Transcript Slide 1

CHRP Narragansett Bay project:
Box Modeling progress
M. Brush
Aug 2008
(1.) Physical box model (after Officer 1980
and Swanson & Jayko (1988) …
(exchanges computed from freshwater inputs
and salinity distributions)
No horizontal exchange
here – a potential
weakness and can revisit
– but right now I can’t see
how to adjust the box
model eqns to let this
happen.
1
2
Exchange from 3 to 4 & 5
is based on the relative
volume of freshwater in 4
& 5.
3
4
6
7
5
8
9
11
12
13
14
RI Sound
10
Schematic of exchanges:
Q = advective (grav. circ.)
E = non-advective (tidal)
Subscript v denotes vertical exch.
Care is taken to
divide bottom area
into both the surface
and bottom layers
based on
hypsographic
analysis
General equations.
Note freshwater inputs
are divided between
surface and bottom layer.
I do that by computing
the volume of freshwater
in each layer and using
the ratio to divide
incoming river flow.
The Hansen-Rattray
parameter (n) divides
exchanges between
advective and tidal
components. I will
compute this later using
the Officer & Kester
method but now have it
set at 0.5
All freshwater inputs are
time-averaged as it is the
longer time scale that
drives estuarine
circulation. Currently the
averaging window is 10
days. I need to test this.
Freshwater Inputs
Currently only rivers and precip. Need to add evaporation and
WWTFs – does anyone have these time series?
USGS daily flows compiled for
the only gauged rivers in the
watershed:
• Blackstone
• Pawtuxet
• Moshassuck
• Woonasquatucket
• Taunton
• Ten Mile
• Hunt
Precip data from TF Green (can
modify if we want but this was
the quickest route for now)
Time period:
2001 – 2006
(time when most buoys operational)
(saving 07-08 for independent model validation)
Freshwater Inputs
• Watersheds were
delineated in GIS for each
NarrBay box.
• Flow at each of the 7
USGS gauges (red circles)
was increased to account for
the ungauged portion on an
areal basis.
• Flow in the completely
ungauged watersheds was
computed from watershed
area and flow/area for the
Hunt watershed, which is a
small, typical (as far as I
recall) RI watershed so
considered reflective of the
others.
Salinity Distributions
• For now, simple cosine
functions for annual salinity
fluctuations were developed
from Insomniacs data
supplemented with data from
the buoys, GSO dock, and GSO
Fox Is. Station.
Box 3
35
30
S3
25
• Insomniacs data at each site
on each date from 1999-2007
were averaged from 0-3m and
3m-bottom. Values were plotted
by day of year (not relative to a
specific year) and cosine fcns fit
by eye.
B3
20
S_bulr
15
B_bulr
S3
10
B3
5
0
J
F
M
A
M
J
J
A
S
O
N
D
• An example is to the right 
• See file “salinity functions.pdf”
for all boxes.
Points: Averaged Insomniac data
Thin lines: Bullocks Reach buoy data
Thick lines: cosine functions
S = surface layer
B = bottom layer
Sample Box Model Output
(units are m3/d, 1/1/01 – 12/31/06)
Q’s
Qv’s
E’s
Ev’s
Computed Residence Times
Greenwich Bay
(Box 6-7) –
seems a little
high
Entire Narragansett Bay – matches
Pilson’s estimates and variability
fairly well.
RT, DAYS
Providence River
(Box 1-3) –
Similar values to
what I recall
1/1/01
12/31/06
Providence River boxes
Computed Residence Times
all y-axes are 0 – 40 days
all x-axes are 1/1/01 - 12/31/06
Greenwich Bay (6-7) &
Mt Hope (10) Bay
Upper Bay boxes
West Passage boxes
Computed Residence Times
all y-axes are 0 – 40 days
all x-axes are 1/1/01 - 12/31/06
East Passage boxes
Box Model Summary:
(1) Running and producing reasonable exchanges and
residence times.
(2) Conserving volume
(3) Need to refine n and add WWTFs & Evaporation
(4) Further down the road:
(a) use daily buoy/PORTS salinities so more eventscale processes can be captured, and
(b) try to formulate salinity functions and pycnocline
depths as functions of freshwater flow (and
other relevant factors) so forecasts are not
tied to present conditions
(5) Has been coupled to novel ecosystem model – see
following pages …
(2.) Ecological model
PAR compiled from Eppley Lab
(thanks to Dan and Lora Harris).
We have through Jun 06. Pell is sending me the
rest and my shop will digitize.
PAR
Nutrient Loading:
We can work on this (e.g. using Don’s timeseries
or Scott’s new estimates), but for now I have used
a quick and easy solution. Also need to add in
WWTFs. Atmospheric loads are based on
historical data (eg Fraher thesis). Here is my
approach for watershed loads analogous to what I
did for the original Greenwich Bay model:
(E/m2/d)
80
Nixon et al (1995) mass balance paper:
70
• Reports annual DIN, DIP load for 5 major rivers
• I used total annual flow to estimate the average
DIN & DIP concentration in the rivers:
60
50
40
30
Blackstone
Pawtuxet
Mean, other 3
20
10
0
12/6/99
4/19/01
9/1/02
1/14/04
5/28/05
10/10/06
DIN uM
125
125
85.7
DIP uM
3.5
12.2
7.9
2/22/08
• I used the average of the Taunton, Moshassuck,
and Woonasquatucket (no WWTFs) for the
ungauged watersheds.
There is still lots to work on, but the model is up and running.
See file “Output_Aug08.pdf” for results with no biological tuning ...
The results are very encouraging! The model produces fairly typical annual cycles of
chl, din, dip, & DO, with typical down-estuary gradients.
The shape of the cycles needs a bit of work, and there is something going on
betweeen the 1st and remaining years.
Also a weird box every now and then (e.g. Box 6 – Greenwich Bay DO).
But look at the broad-scale comparison of predicted vs Insomniacs DO … (and I didn’t
tune any biological/chemical parameters). Wicked cool!