E. Minerogenic Particulate Phosphorus Model · E. Minerogenic Particulate Phosphorus Model 1...
Transcript of 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
10/22/2015 Upstate Freshwater Institute 2
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
10/22/2015 Upstate Freshwater Institute 3
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
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
10/22/2015 Upstate Freshwater Institute 4
PPm/a ÷ PPm ≈ fBAP for PP
10/22/2015 Upstate Freshwater Institute 5
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
1/15/14 Upstate Freshwater Institute 6
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
10/22/2015 Upstate Freshwater Institute 7
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)
10/22/2015 Upstate Freshwater Institute 8
pelagic shelf
TP
PPm/u
Outline
10/22/2015 Upstate Freshwater Institute 9
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
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)
10/22/2015 Upstate Freshwater Institute 10
A mechanistic model for unavailable minerogenic
particulate P (PPm/u) for Cayuga Lake: Background
10/22/2015 Upstate Freshwater Institute 11
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
10/22/2015 Upstate Freshwater Institute 12
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
10/22/2015 Upstate Freshwater Institute 13
• 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
10/22/2015 Upstate Freshwater Institute 14
Fall Creek examples
Consideration for modeling PPm/u: Tributary
conditions, dependence of PP on PAVm
10/22/2015 Upstate Freshwater Institute 15
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
10/22/2015 Upstate Freshwater Institute 16
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
Specifications for modeling PPm/u:
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
10/22/2015 Upstate Freshwater Institute 17
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
Apr May Jun Jul Aug Sep Oct
PP
m/u:P
AV
m
(mg
/m2)
0
4
8
FC
Inlet
Specifications for modeling PPm/u:
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
10/22/2015 Upstate Freshwater Institute 18
Outline
10/22/2015 Upstate Freshwater Institute 19
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
Mechanistic dynamic PPm/u model performance:
Targeted attributes for testing
10/22/2015 Upstate Freshwater Institute 20
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
Mechanistic dynamic PPm/u model performance:
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
Apr May Jun Jul Aug Sep Oct
PP
m/u:P
AV
m (m
g/m
2)
0
5
10
15
20shelf
pelagic
empirical model
10/22/2015 Upstate Freshwater Institute 21
Mechanistic dynamic PPm/u model performance:
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
Apr May Jun Jul Aug Sep Oct
PP
m/u
(µ
g/L
)
0
1
2
313 59
10/22/2015 Upstate Freshwater Institute 22
Mechanistic PPm/u model performance: Consistency with
historic TP and PP observations following runoff events
10/22/2015 Upstate Freshwater Institute 23
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
10/22/2015 Upstate Freshwater Institute 24
issue for PPm/u model
evaluation as with the
PAVm model
The complications of turbidity plumes on the shelf
following major runoff events
10/22/2015 Upstate Freshwater Institute 25
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
10/22/2015 Upstate Freshwater Institute 26
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
10/22/2015 Upstate Freshwater Institute 27
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
10/22/2015 Upstate Freshwater Institute 28
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
10/22/2015 Upstate Freshwater Institute 29
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
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
10/22/2015 Upstate Freshwater Institute 30
2013
Apr May Jun Jul Aug Sep Oct
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
10/22/2015 Upstate Freshwater Institute 31
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
10/22/2015 Upstate Freshwater Institute 32
based on model simulations for the
1999-2013 period
Outline
10/22/2015 Upstate Freshwater Institute 33
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
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
10/22/2015 Upstate Freshwater Institute 34
Summary
10/22/2015 Upstate Freshwater Institute 35
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