Spectral Analysis of Streamflow for Continental USA

Kazi Ali Tamaddun1, Ajay Kalra2, Sajjad Ahmad3

1

Department of Civil and Environmental Engineering and Construction, University of

Nevada, 4505 S. Maryland Parkway, Las Vegas, NV 89154-4015, USA; Phone: (702)
490-1284; email: tamaddun@unlv.nevada.edu

Development Review Division, Clark County Public Works, 500 S. Grand Central
Parkway, Las Vegas, NV 89155, USA; Phone: (702) 455-7531; email:
Ajay.Kalra@ClarkCountyNV.gov
3

Department of Civil and Environmental Engineering and Construction, University of

Nevada, 4505 S. Maryland Parkway, Las Vegas, NV 89154-4015, USA; Phone: 702895-5456; email: sajjad.ahmad@unlv.edu
Abstract:
Evaluating streamflow changes is one of the most important tasks in hydrology,
which provides significant insight regarding the trends that occur in streamflow due
to climate variability and change. This study focused on investigating the presence of
gradual changes (trends) and abrupt changes (shifts) in 240 unimpaired streamflow
stations across the continental United States. Discrete Wavelet Transform (DWT) was
used as the trend detection method, which is one of the techniques used in Spectral
Analysis. The changes in streamflow in water-year were analyzed along with three
dyadic scales (1year, 2 years and 4 years) for 60 years i.e., 1951-2010. The nonparametric Mann-Kendall test was used to identify the trends, whereas the nonparametric Pettitt test was used to identify the shifts. The results showed a significant
increase in streamflow in the Northeast and Midwest (Upper Mississippi and Ohio)
regions and a decrease in the Pacific Northwest region. The central regions, especially
Missouri, had mixed results with a propensity towards increasing trends. The
southern U.S. regions did not show any overall significant trends. The shifts were
found to be more spatially dispersed and were in agreement with the gradual trends.
More stations showed trends and shifts with increasing time scales, which implies the
presence of periodic occurrences at 4 years or higher. The presence of persistence was
also observed to increase with the increasing time-scales. The results from the current
study may assist water managers to efficiently plan and manage the water resources
under changing climatic conditions.
Key words: Streamflow, Spectral Analysis, Mann-Kendall, Pettitt Test, Discrete
Wavelet Transform

Introduction:
Alterations in magnitude and frequency of flow events, as a result of change in
climatic conditions, can threaten critical infrastructures and natural environment
(Dawadi and Ahmad, 2013; Burn et al. 2010; Kalra and Ahmad, 2009; Kalra et al.,
2013a; Puri et al., 2011a,b; Stephen et al., 2010; Choubin et al., 2014; Zhang et al.,
2014). The changes also limit the accessibility of clean water for upcoming
generations (Middelkoop et al., 2001). As a result, meeting the future demand with
present water infrastructure is becoming more and more challenging for water
managers (Dawadi and Ahmad, 2012; Qaiser et al., 2011; Ahmad et al., 2010a; Kalra
et al., 2013b&c). Water required for the increasing energy demand and environmental
needs makes the scenario more challenging (Shrestha et al., 2011; 2012; Kalra and
Ahmad, 2012; Impoinvil et al., 2007). The changes in hydrological cycle can result
from climate variability and in some cases from anthropogenic intrusion (Lettenmaier
et al., 1994; Kalra and Ahmad, 2011; Carrier et al., 2013; Sagarika et al., 2014; Wu et
al., 2013; Ahmad et al., 2010b). The hydrologic cycle can exhibit two types of
changes - gradual (trend) and abrupt (shift). Long-term trends are expected to
continue for a longer period of time into the future while shifts can cause a sharp
change (increase or decrease) in the pattern of the hydrologic variable under
observation (Lins and Slack, 1999; McCabe and Wolock, 2002). If these changes are
not properly addressed while designing projects, the outcome may cause under-design
or over-design of structures (Milly et al., 2008; Forsee and Ahmad, 2011).
In the past, studies have been made to understand the flow pattern of streamflows
over continental United States; most of the studies focused on long term trends in
streamflows (Svensson et al., 2005; Sagarika et al., 2014). Studies suggest that annual
moderate-to-low streamflows showed significant increase whereas peak streamflows
showed relatively less severe increase (Lettenmaier et al., 1994; Kalra et al., 2008).
Several methods have been applied to detect and analyze time series data of
streamflow. Nalley et al. (2012) explained various methods and listed the previously
conducted studies concerning these methods. The Mann-Kendall (MK) trend test
(Mann, 1945; Kendall, 1975) has been widely used as the detection method as it has
several advantages over other methods. MK trend test can analyze a skewed
distribution (nz and Bayazit, 2003) which makes it more effective than other
methods and this feature in turn minimizes the limitations of some of the previous
works. MK trend test is also resilient to missing values and non-stationary data,
which are important features for analyzing hydrologic time-series. Traditional MK
test does not account for auto-correlation, which is considered a weakness of this
method. Studies suggest that non-linear trends might not always be detected by MK
tests (McBean and Motiee, 2006). As a result MK test is often used in combination
with other methods (Nalley et al., 2012) which can account for the non-linearity as
non-linear trends are quite recurrent in hydrology.
In this study, streamflow data were collected from 240 unimpaired streamflow
stations across the continental United States over a period of 60 years (i.e., 19512010). The time-series data were transformed (decomposed) into multiple lowresolution intervals using discrete wavelet transform (DWT). Independent and
modified MK tests (with appropriate modifications accounting for long term

persistence (LTP) and short term persistence (STP)) were used to detect the trends
(gradual changes). Pettitt test (Pettitt, 1979) was applied to determine the changes in
shifts (abrupt changes).The analyses were done based on water-year and 3 different
dyadic scales (i.e., 1 year, 2 years and 4 years) to differentiate the patterns of trends
and shifts in different time scales. Hence, the most significant periodicities which are
affecting the trends and shifts present in the time-series were assessed.
Study area and data:
The continental U.S. is divided into 18 regions by the United States Geological
Survey
(USGS)
in
their
Hydrologic
Unit
Map
(http://water.usgs.gov/GIS/regions.html). USGS Hydro-climatic Data Network 2009
(HCDN-2009), published in 2012, lists 704 stations in the continental USA. The
HCDN-2009 dataset is a revised version of the USGS original HCDN network (Slack
et al., 1992). In this study, 240 stations having a continuous water-year data of 60
years from 1951 to 2010 (October 1950 to September 2010) were used for the
analysis (Figure 1). The raw data were collected on a monthly basis and were
averaged to obtain the water-year data i.e. October of the previous year to September
of the current year. The dyadic scales obtained from decompositions were 1 year, 2
years and 4 years. The length of data and the resolution obtained from the last detail
coefficient of decomposition dictated the choice of dyadic scales.

Figure 1: Map showing the location of 240 unimpaired streamflow stations across 18
hydrologic regions in continental United States.
Method:
Discrete wavelet transform in conjunction with different trend and shift detection
techniques were applied to the dataset (of 60 years) to detect the spatio-temporal
distribution of the streamflow across continental U.S. for water year and 3 different
dyadic scales (i.e., 1 year, 2 years and 4 years). The non-parametric Mann-Kendall
tests have been considered by many researchers as an effective tool since they are free
from any initial assumption regarding the probability distribution and they can also

deal with the presence of outliers in the dataset (Lins and Slack, 1999; Burn, 2008). A
comparison between different change-point detection techniques suggests that the
non-parametric Pettitt test has better accuracy than others in detecting trends
(Villarini et al, 2009).
DWT was applied to the original time-series of the streamflow using decomposition
function found in MATLAB wavelet toolbox. Coiflet wavelet function (Coif1) was
used in this study as the mother wavelet type due to shortened time-series and better
accuracy over other functions (Partal and Kk, 2006).Inflation of variance can lead
to a rejection of the null hypothesis due to the presence of serial correlation (Lag-1
autocorrelation) (Yue et al., 2002). To remove serial correlation, (1) the series was
pre-whitened assuming the presence of lag-1 autocorrelation and (2) the series was
de-trended before removing the lag-1 autocorrelation, which is also known as Trend
Free Pre-whitening (TFPW) (Yue et al., 2002). The presence of long-term
persistence, or the Hurst phenomenon (Hurst, 1951), might lead to an underestimation
of the serial correlation of the data and an over-estimation of the significance of the
MannKendall test. To detect the trend under LTP, the method proposed by Hamed
(2008) was adopted. The trend results are presented at a significance level of p 0.10
Results:
Trend Changes:
Table 1 list the results of all three MK tests for water-year and 3 dyadic scales of 1, 2
and 4 years. In water-year, an increasing trend is present in the northeast and the
central United States (Figure 2). Majority of the stations showing decreasing trends
lie in the northwest region. In water-year, out of 240 stations, 50 stations (21% of the
total stations) showed increasing trends whereas 23 stations (10% of the total stations)
showed decreasing trends. Great Lakes, Ohio, Upper Mississippi, Souris-Red-Rainy,
Missouri, Arkansas-White-Red, Texas Gulf and Rio Grande showed significant
increasing trends. Decreasing trends were mostly observed in Pacific Northwest. The
results of MK2 and MK3 are listed in Table 1 along with MK1.

(a)

(b)

(c)

Figure 2: Spatial patterns of the results from (a) MK1, (b) MK2 and (c) MK3 for
water-year at p 0.10. Significant Increasing and decreasing trends are shown by
upward pointing grey pentagons and downward pointing black triangles respectively.
Dots indicate no significant trend.

Table 1: Results of the three Mann Kendall (MK) tests at each hydrologic region for
water-year and 3 dyadic scales of 1, 2, and 4 years.
Hydrologic
Region No.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

Region Name

New England
Mid-Atlantic
South Atlantic-Gulf
Great Lakes
Ohio
Tennessee
Upper Mississippi
Lower Mississippi
Souris-Red-Rainy
Missouri
Arkansas-White-Red
Texas-Gulf
Rio Grande
Upper Colorado
Lower Colorado
Great Basin
Pacific Northwest
California

Total

Number of
Unimpaired
Stations

18
39
21
10
21
8
19
2
3
22
11
5
3
_
2
5
37
14
240

Number of Stations with Trends

Water-year
1 year
2 years
4 years
MK1+/- MK2+/- MK3+/- MK1+/- MK2+/- MK3+/- MK1+/- MK2+/- MK3+/- MK1+/- MK2+/- MK3+/4/0
4/0
3/0
8/0
8/0
8/0
8/0
11/0
0/0
8/0
14/0
0/0
3/0
3/0
0/0
9/0
9/0
9/0
11/0
20/0
0/0
15/0
29/0
0/0
0/0
0/0
0/0
1/1
1/1
1/1
2/2
3/3
0/0
1/2
5/8
0/0
4/2
4/2
3/0
5/3
5/3
5/3
5/3
6/3
0/0
4/3
7/3
0/0
5/0
5/0
3/0
10/0
12/0
9/0
10/0
12/0
2/0
13/0
15/2
1/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
2/0
0/0
0/1
0/2
0/0
13/0
13/0
5/0
16/0
16/0
15/0
16/0
16/0
2/0
16/0
16/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
1/0
0/0
1/0
1/0
1/0
2/0
2/0
2/0
2/0
2/0
1/0
2/0
2/0
0/0
10/2
11/2
4/0
12/1
12/1
10/1
14/2
14/2
1/0
11/2
13/3
1/0
5/1
5/1
3/1
7/1
7/1
7/1
6/1
7/1
0/0
8/1
9/1
0/0
3/0
3/0
3/0
5/0
5/0
5/0
5/0
5/0
0/0
5/0
5/0
0/0
1/0
1/0
1/0
3/0
3/0
2/0
2/0
3/0
0/0
3/0
3/0
0/0
_
_
_
_
_
_
_
_
_
_
_
_
1/0
1/0
1/0
1/0
1/0
1/0
1/0
1/0
0/0
1/0
1/1
0/0
0/0
0/0
0/0
1/0
0/0
1/0
1/0
1/1
0/0
1/0
3/0
0/0
0/17
0/17
0/16
0/21
0/24
0/20
0/29
0/30
0/7
0/31
0/35
0/0
0/1
0/1
0/1
0/2
0/2
0/2
4/5
4/4
0/0
2/3
3/5
0/0
50/23

51/23

27/18

80/29

81/32

75/28

87/42

107/44

6/7

90/43

126/60

2/0

MK1, MK2, MK3: corresponding MK test.

+ : Total number of stations showing increasing trends
- : Total number of stations showing decreasing trends

In 1 year dyadic scale, a total of 80 stations (33.3% of the total stations) showed
increasing trends whereas 29 stations (12% of the total stations) showed decreasing
trends (Table 1).
In 2 year dyadic scale, a total of 129 stations (54% of the total stations) showed either
increasing or decreasing trends. 87 stations (36% of the total stations) showed
increasing trends whereas 42 stations (18% of the total stations) showed decreasing
trends (Table 1). Dyadic scale of 4 years showed almost identical spatial pattern of
trends across the continental U.S. when compared to 2 years dyadic scale except for
some slight changes in the number of stations in a few hydrologic regions. A total of
90 stations (38% of the total stations) showed increasing trends whereas 43 stations
(18% of the total stations) showed decreasing trends. The analysis of STP and LTP
indicated that northeast U.S. has greater number of stations witnessing persistence
while southern U.S. did not show any significant presence of lag-1 autocorrelation or
LTP. The central portion also had stations exhibiting lag-1 autocorrelation and LTP
but it decreased along eastern to western regions. The results of MK2 and MK3 from
Table 1 correspond to the number of stations with trends in each region
acknowledging the persistence.
Abrupt changes:
A summary of the number of stations showing abrupt changes or shifts are presented
in Table 2 corresponding to each hydrologic region. The table also shows the number
of stations with change points in water-year and dyadic scales of 1, 2 and 4 years.

Great Lakes, Upper Mississippi, Lower Mississippi, Souris-Red-Rainy, Texas-Gulf,

Rio Grande and Lower Colorado had more than 50% of their stations with increasing
shift in water-year. Pacific Northwest was the only region with more than 50% of the
stations showing decreasing shifts. Figure 3(a) shows the spatial distribution of
stations across continental U.S. with increasing and decreasing shifts in water-year.
Figure 3(b) presents the region wise shifts in a comparative graph in water-year.
Table 2: Results of the Pettitt test, reporting shifts for the water-year and dyadic
scales of 1, 2 and 4 years.
Hydrologic
Region No.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Total

Region Name

New England
Mid-Atlantic
South Atlantic-Gulf
Great Lakes
Ohio
Tennessee
Upper Mississippi
Lower Mississippi
Souris-Red-Rainy
Missouri
Arkansas-White-Red
Texas-Gulf
Rio Grande
Upper Colorado
Lower Colorado
Great Basin
Pacific Northwest
California

Number of
Unimpaired
Stations

Number of Stations with shifts

18
39
21
10
21
8
19
2
3
22
11
5
3
_
2
5
37
14

Water-year
+/5/0
12/0
3/0
5/3
7/0
0/0
13/0
2/0
2/0
10/2
5/1
4/0
3/0
_
1/0
0/1
0/19
3/0

1 year 2 years 4 years

+/+/+/7/0
14/0 17/0
19/0 30/0 37/0
9/6
9/7
12/7
5/3
7/3
7/3
13/0 12/1 16/1
2/0
2/0
2/4
15/0 16/0 16/3
2/0
2/0
2/0
3/0
3/0
3/0
13/3 14/4 14/5
8/1
8/1
9/1
5/0
5/0
5/0
3/0
3/0
3/0
_
_
_
1/0
1/0
1/1
2/1
2/1
3/1
0/22 1/34 0/35
2/2
6/4
6/5

240

75/26

109/38 135/55 153/66

+ : Total number of stations showing increasing shifts

- : Total number of stations showing decreasing shifts

(a)

(b)

Figure 3: (a) Spatial distribution of stations with increasing and decreasing shifts in
water-year. (b) Region wise comparison with increasing and decreasing shifts in
water-year.

While analyzing with dyadic scale of 1 year, out of 240 unimpaired stations 109
stations (about 45% of the total stations) showed increasing shifts while 38 stations
(around 16% of the total stations) showed decreasing shifts (Table 2). Number of
stations showing shifts in 1 year dyadic scale showed around a 45% increase
compared to water-year. 2 years dyadic scale showed increase in the number of
stations both in terms of increasing and increasing shifts. Out of 240 unimpaired
stations, 190 stations (about 80% of the total stations) showed either increasing or
decreasing shift. 135 stations (about 56% of the total stations) showed increasing
shifts whereas 55 stations (about 23% of the total stations) showed decreasing shifts.
There was around 29% percent increase in the number of stations showing shifts in 2
years dyadic scale compared to 1 year dyadic scale. Number of stations showing
shifts are seen to increase more in 4 year dyadic scale. Out of 240 unimpaired stations
219 stations (more than 91% of the total stations) were seen to have either increasing
or decreasing shifts. 153 stations (about 64% of the total stations) showed increasing
shift whereas 66 stations (about 28% of the total stations) showed decreasing shift in
4 years dyadic scale. 4 years dyadic scale also showed 15% increase in the number of
stations showing shifts when compared to 2 years dyadic scale.
Conclusion and Discussion:
In this study, streamflow data from 240 unimpaired stations across the continental
U.S. were analyzed for long term trend and abrupt shifts. The analyses of the timeseries data were conducted using non-parametric tests i.e. Mann-Kendall and Pettitt
test. Appropriate modifications to MK test were applied to account for LTP and STP.
DWT was applied to decompose the original time-series to evaluate the underlying
trends and shifts. Trends and shifts were analyzed at three different dyadic scales (i.e.,
1, 2 and 4 years) to find out which periodic scales are more influential.
The water-year trend analysis showed significant increasing trends in the uppercentral and northeast U.S. for all 3 MK-tests. Northwest region on the other hand
showed significant decreasing trend for all the MK-tests. Southwest and southeast
U.S. did not show any significant trends in any of the MK-tests. Number of stations
showing trends in the southern U.S. was also less compared to northern parts of the
United States. For different dyadic scales (1, 2 and 4 years), the spatial distribution of
the stations showing increasing and decreasing trends were similar to water-year
though the number of stations increased with increasing time-scale. MK-3 did not
show significant amount of stations exhibiting trends for 2 and 4 years dyadic scales.
The analyses of the presence of LTP and STP showed that the number of stations
with either lag-1 autocorrelation or LTP increases with the increasing time-scale. For
2 and 4 years dyadic scales, almost all the stations had both lag-1 autocorrelation and
LTP. For abrupt shifts, the stations showing increasing or decreasing step changes
were more spatially dispersed. The central and southern U.S. also showed significant
number of stations with shifts. The prominence of increasing shifts in northeast and
decreasing shifts in northwest showed similar patterns like long-term (trend) changes.
The historical analyses of the step changes have shown that there were presences of
significant increasing shifts during 1970 followed by a significant decreasing shift
from 1973-1976. For water-year and dyadic scales of 1 and 2 years showed presence

of simultaneous increasing and decreasing shifts after 1980s. For 4 year dyadic scale,
the increasing and decreasing shifts showed a reverse pattern.
Overall, the current study provides information regarding the time scales which are
influencing the trends and shifts. It also recognizes the patterns that were present in
the historical data of the study period. The study can be extended by analyzing data of
greater length which would allow decompositions of dyadic scales greater than 4
years. Inclusion of the factor of the seasons can be a scope for future work. The
results from this study may assist water managers to efficiently plan and manage the
water resources under changing climatic conditions. The study also provides a
platform for further research on developing forecasting models by analyzing the trend
lines which can be generated from these analyses.
Acknowledgment
This work was supported partly by NSF under Grants IIA-1329469 and CMMI0846952.
References
Ahmad MM, Ghumman AR, Ahmad S, Hashmi HN (2010a) Estimation of a unique pair
of Nash model parameters: an optimization approach. Water Resources Management
24(12): 2971-2989. DOI: 10.1007/s11269-010-9590-3.
Ahmad, S., Kalra, A., & Stephen, H. (2010b). Estimating soil moisture using remote
sensing data: A machine learning approach. Advances in Water Resources, 33(1),
6980. doi:10.1016/j.advwatres.2009.10.008
Burn, D. H. (2008). Climatic influences on streamflow timing in the headwaters of the
Mackenzie River Basin. Journal of Hydrology, 352(12), 225238.
doi:10.1016/j.jhydrol.2008.01.019
Burn, D. H., Sharif, M., & Zhang, K. (2010). Detection of trends in hydrological
extremes for Canadian watersheds. Hydrological Processes, 24(13), 17811790.
doi:10.1002/hyp.7625
Carrier, C., Kalra, A., & Ahmad, S. (2013). Using Paleo Reconstructions to Improve
Streamflow Forecast Lead Time in the Western United States. JAWRA Journal of
the American Water Resources Association, 49(6), 13511366.
doi:10.1111/jawr.12088
Choubin B, Sigaroodi SK, Malekian A, et al. (2014). Drought forecasting in a semi-arid
watershed using climate signals: a neuro-fuzzy modeling approach. Journal of
Mountain Science 11(5). DOI: 10.1007/s11629-014-3020-6
Dawadi, S., and Ahmad, S. (2012). Changing climatic conditions in the Colorado River
Basin: Implications for water resources management. Journal of Hydrology, 430
431, 127141. doi:10.1016/j.jhydrol.2012.02.010
Dawadi, S., and Ahmad, S. (2013). Evaluating the impact of demand-side management
on water resources under changing climatic conditions and increasing population.
Journal of Environmental Management, 114, 26175.
doi:10.1016/j.jenvman.2012.10.015
Forsee, W., and Ahmad, S. (2011). Evaluating urban stormwater infrastructure design in
response to projected climate change. ASCE J. Hydrologic Eng. 16, 865873.

Hamed, K. H. (2008). Trend detection in hydrologic data: The MannKendall trend test
under the scaling hypothesis. Journal of Hydrology, 349(34), 350363.
doi:10.1016/j.jhydrol.2007.11.009
Hurst, H. (1951). Long-term storage capacity of reservoirs. Trans. Am. Soc. Civ. Eng.
116, 770799.
Impoinvil, D. E., Ahmad, S., Troyo, A., Keating, J., Githeko, A. K., Mbogo, C. M., &
Beier, J. C. (2007). Comparison of mosquito control programs in seven urban sites
in Africa, the Middle East, and the Americas. Health Policy, 83(2), 196-212.
Kalra, A., Piechota, T.C., Davies, R., Tootle, G.A. (2008). Changes in U.S. streamflow
and western U.S. snowpack. J. Hydrol. Eng. 13 (3), 156163.
Kalra, A., and Ahmad, S., (2009). Using oceanic-atmospheric oscillations for long leadtime stream flow forecasting. Water Resources Research, 45(3), W03413,
http://dx.doi.org/10.1029/2008WR006855.
Kalra, A., and Ahmad, S. (2011). Evaluating changes and estimating seasonal
precipitation for the Colorado River Basin using a stochastic nonparametric
disaggregation technique. Water Resour. Res. 47, W05555.
http://dx.doi.org/10.1029/ 2010WR009118.
Kalra, A., and Ahmad, S. (2012). Estimating annual precipitation for the Colorado River
Basin using oceanic-atmospheric oscillations. Water Resources Research, 48(6),
W06527. doi:10.1029/2011WR010667
Kalra, A., Ahmad, S., & Nayak, A. (2013a). Increasing streamflow forecast lead time
for snowmelt-driven catchment based on large-scale climate patterns. Advances in
Water Resources, 53, 150162. doi:10.1016/j.advwatres.2012.11.003
Kalra, A., Li, L., Li, X., Ahmad, S. (2013b). Improving streamflow forecast lead time
using oceanic-atmospheric oscillations for Kaidu river basin, Xinjiang, china. J.
Hydrological Eng. 18 (8), 10311040.
Kalra, A., Miller, W. P., Lamb, K. W., Ahmad, S., & Piechota, T. (2013c). Using largescale climatic patterns for improving long lead time streamflow forecasts for
Gunnison and San Juan
River Basins. Hydrological Processes, 27(11), 1543
1559. doi:10.1002/hyp.9236
Kendall, M.G. (1975). Rank correlation methods: Charles Griffin, London, UK.
Lettenmaier, D.P., Wood, E.F., Wallis, J.R. (1994). Hydro-climatological trends in the
continental United States, 194888. J. Clim. 7 (4), 586607.
Lins, H., & Slack, J. (1999). Streamflow trends in the United States. Geophysical
Research Letters, 26(2), 227230.
McBean, E., & Motiee, H. (2006). Assessment of impacts of climate change on water
resources a case study of the Great Lakes of North America. Hydrology and Earth
System
Sciences Discussions, 3(5), 31833209. doi:10.5194/hessd-3-3183-2006
McCabe, G. J., & Wolock, D.M. (2002). A step increase in streamflow in the
conterminous
United States. Geophysical Research Letters, 29(24), 2185.
doi:10.1029/2002GL015999
Middelkoop, H., Daamen, K., & Gellens, D. (2001). Impact of climate change on
hydrological regimes and water resources management in the Rhine basin. Climatic
Change, 105128.
Milly, P., Julio, B., Malin, F., Robert, M., Zbigniew, W., Dennis, P., Ronald, J. (2008).
Stationarity is dead. Ground Water News Views 4 (1), 68.
Nalley, D., Adamowski, J., & Khalil, B. (2012). Using discrete wavelet transforms to
analyze trends in streamflow and precipitation in Quebec and Ontario (19542008).
Journal of Hydrology, 475, 204228.

nz, B., Bayazit, M. (2003). The power of statistical tests for trend detection. Turk. J.
Eng.
Environ. Sci. 27 (4), 247251.
Partal, T., & Kk, M. (2006). Long-term trend analysis using discrete wavelet
components of
annual precipitations measurements in Marmara region (Turkey).
Physics and Chemistry of the Earth, Parts A/B/C, 31(18), 1189 1200.
doi:10.1016/j.pce.2006.04.043
Pettitt, A. (1979). A non-parametric approach to the change-point problem. Appl. Stat.
28 (2), 126135.
Puri, S., Stephen, H., and Ahmad, S. (2011a). Relating TRMM precipitation radar land
surface backscatter response to soil moisture in the southern United States. Journal
of Hydrology, 402(12), 115125.
Puri S, Stephen H, Ahmad S (2011b). Relating TRMM precipitation radar backscatter to
water stage in wetlands. Journal of Hydrology 401 (3-4): 240-249. DOI:
http://dx.doi.org/10.1016/j.jhydrol.2011.02.026.
Qaiser, K., Ahmad, S., Johnson, W., & Batista, J. (2011). Evaluating the impact of water
conservation on fate of outdoor water use: a study in an arid region. Journal of
Environmental Management, 92(8), 20618. doi:10.1016/j.jenvman.2011.03.031
Sagarika, S., Kalra, A., & Ahmad, S. (2014). Evaluating the effect of persistence on
long-term trends and analyzing step changes in streamflows of the continental United
States. Journal of Hydrology, 517, 3653. doi:10.1016/j.jhydrol.2014.05.002
Shrestha, E., Ahmad, S., Johnson, W., Shrestha, P., & Batista, J. R. (2011). Carbon
footprint of water conveyance versus desalination as alternatives to expand water
supply. Desalination, 280(1-3), 3343. doi:10.1016/j.desal.2011.06.062
Shrestha, E., Ahmad, S., Johnson, W., & Batista, J. R. (2012). The carbon footprint of
water management policy options. Energy Policy, 42, 201212.
doi:10.1016/j.enpol.2011.11.074
Slack, J.R., Landwehr, J.M., Lumb, A. (1992). A U.S. geologic survey streamflow data
set for the United States for the study of climate variations. 19741988, Rep. 92
129, United States Geological Survey, Oakley, Utah.
Stephen, H., Ahmad, S., Piechota, T. C., and Tang, C. (2010). Relating surface
backscatter response from TRMM precipitation radar to soil moisture: Results over a
semi-arid region. Hydrology and Earth System Sciences, 14(2), 193204.
Svensson, C., Kundzewicz, W. Z., & Maurer, T. (2005). Trend detection in river flow
series: 2. Flood and low-flow index series / Dtection de tendance dans des sries de
dbit fluvial: 2. Sries dindices de crue et d'tiage. Hydrological Sciences Journal,
50(5), 3741. doi:10.1623/hysj.2005.50.5.811
Villarini, G., Serinaldi, F., Smith, J. a., & Krajewski, W. F. (2009). On the stationarity
of annual flood peaks in the continental United States during the 20th century.
Water Resources Research, 45(8). doi:10.1029/2008WR007645
Yue, S., Pilon, P., Phinney, B., & Cavadias, G. (2002). The influence of autocorrelation
on the ability to detect trend in hydrological series. Hydrological Processes, 16(9),
18071829.
doi:10.1002/hyp.1095
Wu G, Li L, Ahmad S, Chen X, Pan X (2013). A Dynamic Model for Vulnerability
Assessment of Regional Water Resources in Arid Areas: A Case Study of
Bayingolin, China, Water Resources Management, 27(8): 3085-3101.
Zhang FY, Li LH, Ahmad S, et al. (2014). Using path analysis to identify the influence
of climatic factors on spring peak flow dominated by snowmelt in an alpine
watershed. Journal of Mountain Science, 11(4). DOI: 10.1007/s11629-013-2789-z.