E. Minerogenic Particulate Phosphorus Model

Transcription

1 E. Minerogenic Particulate Phosphorus Model - the unavailable component (PP m/u ) - development, testing, and application 1

2 Outline 1. background, supporting elements, and an empirical PP m/u model 2. conceptual mechanistic PP m/u model, input considerations and development 3. model performance 4. model analyses 5. applicability, advancements, and summary 2

3 Outline 1. background, supporting elements, and an empirical PP m/u model 2. conceptual mechanistic PP m/u model, input considerations and development 3. model performance 4. model analyses 5. applicability, advancements, and summary 3

4 Unavailable minerogenic particulate phosphorus (PP m/u ), other P fractions, and bioavailability total P P fraction bioavailability (assays) total P total dissolved (TDP) (<.45 µm) particulate PP SRP SUP PP o PP m complete mostly mostly limited - (Prestigiacomo et al., 215) soluble reactive soluble unreactive organic particulate minerogenic particulate partitioning PP m = PP m/u + PP m/a PP m/u PP m/a PP m f BAP for PP PP m/a none complete minerogenic unavailable particulate minerogenic available particulate, small fraction here 4

5 5

6 NYSDEC/TAC meeting, Jan. 15, 214 Implications for Water Quality Model Structure PP needs to be partitioned according to PP o and PP m 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 based on: Effler et al Inland Waters 4: consistency with implemented tool 1/15/14 6

7 Independent estimates of PP m/u and the ratio PP m/u :PAV m from an empirical model: A test for the dynamic mechanistic model development and testing for the lake documented (Effler et al. 214) stoichiometry based model for lake PP, PP o, and PP m/u Chl chlorophyll a PP = PP o + PP m/u PP = (PP o :Chl) Chl + (PP m/u :PAV m ) PAV m 1.53 stoichiometric 7.1 mg/m 3 ratios supports the partitioning of PP - from optimization analysis of lake observations, temporal uniformity in stoichiometric ratios invoked 7 PP = PP o + PP m/u

8 Implications of PP m/u concentrations in lakes and related challenges for its modeling TP noteworthy contributions of PP m/u compromise TP concentration as a metric of trophic state because of the unavailability of PP m/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 (PAV m ) runoff event effects important runoff events expected to promote higher PAV m and thereby PP m/u challenges to model PP m/u in lakes, quantification of: 1. delivery, transport, and fate of minerogenic particles (PAV m ), and 2. P associated with these particles (PP m ) and its bioavailability (PP m/u vs PP m/a ) pelagic shelf PP m/u 8

9 Outline 1. background, supporting elements, and an empirical PP m/u model 2. conceptual mechanistic PP m/u model, input considerations and development 3. model performance 4. model analyses 5. applicability, advancements, and summary 9

10 Extensive technical program support for mechanistic PP m/u model for Cayuga Lake lake monitoring LSC program , plus 213 P fractions, Chl-a, PAV m (size classes and type) primary tributaries P fractions, sediment (ISPM, SPM), PAV m (size classes) bioavailability assays (NYSDEC recommended) submodels (1) 2-D transport, (2) PAV m (4 size classes) loading estimates PAV m and PP m/u runoff event sampling focus (NYSDEC recommended), Inlet (extra, UFI and Cornell) sediment trap (extra, UFI and Cornell) 1

11 A mechanistic model for unavailable minerogenic particulate P (PP m/u ) for Cayuga Lake: Background P associated with minerogenic particles (PP m/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 PP m/u has been developed and tested, as described here supporting components long-term and intensive 213 monitoring LSC monitoring ( ) bioavailability assessments loading estimates, PAV m and PP m/u transport and PAV m submodels PP m/u TP pelagic shelf 11

12 Conceptual model for dynamic mechanistic PP m/u model 12 tributary inputs transport submodel Gelda et al. 215a PAV m submodel Gelda et al. 215b PAV m/n loads PAV m/n PP m/u PP m/u :PAV m lake particles sink processes for PAV m PAV m submodel = f(q,pp and PAV m observations); i.e., varies PP m/u loads transport submodel PP m/u apportioned to the PAV m size classes PAV m/n according to their contributions to total PAV m model features PP m/u :PAV m of delivered particles subject to variation in-lake behavior of PP m/u is assumed to be that of PAV m with which it was associated upon entry in-lake dynamics of PP m/u reflect external load of PAV m/n, the dynamics of the PP m/u :PAV m ratio in the tributaries and the progression of in-lake PAV m loss processes

13 Specifications for modeling PP m/u well-mixed upper waters targeted (epilimnion) linear dependency of PP m on PAV m described by Effler et al. (214), conceptual consistency partitioning of PP m according to size class contributions PP m from tributary inputs of PP according to: PP m = PP (ISPM:SPM); ISPM:SPM generally >.9 PP m subject to temporal variability in tributary inputs according to variations in the PP m :PAV m ratio tributary loads of PP m from PAV m load estimates PP m load = PAV m load (PP m :PAV m ) PP m/u loads incorporate tributary-specific bioavailability results PP m/u load = (1-f BAP ) PP m load supported by rapidity of the transformation in-lake behavior of PP m/u equivalent to that of the PAV m with which it is associated f f BAP,t :f BAP,t BAP f BAP Biossay Progression (d) f BAP = fraction bioavailable for completed bioassay f BAP,t = fraction bioavailable through progression of bioassay 13

14 ISPM:SPM Considerations for modeling PP m/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. 27) ISPM (mg/l) PAV m (m -1 ) ISPM ISPM:SPM origins of variance in dependencies e.g., character of stream banks Q (m 3 /s) Fall Creek examples 14

15 Consideration for modeling 1 PP m/u : Tributary y = 4.3x 3.7 conditions, dependence of PP on PAV m strong positive dependence of PP in streams on PAV m most of PP is as PP m/u the P content of clay mineral particles delivered T n PP ( g/l) Fall Creek linear fit (b) PAV m (m 1 ) Salmon Creek 3 linear fit y = 6.62x 38.6 R 2 =.99 (int. sig.) R 2 =.96 (int. not sig.) 2 3 associated with PAV m during a runoff event large quantities of PAV m and associated P (PP m ) are delivered PAV m (m 1 ) Fig. 4c from Peng and Effler 215 (c) 15

16 Specifications for modeling PP m/u : Tributary conditions and loads temporal patterns runoff events dominate sediment loading to the lake PAV m load Q (m 3 /s) PP m/u (kg) (1 6 m 2 ) FC PAV m Inlet PAV m FC PP m/u Inlet PP m/u FC PP m A M J J A S O Apr May Jun Jul Aug Sep Oct 213 Fall Cr. Q time series 213, prominent runoff events cumulative format for PAV m and PP m/u loads early July and August runoff events dominate 16

17 PP:TP Specifications for modeling PP m/u : Tributary conditions and loads positive dependencies on stream flow (Q)-PP, PP:TP, PP:PAV m PP (mg/l) PP:PAV m (mg/m 2 ) PP PP:TP Fall Cr. examples PP m :PAV m ratio supports estiamtes of PP m (and PP m/u ) loads from PAV m loads negative, but variable dependency on Q 17 Q (m 3 /s)

18 Q (m 3 /s) Specifications for modeling PP m/u : Tributary conditions and loads temporal patterns 18 PP m/u :PAV m (mg/m 2 ) PP m/u (kg) example for Fall Cr, PP m/u :PAV m and Q cumulative formats for PP m/u load and the ratio PP m/u :PAV m (mg/m 2 ) PP m/u :PAV m Q FC PP m/u Inlet PP m/u FC PP m Apr May Jun Jul Aug Sep Oct 213 FC Inlet PP m/u :PAV m highly variable, negative dependency on Q dominance of runoff events for PP m/u loads PP m loads just slightly higher, consistent with low f BAP cumulative change for ratio will drive changes in-lake relationship

19 Outline 1. background, supporting elements, and an empirical PP m/u model 2. conceptual mechanistic PP m/u model, input considerations and development 3. model performance 4. model analyses 5. applicability, advancements, and summary 19

20 Mechanistic dynamic PP m/u model performance: Targeted attributes for testing tests of consistency extent of closure with prediction of independently tested empirical PP model in-lake PP m/u :PAV m ratio in-lake PP m/u PP m/u with historic TP and PP shelf observations, following runoff events predicted PP m/u deposition on shelf from runoff event inputs with sediment trap observations 2

21 Mechanistic dynamic PP m/u model performance: Lake PP m/u :PAV m ratio vs. empirical model value predictions of in-lake PP m/u :PAV m ratio only minor shelf vs. pelagic differences averages (shelf 8. mg/m 2 ; pelagic 7.4 mg/m 2 ) closed well with single empirical model value (7.1 mg/m 2 ; Effler et al. 214) PP m/u :PAV m (mg/m 2 ) shelf pelagic empirical model Apr May Jun Jul Aug Sep Oct 213 in-lake variations (cv=.22) driven by tributary variations (cv=.46) tributary dynamics linked to variations in Q 21

22 Mechanistic dynamic PP m/u model performance: Comparison of PP m/u predictions, mechanistic vs empirical models comparisons for both the shelf and pelagic waters predictions for 213, 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) PP m/u (µg/l) 1 (a) shelf pelagic Apr May Jun Jul Aug Sep Oct

23 Mechanistic PP m/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 PAV m (sub)model Table 2 (Gelda et al. 215b) signatures of increased PP m expected on shelf following runoff events 23

24 The complications of turbidity plumes on the shelf following major runoff events issue for PP m/u model evaluation as with the PAV m model 24

25 The complications of turbidity plumes on the shelf following major runoff events issue for PP m/u model evaluation as with the PAV m model 25

26 Mechanistic PP m/u model performance: Consistency with historic TP and PP observations following runoff events variations is predicted PP m/u for long-term ( ) simulations performed reasonably well in explaining observed differences for historic events recall this is an imperfect match of P fractions performance improves somewhat with representation of the effects of PP o (phytoplankton biomass) variation also TP obs (µg/l) PP obs (µg/l) PP obs (µg/l) I =.77 S = r ² = I =.83 S =.33 r ² = PP m/u (µg/l) I = 57 S =.49 r ² = numbers correspond to the various historic runoff events listed in previous table (Table 2, Gelda et al. 215b) PP 1.53 = (PP o :Chl) Chl Effler et al. (214) 26 PP m/u + PP o (µg/l)

27 Consistent mechanistic PP m/u model performance for local deposition from runoff events comparison of simulations of deposition of PP m/u (i.e., associated with PAV m ) 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 (m 3 /s) DF PP (mg/m 2 /d) 1 (a) Fall Creek (b) Site 2 predicted PP m/u 4 deposition 2 traps May Jun Jul Aug DF PP downward flux of PP 27

28 Consistency of the mechanistic and empirical models in representing contributions of PP m/u and PP o to PP for shelf vs pelagic waters recall, Chl-a (e.g., phytoplankton biomass) not significantly different on shelf vs pelagic waters recall PP = PP o + PP m both models higher PP m/u values on shelf vs pelagic greater variability on shelf median values shelf vs pelagic waters, contributions of PP m/u 2 vs 11%, mechanistic 16 vs 8%, empirical reasonably good closure % Contribution to PP shelf PP o pelagic empirical model PP m/u PP o comparisons mechanistic model PP m/u PP m/u much greater, and PP o much smaller contributions on the shelf for the extremes of runoff events Key: 95 th percentile 9 th percentile 75 th percentile mean median 25 th percentile 1 th percentile 5 th percentile 28

29 Outline 1. background, supporting elements, and an empirical PP m/u model 2. conceptual mechanistic PP m/u model, input considerations and development 3. model performance 4. model analyses 5. applicability, advancements, and summary 29

30 Analyses with the tested PP m/u model: Dependence of shelf response on runoff event magnitude, PP m/u vs. Q peaks 3 Fall Cr. peak Q for the earlier ( ) runoff events corresponding predicted peak PP m/u at Site 2 on shelf strong positive dependency of PP m/u on event magnitude sources of variance in the relationship-variations in ambient mixing, limitations in peak Q defining external PP m/u loads Q (m 3 /s) Max. PP m/u (µg/l) examples Apr May Jun Jul Aug Sep Oct Fall Cr. Peak Q (m 3 /s)

31 Analyses with the tested PP m/u model: Interannual variations in predicted PP m/u explains much of the interannual and spatial variations in TP total P (TP) concentration is a trophic state metric, guidance value of 2 µg/l as summer average, in NY PP m/u variations important in regulating interannual and shelf vs pelagic differences in TP PP m/u lower at pelagic site each year interannual differences in PP m/u between sites explained much of the year-to-year differences in TP (45%) these interannual differences in PP m/u explain 47% of the year-to-year differences in TP at both sites P (µg/l) s TDP PP o PP m/u TP limit p PP m/u based on mechanistic model predictions 31

32 Analyses with the tested PP m/u model: Interannual variations in day counts of elevated concentrations on the shelf multiple PP m/u concentration thresholds chosen for Jun-Sept. interval, high flow events represented by number of days Fall Cr. Q was in the upper 1% high PP m/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 PP m/u predicted for 26 and 211 Days Days Days Days Days based on model simulations for the period upper 1% flow (a) PPm/u > 5µg/L (b) 5 PPm/u > 1µg/L (c) 5 PPm/u > 2µg/L (d) 25 5 PPm/u > 5µg/L (e)

33 Outline 1. background, supporting elements, and an empirical PP m/u model 2. conceptual mechanistic PP m/u model, input considerations and development 3. model performance 4. model analyses 5. applicability, advancements, and summary 33

34 Applicable findings from the PP m/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 PP m/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 213) more runoff events and severity of the events 34

35 Summary P associated with minerogenic particles (e.g., PP m ) 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 PP m (PP m/u ) has been successfully developed and tested higher PP m/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 PP m/u and PP m/u :PAV m with those from an independent empirical model (Effler et al. 214) consistency of predicted PP m/u shelf deposition with sediment trap observations a consistent partitioning of the PP pool between PP m/u and PP o (phytoplankton) for historic observations consistency of post-runoff event TP and PP observations with PP m/u predictions this advancement in modeling is invaluable and appropriate for large initiatives addressing the trophic state issue 35