ENSO Induced Droughts Impacts on
Groundwater Levels in the Lower
Apalachicola-Chattahoochee-Flint River Basin
By
Subhasis Mitra, Puneet Srivastava, Lynn Torak and Sarmistha Singh
Biosystems Engineering, Auburn University
Georgia Water Science Center, USGS
Introduction
There is increased pressure on water resources of southeast USA:
• Due to increased urbanization and population.
• Severe climate variability caused by ENSO exacerbates the situation.
El Niño Southern Oscillation (ENSO)
• Cycle of above and below average sea-surface temperatures in the equatorial pacific of
the coast of Peru with 3-7 year periodicities.
• Three phases- El Niño (warm phase); La Niña (cold phase) and Neutral phase.
El Niño is expressed in decreased
temperature
and
increased
winter
precipitation in the Southeastern USA.
La Niña is expressed in increased
temperature
and
decreased
winter
precipitation in the Southeastern USA.
• Major mode of climate variability in the southeast United States.
• ENSO have been found to affect the hydrologic cycle.
• Groundwater is the major source of water resources in the lower ApalachicolaChattahoochee-Flint (ACF) river basin.
• Provides an opportunity to study ENSO effects on GW levels and its interaction with
anthropogenic factors.
Objective:
• Quantify the effect of ENSO-induced climate variability on groundwater levels under
different overburden conditions.
This will allow us to develop forecasting tool to better manage GW & surface water
resources and meet anthropogenic water demands.
Study Area
About 4632 mi2 of land area contributes
groundwater and surface water to the
Upper Floridan Aquifer (UFA).
The climate of the lower ACF River
Basin is humid subtropical with long
summers and mild winters.
The ACF River Basin is at the center
of tri-state water crisis
Nearly 500,000 acres
irrigated with around
4000 wells in the
Upper Floridan Aquifer.
Fig. 1. Location of the study area, GW observation wells and geohydrologic
zones.
Three states namely Georgia, Florida
and Alabama went to court for the
basin’s finite water resources.
Upper Floridan Aquifer system (UFAS)
Overburden
Fig. 2. Conceptual diagram of groundwater and surface water flow in the
inter-connected stream aquifer system in the UFAS. (source USGS)
 Groundwater levels in the UFA respond to seasonal climatic effects such as precipitation,
droughts, stream stage and lake level changes.
Fluctuations in groundwater levels in the UFA also depend on:
•
•
•
Thickness and location specific hydraulic characteristics of the above lying USU,
Proximity to surface streams or lake system.
Groundwater irrigation withdrawal for agricultural, industrial and municipal purposes.
Data Collection and Processing
• Niño 3.4 Monthly Sea-Surface temperature (SST) data were collected from Climate
Prediction Center-NOAA.
• Daily Groundwater Level Data were collected from USGS-Georgia Water Science
Center.
• Twenty-one observation wells with 25 to 30 years of data were used.
• Groundwater level anomalies were calculated and sorted according to recharge
(December-April) and non-recharge seasons (May-November).
• The software package for Wavelet analysis was used from the Matlab code
developed by Aslak Grinsted (http://noc.ac.uk/using-science/crosswavelet-waveletcoherence).
Methodology
Wavelet Analysis
•
Wavelet analysis examines the relationship between two time series to determine the
prevailing modes of variability and their variation over the time period.
•
Used to quantify and visualize statistically significant changes in ENSO SST anomalies and GW
level variance during the historical time period.
Continuous Wavelet Transform
•
The Continuous Wavelet Transform analyses localized recurrent oscillations in time series by
transforming it into time and frequency space.
Cross Wavelet Transform and Wavelet Coherence Transform
•
Cross Wavelet Transform examines whether two time series in regions of time-frequency space
share high common power and consistent phase relationship, which might suggest causalty.
Three wells under shallow (<50 ft), moderately deep (>50 ft) and deep overburden conditions
(>100 ft) were used to study the fluctuations of groundwater levels under different overburden
conditions.
Groundwater level fluctuation with Climate Variability Analysis.
•
Non-parametric Mann-Whitney tests were used to evaluate the impacts of ENSO phases on the
medians of GW level anomalies in the ACF for recharge (December-April) and non-recharge seasons
(May-November).
•
Twenty one observation wells used for climate variability analysis.
•
The year 2000-01 was specially analyzed to study the effects of strong La Niña events (prolonged
droughts) on GW level anomalies.
Recovery Period
•
Recovery period was calculated using 3 month running averages.
•
Defined by the time required for GW level anomalies to remain above -0.25 ft for atleast 6 consecutive
3 month running averages of, after the end of the La Niña phase.
•
For calculation of the recovery periods two particular La Niña events, year 1988-89 and 2000-01,
representing short and prolonged La Niña (prolonged drought) were selected.
Results
•
Well under shallow and moderately
deep overburden conditions showed
regions of high power, though not
statistically significant, in the 3-7 year
ENSO periodicities.
SST
(a)
Shallow
(b)
Period (years)
Continuous Wavelet Spectra
(c)
Moderately
Deep
•
Wells under deep overburden, did not
exhibit any areas with high power.
(d)
Deep
Time (year)
Figure 3: Significant Wavelet Power Spectra shown within the cone-ofinfluence for (a) monthly NINO 3.4 sea surface temperatures (oC), (b)
Groundwater level anomalies (ft) for shallow overburden, (c) Groundwater
level anomalies (ft) for moderately deep overburden, (d) Groundwater level
anomalies (ft) for deep overburden.
Figures are color-mapped to indicate low powers in blue and white and high
wavelet power with reds and oranges. Black outlines indicate areas significant
to 95% confidence.
Cross-Wavelet Analysis and Wavelet
Coherence Transform
•
•
•
The Cross-Wavelet Analysis and Wavelet
Coherence Transform between SST and
GW level anomalies shows high shared
power in the areas that were seen to be
sharing high power in the single wavelet
spectra.
Wells under shallow and moderately
deep shared high and significant power
in the 3-7 year periodicities and the
significant
areas
within
this
periodicities are positively phase
locked.
These areas of shared power in CrossWavelet
Analysis
and
Wavelet
Coherence
Transform
suggests
causalty.
Well under deep overburden did not
show any shared high and significant
power in any period suggesting that
groundwater levels under deep
overburden conditions are not affected
by ENSO.
(b)
Shallow
Moderately Deep
(d)
(e)
Period (years)
•
(a)
(c)
Deep
(f)
Time (year)
Figure 4: Cross Wavelet Spectrum between NINO 3.4 sea-surface temperatures
and monthly groundwater level anomalies (ft) for overburden conditions: (a)
shallow, (b) moderately deep, and (c) deep. Wavelet Coherence Analysis
between NINO 3.4 sea-surface temperatures and monthly groundwater level
anomalies (ft) for (d) shallow, (e) moderately deep, and (f) deep. Arrows
indicate variable’s phase relationship. Black outlines indicate areas significant to
95% confidence. Arrows pointing anti-clockwise represents anti-phase behavior,
while clockwise arrows indicates in-phase behavior.
Table 1: Mann Whitney test results between ENSO phases and
monthly groundwater level anomalies for the entire period of record. P
values are significant at 0.01
Well-ID
El Niño (ft)
La Niña (ft)
Diff (ft)
p
06F001
1.35
-3.96
5.30
0.00
10G313
0.72
-2.82
3.54
0.00
08G001
3.66
-5.50
9.16
0.00
08K001
5.74
-2.42
8.16
0.00
11K003
3.35
-1.52
4.87
0.00
12L030
2.33
-2.56
4.88
0.00
12L028
1.76
-3.19
4.96
0.00
13L049
2.89
-3.68
6.57
0.00
13M006
3.24
-1.62
4.86
0.00
07H002
3.79
-2.76
6.56
0.00
12L029
1.82
-2.24
4.07
0.00
11K015
0.36
-2.03
2.39
0.00
15L020
0.60
-1.95
2.56
0.41
Median
1.36
-2.02
3.66
• Significant differences were found in
GW level anomalies for all wells,
except for well 15L020.
• High level
value<0.01).
of
significance
(p-
• The median of GW level anomalies
during the El Niño and La Niña
phases were above average and
below average respectively.
• Well 15L020 did not exhibit
significant difference due to high
overburden conditions.
Table 2: Mann Whitney test results between ENSO phases and monthly groundwater level anomalies for recharge and non-recharge
seasons. p values are significant at 0.05.
Non – Recharge (May–November)
Recharge (December-April)
p-value
Well-ID
El Niño (ft)
La Niña (ft)
diff (ft)
p-value
11.17
0.00
06F001
0.10
-3.38
3.48
0.00
-3.37
5.29
0.00
10G313
-0.30
-2.40
2.10
0.00
7.31
-7.93
15.23
0.00
08G001
2.18
-4.35
6.53
0.00
08K001
4.54
-2.21
6.75
0.00
08K001
6.27
-3.30
9.56
0.00
11K003
4.61
-3.23
7.84
0.00
11K003
2.06
-0.86
2.92
0.04
12L030
3.64
-2.77
6.42
0.00
12L030
1.47
-2.45
3.92
0.00
12L028
4.02
-4.06
8.08
0.00
12L028
0.51
-3.17
3.68
0.01
13L049
6.31
-4.42
10.73
0.00
13L049
1.17
-2.82
3.99
0.00
13M006
3.35
-1.51
4.86
0.00
13M006
2.93
-1.73
4.66
0.00
07H002
3.78
-2.47
6.25
0.00
07H002
3.81
-4.16
7.97
0.00
12L029
3.14
-2.87
6.01
0.00
12L029
1.02
-1.78
2.81
0.00
11K015
3.34
-3.21
6.54
0.00
11K015
-0.41
-1.26
0.86
0.92
15L020
0.7
-3.91
4.61
0.16
15L020
0.53
1.28
-0.75
0.75
Median
3.13
-2.83
6.13
Median
0.58
-1.68
2.32
Well-ID
El Niño (ft)
La Niña (ft) diff (ft)
06F001
5.40
-5.77
10G313
1.91
08G001
Table 3: Comparison of monthly averaged groundwater level anomalies for
severe (2000-01) and average La Niña phase during recharge and non-recharge
seasons.
Recharge
Non Recharge
Well_Id
Minimum-2000-01
La Niña
Year 2000-01
La Niña Year 2000-01
06F001
-5.20
-4.49
-2.86
-4.24
-10.44
10G313
-2.90
-6.70
-2.15
-6.38
-9.60
08G001
-5.83
-8.83
-3.68
-7.90
-14.31
08K001
-3.58
-3.02
-3.48
-6.19
-15.26
11K003
-3.21
-8.47
-1.26
-7.37
-13.66
12L030
-2.28
-3.79
-1.65
-3.83
-6.71
12L028
-3.56
-6.21
-2.09
-4.73
-10.13
13L049
-3.57
-6.13
-2.12
-5.38
-8.50
13M006
-1.93
-2.06
-3.01
-3.93
-17.43
07H002
-2.56
-3.19
-1.77
-2.18
-5.83
12L029
-3.79
-3.70
-3.76
-4.96
-10.44
11K015
-3.78
-9.07
-1.03
-7.15
-14.38
10K005
-0.51
-0.32
-0.49
-1.93
-5.48
Mean
-2.81
-4.42
-1.78
-4.29
• Average GW level anomalies were
approximately twice lower during
2000-01 than average La Niña phase
in both the recharge and the nonrecharge seasons.
• Minimum GW level anomalies for
year 2000-01 were almost 3 times
lower than average La Niña phase
values with GW level anomalies going
below 10 ft at 8 well locations and
below 5 ft at 20 at well locations.
• Wells 08G001, 08K001 and 13M006
groundwater
levels
fell
to
approximately below 15 ft during
2000-01, which demonstrates the
effect of severe and prolonged La
Niña can have on groundwater levels.
Table 4: Comparison of recovery periods
(months) for prolonged (2001) and short
(1989) La Niña phase.
Well-ID
Year 2001
Year 1989
06F001
18
1
10G313
24
8
08G001
18
1
08K001
18
0
11K003
25
2
13L012
25
0
12L030
26
2
12L028
26
6
13L049
26
1
13M006
25
1
12L029
25
1
11K015
25
8
10K005
25
0
Mean
22
2
• Year 2000-01 representing severe La Niña phase
shows significantly higher recovery times than the
year 1988-89.
• The average recovery time for year 2000-01 was 22
months as compared to 2 months for 1988-89.
• Might be due to increased irrigation.
Conclusion
• Wavelet Analysis showed that wells under shallow and moderately deep overburden
conditions exhibit ENSO signals while wells under deep overburden conditions does
not exhibit such a relationship.
• Mann Whitney test results validates the above relationship.
• GW level anomalies tended to be above average during El Niño phase and below
average during La Niña events.
• ENSO signals are stronger during recharge season than non-recharge.
• Severe La Niña events can severely affect groundwater resources and their recovery
periods thereby threatening sustainability.
• Results indicate a potential for possible groundwater level prediction with respect to
ENSO phases.
Future Research
• Quantify how pumping for irrigation exacerbates the effect of La Niña on
groundwater levels, and
• Develop procedure for forecasting groundwater levels using ENSO forecasts.