Transcript Slide 1

Validation and improvement of Effective Monthly Recharge (Wem) model
for simulating wetland hydrology
G. Richard Whittecar, Tracy Thornton, John Smith, John McLoed, and Brian Jolemore
Department of Ocean, Earth and Atmospheric Sciences, Old Dominion University, Norfolk, Virginia
ABSTRACT
RESEARCH GOALS
The Effective Monthly Recharge (Wem) model generates a synthetic hydrograph of water table elevations for
toe-slope wetlands dominated by groundwater discharge. Use of post-construction water-level data
gathered for 110 months (2000-2006) at a 2.6-acre mitigation wetland in Virginia Beach permits validation of
model parameters calibrated using pre-construction data. Model improvements based on evaluation of error
sources and sensitivity analyses greatly improve the repeatability and reliability of model parameters
developed by future model users. Other changes to the model decrease its complexity making it more
flexible and adaptable for inclusion in a package of water-budget models under development for use by
wetland design professionals.
One goal of this research is to evaluate the pre-construction model calibration by using post-construction
monitoring data. A second goal is to test and develop the uses of the Wem model to make its application
easier to use and adaptable in more situations.
MODEL DEVELOPMENT
In order to improve the model, we addressed issues related to both data collection errors and calculation
procedures. After examination of data collection records and the history of daily rainfall, we established the
following criterion: eliminate any water level data if more than 0.75 cm rain fell in the month prior to data
collection. In order to make more precise and repeatable comparisons of model results with field data, we
started using correlation coefficients (R2) to test model runs, replacing the original curve-matching
procedure. Users can now compare R2 values generated by multiple model calculations using both n
(number of months) and d (decay factor) values to identify more accurate model parameters.
PREVIOUS WORK
Study Area: VDOT designed a mitigation wetland for a site on a relict barrier island where only precipitation and
groundwater flow through the sandy soils can provide water. Fourteen months of water level data collected in
1999-2000 were used to design an excavation pattern for maximizing seepage into the wetland .
Compare: Matrix of model results by number of months (n) and decay factor (d) used in present model. Also, pattern of
Well#10
& Wem12 (2001-2007)
Wem and the water table fluctuations over time, such as used in the previous curve-matching
procedure.
335
10
11
12
13
N
20 m
d
marsh
h
itc
2.5
315
Well#10 (cm)
Nov. 1, 1999
3.5
Correlation coefficients (R2) by n and d
1.5
295
0.5
275
Elev#10 cm
Wem12
landfill
255
-1.5
Oceana Ridge
Groundwater Flow
Month
Original Wem Model : The Effective Monthly Recharge model estimates past water-table fluctuations.
Monthly groundwater recharge values (precipitation minus ET) are combined in time-weighted averages.
Model parameters were calibrated by comparing shapes of curves generated for different input values.
Effective Monthly Recharge Values and East Pond Levels
A time-weighted recharge value
Normalizing
factor
n
S
a=1
a-1
d
Each
month’s
recharge
(Ppt - ET)
Responsedecay
factor
(<1.0)
n= # preceding months
340
330
330
320
320
15
a-1
d
April-September
340
310
310
y = 13.813x + 282.25
2
R = 0.7169
10
Well#10 (cm)
D
Wmox
S
a=1
Well#10 vs. Wem12
Wem-2mo curve
matches best
20
Centimeters
Wem =
n
Wetland designers using this model will ask “When is the best time to measure water levels?” and “Are any
of these data I’ve just collected better than others?” To address these issues we plotted data according to
standard NRCS wet/normal/dry criteria (3-month window), and by seasonality.
5
0
1
2
3
4
5
6
7
8
9
10
11
12
300
Dry
290
Normal
Wet
280
Total
13
Wem-1 mo
Wem-2 mo
Wem-3 mo
Wem-4 mo
280
270
R2 = 0.87
260
250
Wem-6 mo
y = 14.57x + 273.55
2
R = 0.8677
290
R2 = 0.72
260
May 1999 - May 2000
300
Linear (Total)
270
-5
Well10 (cm)
Effective Monthly Recharge: Wem
-2.5
Jul01
Aug01
Sep01
Oct01
Nov01
Dec01
Jan02
Feb02
Jun02
Aug02
Oct02
Nov02
Dec02
Jan03
Feb03
Apr03
May03
Jun03
Aug03
Sep03
Oct03
Nov03
Dec03
Jan04
Mar04
Apr04
Jun04
Jul04
Sep04
Oct04
Nov04
Dec04
Jan05
Feb05
Mar05
Apr05
May05
Jun05
Jul05
Aug05
Sep05
Oct05
Nov05
Dec05
Feb06
Mar06
Apr06
May06
Jul06
Aug06
Oct06
Nov06
Dec06
Jan07
Feb07
Mar07
235
From Mixon and others, 1989
Ws
250
East Pond
240
-3
-2
-1
240
0
1
2
3
4
5
-3
-2
-1
0
Wem12(d=0.89)
Initial Results: Model calculations that use all available weather data and calibrated model parameters
produce a synthetic hydrograph – an estimate of past water-table fluctuations. The elevation selected for
the new wetland by this method received groundwater discharge consistently during even the driest months.
Wem-2 mo Values, 1952-95, Oceana, Va
15
Deep (central) Pond looking East
after wetland construction
10
(cm)
Effective Monthly Recharge
-0.5
1
2
3
4
CONCLUSIONS
1. Analyses of post-construction monitoring data from the site verified the original calibration for the study area.
Originally done using pond elevations, model runs with well data generate stronger correlations.
2. Use of linear regression correlation coefficients makes the model more robust and flexible, mostly by allowing
for more precise comparisons of model results that use different input values and model parameters.
3. Our analyses to date suggest that the best field data come from Spring and Summer months of any wetness,
plus other months that are “dry.”
ACKNOWLEDGEMENTS
0
-5
Wem during driest months of
calibration 1952-1995
in 1999-2000
Consistent groundwater seeps even
during dry months
5
Wem12 (d=0.89)
5
-10
Wem12 (d=0.89)
Oceana VDOT/Navy Mitigation Site
We acknowledge the assistance of M. Richardson (ODU) and the guidance of T. Wynn and W.L. Daniels
(Virginia Tech) and M. Rolband and L. Giese (Wetland Studies and Solutions, Inc.). Funding provided by a
grant from the Peterson Family Foundation.