cbeo year 3 planning
DESCRIPTION
CBEO Year 3 Planning. Rebecca Murphy Dec. 9, 2008. CBEO research. Interpolation methods Test and compare kriging methods in Bay 2D single depth (x,y) 2D along main stem (z,y) 3D kriging 3D kriging using water distance …3D (or 4D) kriging using travel time (extension project…) - PowerPoint PPT PresentationTRANSCRIPT
CBEO Year 3 Planning
Rebecca Murphy
Dec. 9, 2008
CBEO research
Analyze changing relationship of hypoxic volume to N load over 50 years. Hypotheses:• Artifact of interpolation method• Artifact of sampling density
Calculate and analyze Bay stratification:• Its relation to hypoxic volume• When, where and why changes in stratification are occurring
For use by others in CBEO science questions
and network tools
Interpolation methods• Test and compare kriging methods in Bay
2D single depth (x,y) 2D along main stem (z,y) 3D kriging 3D kriging using water distance …3D (or 4D) kriging using travel time (extension project…)
• Use water quality model as covariate in all of these
July 1-15, 2004 DO interpolated
Science Q: Why the shift in relationship between nutrient loading and hypoxia after 1980?
Artifact: interpolation leads to miscalculation of hypoxia volume
1. 2D kriging main channel data: expand laterally to get 3D
Hypoxic Volume (DO<2mg/L) Interpolated from Average July Datasets
0
2
4
6
8
10
12
14
1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1999 2004
Hyp
oxic
Vol
ume
(km
3 )
0
50
100
150
200
250
300
350
400
450
Spr
ing
N lo
ad (
Mg/
day)
Kriging DO<2 mg/LSpring N load
Hypoxic Volume (DO<2mg/L) Interpolated from Average July Datasets
0
2
4
6
8
10
12
14
16
Hyp
oxic
Vol
ume
(km
3 )
1949 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005
Kriging DO +/- 1 SD .Kriging DO<2 mg/LHagy et al. 2004
1963: There were no samples taken south of the Potomac anywhere in the Bay from July-Aug. This means that there is large uncertainty in the statistical procedure in that region. Considering some methods of accounting for this, since we know the DO won’t be too low in the southern Bay.
Analysis of Hypoxic Volume
Shift in hypoxic volume related to N load does not appear to be an artifact of interpolation method
Early July appears to have the strongest increase in hypoxia/N of each 2 week period in the summer
Early July Cruises: Hypoxic Volume (<2 mg/L) per Spring N Load
y = 1.0589x - 2078.8
R2 = 0.208
0
10
20
30
40
50
60
70
80
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
Hyp
oxi
c V
olu
me
/ N
loa
d [
km3
/(G
g/d
ay)
]
P-value on slope = 0.03
Late July Cruises: Hypoxic Volume (<2 mg/L) Per Spring N-Load
y = 0.1323x - 228.52
R2 = 0.0044
0
10
20
30
40
50
60
70
80
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
Hyp
oxic
Vol
ume
/ N lo
ad [k
m3/
(Gg/
day)
]Late JulyLinear (Late July)
Science Q: Why the shift in relationship between nutrient loading and hypoxia after 1980?
Artifact: interpolation leads to miscalculation of hypoxia volume
1. 2D kriging main channel data: expand laterally to get 3D
2. 2D kriging at single depths: sum volumes from single depths to get 3D
Early July: Hypoxic Volume (DO <2 mg/L) Per Spring N-loadUsing Different Interpolation Methods
y = 1.0589x - 2078.8
R2 = 0.208
y = 1.0358x - 2029.4
R2 = 0.2127
0
10
20
30
40
50
60
70
80
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
Hyp
oxic
Vol
ume
/ N lo
ad (
km 3 /(
Gg/
day)
)
2D profile interpolationsSingle depth slices summedLinear (2D profile interpolations)Linear (Single depth slices summed)
Science Q: Why the shift in relationship between nutrient loading and hypoxia after 1980?
Artifact: interpolation leads to miscalculation of hypoxia volume
1. 2D kriging main channel data: expand laterally to get 3D
2. 2D kriging at single depths: sum volumes from single depths to get 3D
Artifact: interpolation leads to miscalculation of hypoxia volume
3. Sub-sample post-1984 data as if it was pre-1984 data and interpolate
DO samples taken July 9-12, 1970
-
2
4
6
8
10
12
14
1949 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005
Hyp
oxi
c V
olu
me
(km
3 )
0
2
4
6
8
10
12
14
Subsampled avg hypoxic vol +/-SD
.
Avg subsampled set hypoxic vol
Full data set, hypoxic vol
July Mainstem Hypoxic Volume (DO <2 mg/L): Kriging Mainstem Data with Sample Density Analysis
Process Hypothesis to Hypoxia Q: Change in StratificationLooking at:
Long term trends in volume of water below pycnocline in July Long term trends in interpolated salinity and temperature data Correlations between pycnocline volumes, DO, temp, salinity, flow
upper and lower pycnocline
Early July Cruises: Hypoxic Volume and Pycnocline Volume
y = 0.1354x - 240.47
R2 = 0.1556
y = 0.176x - 343.04
R2 = 0.0997
y = 0.1761x - 326.71
R2 = 0.2202
-
5
10
15
20
25
30
35
40
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
Vo
lum
e (k
m3 )
Vol Below Upper Pycnocline
Vol within pycnocline
Hypoxic Vol (<2 mg/L)
Linear (Vol Below Upper Pycnocline)
Linear (Hypoxic Vol (<2 mg/L))
Linear (Vol within pycnocline)
p-value = 0.06
p-value = 0.02
p-value = 0.14
Early July: Hypoxic Volume and Water Temperature
y = 0.0497x - 72.877
R2 = 0.0796
-
5
10
15
20
25
30
35
40
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
Vo
lum
e (k
m3 )
20
21
22
23
24
25
26
27
28
29
30
Su
rfac
e T
emp
(C
)
Volume below upper pycnocline
Interpolated Hypoxic Volume (<2 mg/L)
Interpolated Avg Surface Temp (C)
Linear (Interpolated Avg Surface Temp (C))P-value = 0.19
Identify Locations with Temperature Change Map shows significance level of non-
zero slope in the regression for each station:
Temperature = + Year +
(for surface temperature measurements from each early July, 1984-2006)
Increase in surface temperature appears to be in northern part of Bay
CB1.1
CB3.3C
Summary: Hypoxia and Stratification With improved interpolation, we still observe increased hypoxic
volume per nitrogen load in recent years
Increase in hypoxic volume appears to be strongest in early July
Chesapeake Bay stratification is increasing in early July
This could be a reason for the increased hypoxic volume
Surface water temperature appears to be increasing in early July, and is strongly correlated to hypoxic volume
Temperature could be the reason stratification is increasing
Temperature could be affecting hypoxic volume through stratification OR other means (solubility, increased phytoplankton growth rates, etc)
Year 3 Plan 3D kriging/water distance
Investigate temperature and pycnocline changes and relations to hypoxic volume Recent Climate Change Report for Bay (CBP STAC, Oct 08) Analyze suspended sediment, clarity, and chlorophyll (ideas from student
meeting) Analyze lateral stratification and temperature Pycnocline trends using 3D kriging Model results and regressions
Collaborations with others, including: Kemp, DiToro teams for ideas on pycnocline analysis Jeremy: Interpolations of N, P Jeremy: Analysis of model results Jen: Interpolations of DO at benthic sites
Other CBEO support Testbed organization/documentation/new data GEON data and tool transfer Spatial querying and creating user-friendly model data queries
Year 3 Plan Possible CBEO-related papers:
Hypoxia trend with multiple interpolation methods and digging in more to post-1984 (possibly partner with analysis of N trend)
Interpolation comparisons/method development studies One submitted 3D kriging and water distance possible
Hypoxic volume trend relates to temperature and stratification