1. data analysis 2. model formulation 3. calibration status...revisiting kd story. note: low kd...
TRANSCRIPT
1
1. Data Analysis
2. Model Formulation
3. Calibration Status
4. Moving Forward
2
These data provided guidance into the selection of the algal groups to model, as well as
provided calibration data (Thanks to Harold Marshall and Todd Egerton)
3
“borrowed” from Todd’s presentation at the March 2014 SAP meeting
4
Challenges in developing light attenuation relationships as a function of water quality
variables.
Want to check Poole-Atkins coefficient – Kd = P.A./SDD -> P.A. = Kd*S.D.D. = note variability
– mean value of 1.5
Regression slope for Kd = f(Chla) = 0.0165 (models usually 0.010-0.025 – I’ve used 0.017
pretty consistently), but low R2
Regression slope for kd = f(TSS) = 0.0565 R2=.526 Multiple linear regression for Chla
(0.0133), TSS (0.0528), CDOM (0.243) yields R2=0.6
Good R2 for Kd vs. Turbidity and TSS vs. Turbidity, but note the product of the regression
slopes 0.0755 * 0.726 = 0.055 which is similar to slope for Kd vs TSS
5
We performed similar regressions for each salinity regime – similar type of results
6
As will be presented in the calibration section, the spatial/temporal distribution of light
attenuation data (based on monthly grab samples) were quite variable and made
calibration of challenging. If the river was sampled just after a storm event with high light
attenuation do you use that high value for the entire month. Decision was made to use
median Kd data. We looked at both available Kd and Kd estimated using 1.5/SDD. First,
however, the observed Kd were “corrected” for chla -> Kdbase = obs Kd – 0.017*Chl-a
7
Plot of Kdbase – spatially variable, but constant in time
8
Looked C:Chla ratios and C:N, C:P and C:Si ratios by river segment
9
.wn 21Z- r1 vorr 0.4411 igtvr2.1„;. 1.170
w.. we11-oa12 1$7.7a-"
00DA I
1.0 2.0 3.0 4.0 5.0
BioSi (mg/14
10
flops • 8.1814 Puort • 13.76
wrM = ken - x17,1;°-
slope • 19.2708 1-score • 49.316 car Goer, • .0.41,
0 0 .0 I .0 1.0 3.0 4.0 50
BioSi (mgIL)
11
Data Analysis 0
PN vs PC 20 20 Chl-a vs PC
' st , )L i L
At, 20 PP vs PC
'.4.1.7i:ir-P ,,,ii! s4epe . 01.411,3 Irma°:4. V-°4-" 16 16
14.ve = 0.4004 ita
- 0.150
J 12 01 E 0 8 0.
4
1( 12 rn E LI 8 0.
4
;) 12 04 E (-) 8 0-
4
0
MOM 14.81.66 54.169
.60441 6 010 6,,,54:804:47; 6 014 16
slope • .999 9900 .... • 54 100 err coal • 599990
p.eacape • 443900 .9s9980.000
16.0 me,.. • 59499 0 0
ee. • 54169 cc, c•501 • 599.990 V.Z %%Vain"'
80 100
40 80 120 160 200 CHLA (ug Chl/L)
0 0 0.6 1.2 1.8 2.4 3 0 PN (Foga.)
0 0 0.2 0.4 0.6 0.8 1 0 PP (mgIL)
01001 YS PC
12.0
E
O.
4.0
0.0 00 1.0 2.0 3.0 4,0
BioSi (mgiL)
VSS vs.T56
E 50 40 03
20
100 200 300 400 TSS (mglL)
5.0
20.0
500
Elizabeth Stations, James River
12
What caused the change in PP in the upper freshwater portion of the James R.? Change in
methodology?
13
Google map under the USGS Nav chart, shows the locations of two of Iris Anderson’s
sediment flux stations – located in 1 and 2 m of water. However, model grid includes
channel and model depths are 3 m. Therefore, not good for model vs data comparisons,
since light attenuation limits light reaching bottom and precludes development of benthic
algae. Will consider using nearby model segment in Tar Bay
14
Overview of RCA eutrophication model structure. Very similar to USACE/CBP CE-QUAL-
ICM, but we are not directly modeling CDOM, TSS, not benthic filter feeders
15
Overview of structure of the HAB model
16
Algal growth rates as a function of temperature with current model coefficients
17
Effect of salinity on reducing freshwater and marine algal growth rates
18
Some calibration results. Top panel: green is 5-day average Chl-a, blue is max chla
computed within the 5-day period and red is the min computed within the 5-day period.
Revisiting Kd story. Note: low Kd values in spring 2009 despite high flows and high values
of Kd in late summer and fall 2009 despite lower flows. Led to decision to use “average” Kd
rather than time-variable Kd
19
Plot PO4 and NH4 using logs in order to better see if nutrient limitation is occurring)
Michaelis-Menton coefficients for PO4 are about 0.001 mgP/L and for NH4 are about 0.010
mgN/L. If PO4 concentrations are 0.004 mgP/L then you have a 20% reduction in growth
rate -> PO4/(KmP+PO4) = 0.004/(0.001+0.004) = 0.80
Data suggest some P limitation in upper tidal James
20
Looking to complete WQM calibration/validation in Q1 of 2015.
21
Looking to complete WQM calibration/validation in Q1 of 2015.
22
Top panel – same as earlier. Middle panel: black is total phyto C, green is diatom C, blue is
greens C, red is Microcystis and black triangles are observed algal C based on Todd
estimates
23
Similar analysis for lower estuary (ignore purple line – was exploring an alternative method
for determining algal Chla based on a paper by Finenko et al.)
24
LEGEND Alark1Su.ow.
..:''':"E Results for Station LE5.3 Segment 279,93
25
2010 2009 2008 2014 2012 2011 2007 2013
----- LEGEND -
• Su.. NY • Bottom 0.
Results for Station LE5.3 Segment 279,93
Preliminary Results
0.20
0.16
7,7 O. 0.12
0,08
2 0.04
9 00 20
1.8
2' 1 .2 z 0.0
2 0.4
0
:7 r.
VE
O
4
3
2 0
• •
•
•
•
•
• •
26
8
4 0 0
C' 2
0
Run Number 43JPIA
----LEGEND-- - Ito. Surface - 11.60 Boson
• SMz OM ✓ V.v. Ovt.
Results for Station LE5.3 Segment 279.93
Preliminary Results
020
010
a. 0,12
0.08 0
• 0.04
090
20
1.0
2 1.2
E o O
0.4°
00
10
27
Looking to complete WQM calibration/validation in Q1 of 2015.
28
Similar analysis for LE5.5-W – here we have phyto C estimates from Todd. Question – do
we believe phyto C based on biovolume and cell counts? Note how low phyto C estimates
are despite reasonable fit to chl-a. Would have to have much lower C:Chla ratio to fit both
chla and phyto C.
29
Area needing work – very low cochlodinium (red)
30
Another issue in Lafayette – low DO not captured by model. Where is O2 demand coming
from?
31
Looking to complete WQM calibration/validation in Q1 of 2015.
32
Looking to complete WQM calibration/validation in Q1 of 2015.
33
Can low DO be driving PO4 release from the sediment – note increase in summer. Likely
not due to loads since flows would be low, but maybe due to low DO
34
Model output can show if algal growth is light or nutrient limited. Above plot show model
output for a surface layer segment. Blue line is reduction in algal growth rate due to light
limitation (low solar radiation in winter, high in summer), hence sinusoidal shape. Green
line is reduction in algal growth rate due to nutrient limitation. This needs to be depth
integrated to get true reductions in growth rate for a segment
35
Type of skill assessment we are looking to perform
36
What needs to be done
37
38