20080868 Ver 2_Kimmel SalinityTrends_20160708

Trends in salinity in the Tar -Pamlico River 1988-2014
David G. Kimmel', Associate Professor, Department of Biology, Institute for Coastal Science and
Policy, East Carolina University
'Present Address: Lead Research Oceanographer, NOAA, EcoFOCI group, Alaska Fisheries Science Center, 7600
Sand Point Way NE, Building 4, Seattle, WA; [e-mail]
Introduction
A program of routine measurements of Pamlico nutrients and related hydrographic
parameters was begun in 1967 by Dr. John Hobbie at N.C. State University. That study, which
continued through 1973, was supported by funds from two sources: 1) the Office of Water
Research, U.S. Department of the Interior, through the University of North Carolina Water
Resources Research Institute, and 2) Texas Gulf Sulfur Company. The initial objective was to
study the effects of phosphorus from the phosphate mining operation (Copeland and Hobbie
1972). Later, when it became obvious that phosphorus was not the only factor controlling algal
growth in the estuary, the scope of the project was broadened to include nitrogen. After the
N.C. State University sampling ended, there was an 18 -month lapse until East Carolina
University began a new program in 1975 and this program continues to the present. The
purpose of this study was to determine if there exists a trend in Tar -Pamlico River salinity over
the course of the data record.
Methods
Data were compiled from the Tar -Pamlico river transect data that have been collected
since 1975. At the time of this report, data have been compiled and digitized from 1988-2014.
Data were compiled from 6 stations that have been part of the regular monitoring of the Tar -
Pamlico River (Figure 1; Table 1). Salinity has been measured using a YSI meter. Also used in the
analysis were discharge data from the Tar River at Tarboro
(http://waterdata.usgs.gov/nwis/uv?site no=02083500) and Palmer Hydrological Drought
Index (PHDI) data from http://climate.ncsu.edu/climate/climdiv.php. The PHDI was selected
from the Central Coastal Plain climate region. PHDI is a long-term drought index that takes into
account hydrological impacts and indicates the relative "wetness" or "dryness" of a region and
is based primarily on soil moisture.

Trend testing was conducted using the Seasonal Kendall test for trend (Hirsch et al.
1982; Millard and Neerchal 2001). This test tests for trend across seasons (in this case, months)
when the trend is in the same direction from year to year. This test allows an overall trend
(slope) to be computed while accounting for seasonal variability in the parameter of interest.
Salinity varies significantly over the year, thus the reasoning for choosing this particular trend
test.
Results and Discussion
Each station represents a different salinity region of the estuary, having means that
differ (Table 2). The mean salinities are relatively stable across all years as indicated by the
similar standard deviations seen at all stations (Table 2). The similarity among standard
deviations suggests that the degree of variability in salinity as each station is not appreciably
different compared to any other station. In other words, each station may act as a proxy for
salinity trends of the entire estuary. Even stations at the landward and seaward end of the
estuary appear to be no less variable than those in the central, mesohaline area. The range of
salinities experienced at each station is large (Table 2). This is expected as the Tar -River is
primarily a wind -driven system and is subject to events in the form of strong nor'easters in the
fall and winter and extra -tropical cyclones (hurricanes) during the summer and fall.
Tar River discharge showed distinct seasonality from year to year (Figure 2). Tar River
discharge had weak negative correlation with time as well as a negative slope (Table 3). Salinity
also showed distinct seasonality (Figure 3) and was negatively correlated to discharge (R = -
0.34, p = 0). This correlation was not as strong as might be expected, given the clear
relationship between river discharge and salinity; however, the role of wind in water movement
as well as some distance (approximately 80 km) between the discharge measurement point
(Tarboro) and the first measurement station, may explain the lack of a stronger correlation.
Salinity had weak correlation with time and small positive slopes (Table 3). These data suggest
that there is a minimal, positive trend in salinity detectable over the entire data record. There
does appear to be a significant change point in the salinity record that occurred in 2003 (Figure
3).
Tar River discharge from 1988-2003 showed a slightly weaker relationship with time
that resulted in a negative slope, but one that was less negative than that of the entire data
record (Figures 2, 4; Table 4). Measured salinities from 1988-2003 had very weak relationships
with time and slopes that were mostly positive, though the slope was negative at Station 1
(Figure 5, Table 4). In contrast, the latter part of the data record showed major differences. Tar
River discharge from 2003-2014 showed a stronger correlation with time and a more negative
slope compared to the other time periods (Figure 6, Table 5). Salinities values increased
significantly over the 2003-2014 period (Figure 7, Table 5). The relationship with time was

strong and an increase of approximately 1 salinity unit per year was observed in this period, on
average (Figure 7, Table 5). The trend was system -wide and consistent across all stations.
A time -series of mean, monthly salinity was compared to the Palmer Hydrologic Drought
Index, as discharge showed a weaker relationship to salinity (r2 = 0.23, p = 0). Salinity appeared
to relate to the PHDI more closely (Figure 8). This suggests that salinity is responding in a cyclic
manner with the regional climatological patterns that oscillate between wet and dry periods.
Discharge is part of this signal as well, but its effects are moderated by wind influence and lags
that are not captured by these simple, statistical models. When the residuals from the PHDI
regression are used tested for trend, the slope is 0.06 units per month. Thus, the overall trend
in increasing salinity appears to real and not an artifact of wet dry cycling.
Conclusion
Overall, a positive trend for salinity was found in the Tar -Pamlico River. This trend was
found to be robust over the entire data set, even when accounting for climatological variability
in drought and flow. The overall trend in salinity is comparable to estimates made from the
Delaware River estuary (2.5 to 4.4 units per m of sea level rise) (Ross et al. 2015). This estimate
was similar to what was estimated from the Tar -Pamlico data (Table 3.1 to 5.2 units by 2100).
Sea level rise in North Carolina is expected to be on the order of 1 m by 2100, depending on
location (Kopp et al. 2015). Thus, it would be reasonable to conclude that sea -level rise is likely
the cause of the upward trend in salinity; however, more thorough analyses would further
elucidate the validity of this trend
The more recent rise in salinity (2003-2014) appears to be a function of the recent
drought experienced in North Carolina (Figure 8). Salinity values were correlated to this
prolonged drought period and the severity of the drought can be seen the salinity projection to
2100 based on raw salinity (+72.24 units!) data and a linear model (Table 6). This projection is
clearly absurd, as open ocean salinity is on the order of 35. Using the residuals from the
regression against PHDI yields a more modest salinity projection (+13.76; Table 6); however,
this rate of salinity increase would appear to be too large to be maintained when compared to
the entire data record. Finally, the differences between the two time -periods may also be due
to differences in sampling intensity (Table 7), though sufficient data sampling appears to have
occurred in the latter period. Whatever the case, the more recent and strong linear increase in
salinity is likely to dissipate as the region recovers from drought conditions. However, I do
expect a modest salinity increase over time based on this analysis and it is likely that sea -level
rise is the cause of this trend. In the absence of other major events, such as a breach of the

Outer Banks and the formation of a new inlet, salinity values in the Tar -Pamlico River will rise as
the next century approaches in concert with sea -level rise.
Acknowledgements
I wish to express thanks to Laura McKenna for data collection and quality control. Rebecca
Woody -Cooper helped with data preparation and quality control. Thanks also to Dr. Donald
Stanley for beginning the monitoring program and Dr. Lisa Clough for continuing the program.

References
Copeland, B.J. and J.E. Hobbie. 1972. Phosphorus and eutrophication in the Pamlico River
estuary, NC, 1966-1969 - a summary. University of North Carolina Water Resources
Research Institute, Report No. 65. Raleigh. 86 pp.
Davis, G.J., M.M. Brinson, and W.A. Burke. 1978. Organic carbon and deoxygenation in a
coastal plain estuary: phytoplankton growth in large scale continuous cultures. University
of North Carolina Water Resources Research Institute, Report No. 131. Raleigh. 123 pp.
Hirsch, R.M, J.R. Slack, R.A. Smith. 1982. Techniques of trend analysis for monthly water quality
data. Water Resources Research. 18(1): 107-121.
Kopp, R. E., B. P. Horton, A. C. Kemp, and C. Tebaldi. 2015. Past and future sea -level rise along
the coast of North Carolina. Climatic Change 132: 693-707.
Kuenzler, E.J., D.W. Stanley, and J.P. Koenings. 1979. Nutrient kinetics of phytoplankton in the
Pamlico River, North Carolina. University of North Carolina Water Resources Research
Institute, Report No. 139. Raleigh. 155 pp.
Millard, S. P. and N. K. Neerchal (2001). Environmental Statistics with S -Plus. Boca Raton,
Florida, CRC Press.
Ross, A.C., R. G. Najjar, M. Li, M. E. Mann, S. E. Ford, and B. Katz. 2015. Sea -level rise and other
influences on decadal-scale salinity variability in a coastal plain estuary. Estuarine, Coastal,
and Shelf Science. 157: 79-92.

Table 1. Descriptive characteristics for Tar -Pamlico River stations.
Station
Latitude (°N)
Longitude (°W)
Description
1
35.35000
76.48333
Between Pamlico Point and Rose Bay
3
35.37306
76.64639
Lighted marker "3" off Indian Island
5
35.40028
76.73833
Between lighted marker "1" in
Gaylord Bay and lighted marker "1"
at ferry terminal
7
35.43056
76.84167
Lighted marker "5" off Core Point
8
35.45278
76.91944
Lighted marker "7" at Mauls Point
10
35.48194
76.98750
Lighted marker "12" at Camp Hardee

Table 2. Descriptive statistics for salinity measurements in the Tar -Pamlico River.
Station
N
Mean ± SD
Salinity range
1
2771
13.42 ± 4.24
0-27
3
3555
10.18 ± 4.31
0-22
5
5663
9.76 ± 4.67
0-24.5
7
4653
7.74 ± 4.72
0-21.5
8
3992
6.36 ± 4.68
0-21.5
10
3822
4.78 ± 4.51
0-22

Table 3. Seasonal Kendall trend statistics for Tar River discharge and Tar -Pamlico salinity
measurements at each station from 1988-2014.
Variable
Kendall's r
Slope
p -value
Tar River Discharge
-0.10
-0.31
0
Salinity ST1
0.04
0.04
0
Salinity ST3 0.09
Salinity ST5 0.07
Salinity ST7 0.08
Salinity ST8 0.05
0.10 0
0.07 0
0.10 0
0.06 0
Salinity ST10 0.07 0.03 0

Table 4. Seasonal Kendall trend statistics for Tar River discharge and Tar -Pamlico salinity
measurements at each station from 1988-2003.
Variable
Kendall's r
Slope
p -value
Tar River Discharge
-0.05
-0.27
0
Salinity ST1
-0.04
-0.07
0.02
Salinity ST3 0.03
Salinity ST5 0.01
Salinity ST7 0.03
Salinity ST8 0.001
0.08
0.002
0.05
0.02
0.07
0.001
0.03
0.07
Salinity ST10 0.02 0.01 0.03

Table 5. Seasonal Kendall trend statistics for Tar River discharge and Tar -Pamlico salinity
measurements at each station from 2003-2014.
Variable
Kendall's r
Slope
p -value
Tar River Discharge
-0.16
-1.58
0
Salinity ST1
0.44
1.35
0
Salinity ST3
0.48
1.14
0
Salinity ST5
0.45
1.10
0
Salinity ST7
0.44
1.10
0
Salinity ST8
0.41
1.00
0
Salinity ST10 0.42 0.82 0

Table 6. Monthly and yearly linear slopes for mean, monthly time -series of salinity and residuals
of a regression between PHDI and salinity. Final column shows the estimated rise in salinity by
2100 based on extrapolation of the linear model.
Time -Period
Monthly slope
Yearly slope
Salinity by 2100
1988-2013
0.003
0.036
+3.10
1988-2013 (PHDI residuals)
0.005
0.06
+5.16
2003-2013
0.07
0.84
+72.24
2003-2013 (PHDI residuals)
0.01
0.16
+13.76

Table 7. Sample sizes for the two time periods (1988-2003, and 2003-2014) at three
representative stations (1, 5, and 10).
Month Station 1988-2003 N 2003-2014 N
1
1
100
54
2
1
144
32
3
1
102
89
4
1
115
75
5
1
94
47
6
1
149
161
7
1
200
141
8
1
216
157
9
1
123
72
10
1
258
107
11
1
121
54
12
1
89
71
1
5
217
93
2
5
227
155
3
5
219
168
4
5
239
159
5
5
292
135
6
5
445
252
7
5
381
225
8
5
408
217
9
5
316
206
10
5
350
183
11
5
251
160
12
5
203
162
1
10
143
61
2
10
168
98
3
10
157
110
4
10
163
97
5
10
199
91
6
10
287
162
7
10
250
148
8
10
273
141
9
10
232
136
10
10
248
118
11
10
196
106
12
10
137
101

Estuarine Monitoring Program
Sampling Sites
Washington
Tar River
I Broad Creek
ChOCOW"fy 1
ay 10
Blounts say
Biounts Creek
ljN1
Created by: S. Ludwig -Monty
Duke University - NSCE
s
Pungo River
Bath Creek
7
5N
S 406
S •3
e
a�Gx dS`' MIA
r
S
� nC, � .00 Sou'rv,
4
Goose Caeek
Figure 1. Map of sampling locations in the Tar -Pamlico River. Black circles represent the stations
used in this analysis. Blue squares represent other stations that have been sampled historically
(map created by S. Ludwig -Monty, Duke University).

2.4
a
0.1
O.0
I'a�lJ I�J�JD LUUU LUUU LV IU LU 13
Figure 2. Tar River discharge (log10 M3s-1)at Tarboro, NC (US gage 02083500) from 1988-2014.
Line represents linear model fit.

w
25
20
75
10
5
0
25
20
15
10
5
0
1999 7995 2000 2005 2070 20115
Bi 7
8
+
• • H
W 14
y�
8
1999 7995 2000 2005 2070 20115
Bi 7
B
+
• • H
W 14
�• moi `�° �w
�t � �;
19915 1495 2000 2005 2010 3019
25
20
915
10
5
0
9
25
s
5
S
R vg°
g 8
K
®yR�
6 w
Ea
P
1890
1895
2000
2005
2010
2(175
25
20
915
10
5
0
9
1940 4495 2000 2005 2010 2715
00
25
r
1940 4495 2000 2005 2010 2715
00
25
5
$ 6
v •$
1990 1995
70L0 700.5 7010 2015
Figure 3. Tar -Pamlico River salinity at stations 1, 3, 5, 7, 8, and 10 from 1988-2014. Line
represents linear model fit.

3
LTJ
0
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Figure 4. Tar River discharge (log10 M3s-1)at Tarboro, NC (US gage 02083500) from 1988-2003.
Line represents linear model fit.

2.5
20
15
t©
5
e $!4 9 a:.
aoe a .: �•• ; ��e � ii
M 19E9 1990 1991 1998 1943 1994 3495 199E 1997 1996 1900 2000 2009 20U22003
25
1906 1989 1990 1991 1992 199319941991519%199719%B 19D200D 2001 3007 2003
25
20
15
10
9
0
19154 1989 1990 1991 1992 1993 7994 1995 14.4% 1991 t998 1949 2000 200t 20023003
25
29
915
10
5
26
20
1S
10
5
0
$8° '�6 >f a 8a A
a + s•
55bd �i
� 8
iS61489 1990 MI 1992 199a 1994 1996 IM 100111%819020M 2001 20Q, M.
° o
A_ IPad
$ pfIf
1988 1960 1990 1991 1992 1993 1994 199$ 1994 1997199$1459 2000 2001 2002 2003
25
b
ro
�
ga IrNg
• .�
$ e • 4
16.1
�.
10.1 .f
_4S
19154 1989 1990 1991 1992 1993 7994 1995 14.4% 1991 t998 1949 2000 200t 20023003
25
29
915
10
5
26
20
1S
10
5
0
$8° '�6 >f a 8a A
a + s•
55bd �i
� 8
iS61489 1990 MI 1992 199a 1994 1996 IM 100111%819020M 2001 20Q, M.
° o
A_ IPad
$ pfIf
1988 1960 1990 1991 1992 1993 1994 199$ 1994 1997199$1459 2000 2001 2002 2003
25
20
Z
ae�`
-�..
.�,,, 4"
6
P1.. ii .
14P81 ON 1A001901 11M 1093 tIA4 199x. 19ti; 11?37 1998 1999 7&k1 3001 711107 7004
Figure 5. Tar -Pamlico River salinity at stations 1, 3, 5, 7, 8, and 10 from 1988-2003. Line
represents linear model fit

J
!
' ! !
a{/K! ]/
\2 /\�g #/ §
7 � \
_ \ \,
!
a
!!!
;/
MI
j
!
a
!!!
MI
i
�
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Figure 6 Tar qvr discharge(oEE M3 s-1) at Tarboro,NC(US gage 02083500) from 2003-2014.
Line represents linear model §t

25
20
m i5
10
5
7,003 7006 202 700E 7007 7000 2009 70tO 2.011 2017 7013 M
i5 m
2.0°
L Fa ���� °• rer��
ais s ema % gr
1&dL m d e &�, •, '�,o
10
e
e
0
7003 7004 2W,, 700E 2007 2000 74'19 70t0 2,011 2012 2013 2M4
25
20
�15
i0
5
0
700:1 7004 2Off 2006 2007 2000 2009 2M 2011 2012 7013 2014
25
20
� 15
10
5
2003 7004 2Dor 200E x.007 2000 2009 7010 7011 7012 7013 7M,
25
20-
2003
0 7003 2004 200.5 2006 7007 2000 ?009 7010 2011 2012 2013 2M4
zs
zo
a �
5 • °8ga B�' .j� °� ° i.
0 • a'
PY11:1 2104 7004 ?ML 700? 2n{19 ?1rvi xi" 2 11 nu 701, nail
Figure 7. Tar -Pamlico River salinity at stations 1, 3, 5, 7, 8, and 10 from 2003-2014. Line
represents linear model fit

F)
A
25
20
m
15
10
5
0
1990 199', 2000 2005 2010
25
20
-5 -4 -3 -2 -1 0 1 2 3 4 5
PHDI
Figure 8. Mean, monthly salinity (as calculated from all stations) compared to Palmer
Hydrologic Drought Index. Top panel shows time -series and bottom panel shows a regression
(r 2 = 0.43, p = 0)
• + s •
•. s
25
20
-5 -4 -3 -2 -1 0 1 2 3 4 5
PHDI
Figure 8. Mean, monthly salinity (as calculated from all stations) compared to Palmer
Hydrologic Drought Index. Top panel shows time -series and bottom panel shows a regression
(r 2 = 0.43, p = 0)