considerations for phase ii water quality modeling cayuga lake · 05/11/2014 · linking of...

Considerations for Phase II Water Quality Modeling

Cayuga Lake

11/5/2014 Upstate Freshwater Institute 1

Talk Outline1. Introduction/Background2. Shelf-Pelagic Disconnect 3. Other Lake-wide

Signatures4. Model Needs5. Submodels


the potential for phosphorus (P)-driven cultural eutrophication problems in the southern end (shelf) of Cayuga Lake

shelf context/setting– localized tributary (dominant) and point

source inputs• 40% of water inflow• similar to many reservoirs

– water quality listings• phosphorus (irregular exceedances of

TP guidance value), trophic state the concern

• sediment (metric and limit not stated)• bacteria

11/5/2014 3

The Issue

Upstate Freshwater Institute

shelf

development, testing, and application of a credible mechanistic P-eutrophication model

to be used to guide related management deliberations

– focus on conditions on the shelf, but lake-wide capability necessary


Required: Quantitative Management Tool for P-eutrophication Issue for Cayuga

Lake

Modeling Objectives

model(s) to provide a quantitative tool with which to evaluate water resources management alternatives

linking of watershed and lake water quality models

resolve drivers/processes responsible for prevailing conditions


Identifying Key Model Needs from Limnological Review of Monitoring Data

1. 2013 observations the most complete in time and space will support calibration

2. earlier observations, in support of LSC monitoring (also, CSI tributaries) will support validation

3. model needs considered temporal scales to be resolved spatial scales to be resolved processes to be represented model state variables model drivers


TP(µ

gP/L

)

01020304050

studysummer

Chl

-a(µ

g/L)

0

2

4

6

Sites1 2 3 4 5 6 7

SD

(m)

2468

10

Shelf-Pelagic Disconnect in Trophic State Metrics, 2013


TPshelf > TPpelagic

Chlshelf ~ Chlpelagic

SDshelf < SDpelagic

shelf pelagic

Shelf-Pelagic Disconnect in Trophic State Metrics is Recurring,

1998-2012


the disconnect must be effectively represented in the model

“the disconnect” -worse trophic state on shelf indicated by TP and SD data, but not supported by Chl-a

Chl

-a(µ

g/L)

0

3

6

9

TP(µ

gP/L

)

0

10

20

30

site 1 & 2 avg., shelfsite 3, pelagic

site 1 & 2 avg., shelfsite 3, pelagic

Year 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12

SD

(m)

0

3

6

9

site 2, shelfsite 3, pelagic

Runoff Events Contribute to the

Shelf-Pelagic Disconnect


shelf more strongly impacted by runoff events

TP increasing and SD decreasing linked to runoff events

effects of runoff events must be simulated (i.e., short time scales addressed)

TP(µ

gP/L

)

010203040

Flow

FC

(m3 /s

)

1

10

100

20081

2013 Jun Jul Aug Sep

SD

(m) 3

6

9

site 1 & 2 avg., shelfsite 3, pelagic frequent


Minerogenic Particles Delivered During Runoff Events Cause the Shelf-Pelagic

Disconnect


shelf more strongly impacted by runoff events

PAVm – projected area of minerogenic particles per volume

FSS and PAVm increasing linked to runoff events

the need to simulate minerogenic particle dynamics– short-term loads

FSS

(mg/

L)0.1

1

10

100


PA

V m(c

m2 /L

)

0.1

1

10

100

Flow

FC

(m3 /s

)

1

10

100


Q (m3/s)

1 10 100

PA

V m (1

/m)

0.01

0.1

1

10

100

Minerogenic Particles Delivered During Runoff Events Causes the Disconnect: TP


June - September particulate P, a stoichiometric approach

PP = PPm + PPo; m – minerogenic, o – phytoplankton

PP = (PPm:PAVm)·PAVm + (PPo:Chl-a)·Chl-a

Effler et al. 2014. Inland Waters (see manuscript) PAVm as a state variable, need loads

shelf pelagic

Site1 2 3

P (µ

gP/L

)

0

10

20

30

40

TP limit

TP35.1

16.6

6.8

12.2

TP19.4

4.7

6.5

8.3

TP13.7

0.27.0

7.1

TDP PPo PPm

refinements may evolve in Phase 2

TDP, PPo and PPm to be predicted

stoichiometric ratios


Low Bioavailability of Runoff Event PP Consistency with the Disconnect

runoff event of July 1, 2013 fBAP – fraction of PP bioavailable

SitePP

(µg/L)

Fall Cr. 444

6 Mile Cr. 271

Cay. In. 202

Inlet Channel 104

shelf 46

• demonstrated: shelf PP (post-runoff event) is essentially unavailable to support algae growth; i.e., uncoupled from trophic state

• implications: these features are not supportive of the inclusion of post-runoff event TP observations for assessment of trophic state status

shelffBAP = 3%

InletfBAP = 10%

6 MilefBAP = 6%

Fall Cr.fBAP = 18%

Minerogenic Particles Delivered During Runoff Events Cause the Disconnect: SD


clarity, measured by Secchi depth (SD)SD−1 bpbp = scattering coefficient for particulate material, bulk measurementsbp = bm + bobm = scattering coefficient associated with minerogenic particlesbo = scattering coefficient associated with organic (e.g., phytoplankton) particles

increase in bm from runoff events cause decrease in SD (Effler and Peng 2014)bm = 2.3 × PAVmPAVm as a model state variable

Second Part of the Shelf-Pelagic Disconnect


elevated SRP (phytoplankton growth potential) on shelf does not result in higher Chl-a

contributing processes– rapid flushing– dilution from tributaries– reduced light

availability, particularly during runoff events

Chl-a pattern reflects lake-wide conditions

SR

P(µ

gP/L

)0369

12

Flow

FC

(m3 /s

)

1

10

100

3517

site 1 & 2 avg., shelfsite 3, frequent


Chl

-a(µ

g /L

)

0369

12

Other Lake-Wide Signatures of Interest


POC alternate metric of phytoplankton biomass

NOX seasonal depletion, but non-limiting levels

Si interaction with diatom dynamics

PO

C(m

gC/L

)

0.0

0.5

1.0

1.5

NO

X (µ

gN/L

)

800

1200

1600

2013 Apr May Jun Jul Aug Sep Oct

Si

(mgS

iO2/L

)

0

1

2

Metrics of Phytoplankton

Biomass



PO

C(m

gC/L

)

0.0

0.5

1.0

1.5

2.0



Chl

-a(µ

g /L

)

0369

12

PO

C(m

gC/L

)

0.0

0.5

1.0

1.5

2.0


Biomass

difference in patterns of the two metrics


site 3 temporal variations dependency on ambient

conditions– nutrients– light

within literature range dynamic drivers not

empirically obvious indicates limitation in a

metric of phytoplankton biomass


Biomass

Chl

-a(µ

g /L

)

0369

12


Chl

-a/P

OC

(µ

gChl

/mgC

)

010203040

PO

C(m

gC/L

)

0.0

0.5

1.0

1.5

2.0

POC Performs Better Than Chl-a, Optically


bp – scattering coefficient for particulate material, bulk measurements SD−1 bp (Davies-Colley et al. 2013) bp = bm + bo; m – minerogenic, o – organic (Peng and Effler 2012) bm = 2.3 × PAVm; in Cayuga (Effler and Peng 2014) and others bo = bp − bm

0.0 0.4 0.8 1.2 1.60

1

2

3

4

0.2 1 10 200.1

1

5

Stramski et al. 2008

Oubelkheir et al. 2005

Model II fit; N = 52

y = 2.001[POC] 0.067

R2 = 0.86

RMSE = 0.233

bp b

m (

m

1)

POC (mg L1)

y = 0.32[Chl-a]0.74

R2 = 0.31

bp b

m (

m

1)

Loisel&Morel 1998

Huot et al. 2008

[Chl-a] (mg m3)

dependencies of bo on POC and Chl-a consistent with open ocean literature POC-based relationship much stronger

Lake-wide Role of Quagga Mussel Metabolism?Example – Phosphorus Excretion


a process(es) diminishes the effective fluxes on a water column basis– will need to be identified, integrated into model, and be tested as part of

overall model testing– work of Boegman et al. on this issue is being considered

laboratory fluxes

SRP (µg/L)0 7 14 21

Dep

th (m

)

132117

1028772574227

12observedpredicted

severe overprediction

prevailing simulations over-represent the effects of mussel excretion for case of adopting laboratory flux determinations


wide variations in hydrology and meteorological conditions

causes wide interannual variations in the estimated load of bioavailable P (BAPL)

Example: turbidity in NYC reservoir intakeprobabilistic model projections

(a)

0 4 8 12 16 20 24

Freq

uenc

y (%

)

0102030405060

(b)

Tn,w (NTU)

1 10 100C

FD (%

)0

20

40

60

80

100

15 N

TU

BAP L (

mt)

0

10

20

30

40

2000-201212y

no 201113y tribs SC PtS

2013

14.9

2.34 0.65

29.625.1

summed pointsources

long-term probabilistic projections with P-eutrophication model may demonstrate small changes in loading masked by interannual variations in hydrology

variability

Need for Probabilistic Model Predictions that Represent the Effects of Natural Variations in Drivers

Process Representation

model philosophy of parsimony – only as complex as necessary to address the issue and management alternatives


Fe-Oxygen cap

Psediment submodel

mussel metabolism:e.g., grazing, excretion, respiration

SRP

?plankton:we don’t know,may be potential should be built-in mussel dynamics:

e.g., bioenergeticssubmodel

boundary layer effect?collaboration with Cornell Biological Field Station, Shackelton Point

mixing

flux

metalimnion

hypolimnion

epilimnion

x

stratification, mixing regimesT of layers, stratification, ambient mixing

x

Model Needs from the Shelf-Pelagic Disconnect Analysis

1. temporal scales – broad– short-term, days – e.g., runoff events– seasonal – lake-wide effects, regulatory concerns– multi-year – meteorological/hydrologic variability

2. spatial scales – broad– within shelf– shelf extended to pelagic– entire water column

3. drivers – both short-term resolution and long-term capabilities– hydrology– constituent loads – e.g., P forms and minerogenic particles– meteorological conditions

4. state variables – direct and derived– forms of P– metrics of minerogenic particles (e.g., PAVm)– SD and light levels– metrics of phytoplankton biomass (e.g., Chl-a)


supported by 2-D model

see P loading paper

Model Testing Targets – Evolved from Monitoring and Analysis

primary1. shelf vs. pelagic waters – central role of runoff events

a) TP, TDP, SRP, PPb) PAVm, FSS, Tn, clarityc) the phytoplankton/Chl-a disconnect

• absence of higher shelf levels despite higher P (including SRP)• POC and Chl-a in 2013, Chl-a for < 2013• includes years of higher local loads from point sources• comparative light availability

2. pelagic and shelfa) phytoplankton – upper waters

1) calibration – seasonality for POC, summer avg. Chl-a2) validation – summer avg. Chl-a

b) clarity – contribution of phytoplankton and minerogenic particles, summer avg

c) representation of metabolic effects of prevailing mussel population on pelagic waters

d) effects of variations in drivers


elevated on shelf, the role of minerogenic particles calibration

validation

calibrationvalidation

Tentative List of State Variables


Derived State Variable Names Abbr.

Dissolved organic carbon DOC

Particulate organic carbon POC

Total organic carbon TOC

Dissolved organic phosphorus SUP

Particulate organic phosphorus POP

Total organic phosphorus TOP

Total phosphorus TP

Total chlorophyll a CHLA

Total suspended solids TSS

Total inorganic suspended solids FSS

Total turbidity Tn

Total PAV PAVm

State Variable Names Abbr.

Soluble reactive phosphorus SRP

Labile dissolved organic carbon LDOC

Refractory dissolved organic carbon RDOC

Labile particulate organic carbon LPOC

Refractory particulate organic carbon RPOC

Phytoplankton biomass ALG

Labile soluble unreactive P LSUP

Refractory soluble unreactive P RSUP

Labile particulate organic P LPOP

Refractory particulate organic P RPOP

Labile particulate inorganic P LPIP

Refractory particulate inorganic P RPIP

Turbidity Tni

PAV PAVm,ioptics state variables: SD, Ko(PAR), IrradianceSUP ≈ DOP* silica and nitrogen signatures may be tested

Driver Information Availability

1potential years involved in validation; evolving – LSC monitoring, CSI monitoring2gaged tributaries – Fall Creek, Inlet, Sixmile3land-based before 20124CSI monitoring and 2013 conc.-driver relationships


Driver Type Calibration 2013

Validation 1998 – 2012 1

Hydrology 2

Meteorology 3

Loads

Nutrients 4

Sediment 4

Model Submodels for Cayuga Lake Initiative

1. transport submodel – 2D, calibrated and validated, applications ongoing

2. minerogenic particle submodel – supporting data sets3. optics submodel – early stages supported by NASA grant4. tributary loads specification

a) empirical – e.g., concentration-driver relationshipsb) mechanistic – watershed/land use

5. nutrient (P) cycling submodel6. phytoplankton growth and biomass submodel



part of CE-QUAL-W2 2-D, longitudinal-vertical

hydrothermal/transport model setup, calibrated (2013), and validated (1998-

2012; continuous simulation) high performance

– seasonal thermal stratification– seiche activity – oscillations, upwelling

events– long-term simulations – applicability for

probabilistic projections applications related to water quality issues –

shelf residence time, plunging tributaries, vertical transport

time and space features consistent with water quality issues

see manuscript

Transport Submodel

Distance from downstream boundary (railroad bridge; km)

051015202530354045505560

Ele

vatio

n (m

)

-20

-10

0

10

20

30

40

50

60

70

80

90

100

110

120

LSC

dis

char

ge

LSC intake

Cayuga-AES power plantintake

partitions the minerogenic particle populations according to the contributions of multiple (e.g., n = 4) size classes

state variable PAVm – projected area of minerogenic particles per unit volume predicts minerogenic components of PP, Tn, TSS, SD, and Ko (PAR)


Minerogenic Particle Submodel

PAVm,1 Size Class 1

PAVm,2 Size Class 2

PAVm,n Size Class n

PAV

m,i

Q

settling

ag

gr.

dis

ag

gr.

𝑖=1

4

PAV𝑚,𝑖

bm

am

optics submodel

𝑖=1

4

PVV𝑚,𝑖

PPm= f(PAVm)Tn = f(PAVm)

FSS = f(PVVm)

External Loads

phosphorusturbidity

inorganic TSS

clarity (SD)attenuation

minerogenic components

Processes: settling (Stokes Law); aggregation/disaggregation (calibration); resuspension (?)* similar approach for turbidity in NYC reservoirs (Gelda et al. papers)

PVV: projected volume of particles/ volume bm: minerogenic scattering; am: minerogenic absorption

Tribs

submodel

Optics Submodel for Cayuga Lake


mechanistic – quantifies relationship between optically active constituents (OACs; e.g, Chl-a), inherent optical properties (IOPs), and in turn, apparent optical properties (e.g., Secchi depth, SD)

simple empirical relationships [e.g., SD = f(Chl-a)] perform poorly the supporting advanced measurements funded under a parallel NASA grant

OACsChl-aPOCPAVm

ISPMaCDOM

a*x( ), b*

x( )

IOPsa( ) = ax( )b( ) = bx( )

SD = (KL+cL)-1

Kd() = f[a( ), b( )]

cross-sectionsradiative-transfer

expressionsAOPs

SDKd() Ko(PAR)

IOPsa( ), b( ), c( )

**

a*x( ) = ax OACax

b*x( ) = bxOACbx

closure testsme

asu

rem

en

ts

pre

dic

ted

calculated

measurements

* advanced or tested here

closure tests

Water Quality State Variables

4. Tributary Loads Specification


example concentrations driver relationshipsa). Empirical

driven by records of ambient drivers – multiple time scales possible– stream Q– air T

b). Landuse Model output becomes input to lake water quality model

Q (m3/s)0.1 1 10 100 1000

PP

(µg/

L)

1

10

100

1000

2013

SR

P L

oad F

C (k

g/d)

0.1

1

10

100

1000

J F M A M J J A S O N D

Phosphorus Submodel


LPOP: labile particulate organic PRPOP: refractory particulate organic PLSUP: labile dissolved organic PRSUP: refractory dissolved organic PLPIP: labile particulate inorganic PRPIP: refractory particulate inorganic P

details may track:CE-QUAL-W2Lake2Kother models

additionally, include quagga mussel recycling

ZOOPLANKTON

ALGAE (P)

SRP

LPOP

RSUP

LSUP

RPOP

PredationRespiration

Respiration

Gra

zing

Settling

Adso

rpti

on

Growth

Hydrolysis

RPIP

Hydrolysis

Min

eral.

Min

eral.

Minerogenic Particles Submodel

LPIP

Phytoplankton Growth/Biomass

issues – sources/sinks representation, metrics

metric of phytoplankton biomass– carbon (POC; organic

matter) model – most models– Chl-a, secondary


Phytoplankton Biomass

GrazingMussels?Zooplankton?

Settling

Respiration

Growth

ambient drivers of growth1. irradiance2. temperature3. nutrients

phosphorus

nitrogen xsilica ?

details may track:CE-QUAL-W2Lake2Kother models

Tentative Timeline


No. Component Description 2015 2016

Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

1 Individual Constituents Modeling Analyses NOX, DOC, TP, SUP, POC

2 Inlet Channel – adjustment to loads

3 Minerogenic Particle Submodel

4 Optics Submodel

5 Nutrient-Phytoplankton Submodel

6 Linking of Submodels

The End

Questions?


Next Steps1. TAC and MEG comments and responses (ASAP)2. complete Phase I report – December 15, 20143. prepare modeling amendment for QAPP

• items 1-3 in parallel4. commence modeling program beginning of 2015



Optics Submodel for Cayuga Lake

Specification of symbols

OACs – optically active constituentsChl-a – conc. chlorophyll aPOC – particulate organic carbonPAVm – projected area conc. of minerogenic particlesOACbx – OAC for bxISPM – conc. inorganic suspended particulate materialb() – spectral scattering coefficientaCDOM - absorption coefficient for CDOMax*() – spectral absorption cross-section for component xbx*() – spectral scattering cross-section for component x ax – absorption coefficient for component x

bx – scattering coefficient for component xOACax – OAC for axOACbx – OAC for bxa() – spectral absorption coefficientb() – spectral scattering coefficientc() – spectral beam attenuation coefficientSD – Secchi depth - coefficient for SD radiative transfer functionKd() – spectral downwelling attenuation coeff.KL – downwelling attenuation illuminance coeff.Ko(PAR) – scalar attenuation coeff. for PARcL – beam attenuation illuminance coeff.

considerations for phase ii water quality modeling cayuga lake · 05/11/2014 · linking of...

Documents