

PP

obs (µ

g/L

)

1

10

100

1 2 3

4

5 6 7

8

9 10 11 12

13

14

15

I = 57S = 0.49

r ² = 0.53

Consistent mechanistic PPm/u model performance

for local deposition from runoff events

comparison of simulations of deposition of PPm/u (i.e., associated with PAVm) with sediment traps observations of PP downward flux

simulations and observations both elevated for major runoff events

semi-quantitative support, given the variable operation and trajectories of turbid plumes

Q (

m3/s

)

0

20

40

60

80

100(a) Fall Creek

(b) Site 2

May Jun Jul Aug

DF

PP

(mg/m

2/d

)0

20

40

traps

predicted PPm/u

deposition


DFPP – downward flux of PP

Consistency of the mechanistic and empirical models in

representing contributions of PPm/u and PPo to PP for

shelf vs pelagic waters


recall, Chl-a (e.g., phytoplankton biomass) not significantly different on shelf vs pelagic waters

recall PP = PPo + PPm

both models higher PPm/u values on shelf

vs pelagic greater variability on shelf

median values – shelf vs pelagic waters, contributions of PPm/u

20 vs 11%, mechanistic 16 vs 8%, empirical

reasonably good closure

PPo PPm/u

empiricalmodel

% C

ontr

ibution to P

P

0.00

0.25

0.50

0.75

1.00

mechanisticmodel

PPo PPm/u

median

mean

75th percentile

25th percentile

Key:

90th percentile

95th percentile

10th percentile

5th percentile

• PPm/u much greater, and PPo much

smaller contributions on the shelf for

the extremes of runoff events

shelf pelagic

comparisons

Outline



model


and development


4. model analyses


Analyses with the tested PPm/u model: Dependence of

shelf response on runoff event magnitude, PPm/u vs. Q

peaks

Fall Cr. peak Q for the earlier

(1999-2006) runoff events

corresponding predicted peak

PPm/u at Site 2 on shelf

strong positive dependency of

PPm/u on event magnitude

sources of variance in the

relationship-variations in

ambient mixing, limitations

in peak Q defining external

PPm/u loads Fall Cr. Peak Q (m

3/s)

10 100M

ax.

PP

m/u(µ

g/L

)1

10

100

1000

1 2 3

4 5

6

7 8

9

10

11 12

13

14 15


2013


Q (

m3/s

)

0

20

40

60

80

100

examples

Analyses with the tested PPm/u model: Interannual

variations in predicted PPm/u explains much of the

interannual and spatial variations in TP


total P (TP) concentration is a trophic state metric, guidance value of 20 µg/L as summer average, in NY

PPm/u variations important in regulating interannual and shelf vs pelagic differences in TP

PPm/u lower at pelagic site each year

interannual differences in PPm/u between sites explained much of the year-to-year differences in TP (45%)

these interannual differences in PPm/u explain 47% of the year-to-year differences in TP at both sites

PPm/u – based on mechanistic

model predictions

1999 2000 2001 2002 2003 2004 2005 2006P

(µ

g/L

)

0

10

20

30

TDP PPo

PPm/u

TP limit

s

p

Days

0

25

50

75

Days

0

25

50

upper 10% flow (a)

PPm/u > 5µg/L (b)

Days

0

25

50 PPm/u > 10µg/L (c)

Days

0

25

50 PPm/u > 20µg/L (d)

Days

0

25

50 PPm/u > 50µg/L (e)

99 01 03 05 07 09 11 13

Analyses with the tested PPm/u model: Interannual

variations in day counts of elevated concentrations on

the shelf

multiple PPm/u concentration thresholds

chosen for Jun-Sept. interval, 1999-2013

high flow events represented by number of

days Fall Cr. Q was in the upper 10%

high PPm/u concentrations coupled to runoff

event occurrences

major interannual variations predicted;

consistency with observations

timing of monitoring could be important to

reported summer average and guidance

value status

extreme cases of high PPm/u predicted for

2006 and 2011


based on model simulations for the

1999-2013 period

Outline



model


and development


4. model analyses


Applicable findings from the PPm/u modeling initiative

there has been a qualitative recognition of the interference of minerogenic particles for metrics of trophic state for decades

the importance of these particles depicted here, relative to the application of TP, is likely broadly occurring

advancements from, or value of, this PPm/u modeling

improved usage of TP as a trophic state metric

applicable to the numerous systems of similar setting and issues

advancement of water quality models to represent this issue and effects of minerogenic particles

these advancements will be necessary to address the implications of predicted features of climate change (NOAA 2013) – more runoff events and severity of the events


Summary


P associated with minerogenic particles (e.g., PPm) delivered from the watershed

interferes with the use of TP as a trophic state metric because of its limited bioavailability

a mass balance-type model for unavailable PPm (PPm/u) has been successfully developed

and tested

higher PPm/u on the shelf following runoff events is predicted , that causes irregular

exceedances from TP guidance value, that is uncoupled from trophic state

positive features of model performance included:

reasonable closure of predicted shelf and pelagic levels of PPm/u and PPm/u:PAVm with those

from an independent empirical model (Effler et al. 2014)

consistency of predicted PPm/u shelf deposition with sediment trap observations

a consistent partitioning of the PP pool between PPm/u and PPo (phytoplankton) for historic

observations

consistency of post-runoff event TP and PP observations with PPm/u predictions

this advancement in modeling is invaluable and appropriate for large initiatives

addressing the trophic state issue through TP

e. minerogenic particulate phosphorus model · e. minerogenic particulate phosphorus model 1...

Documents