2017 dengetalnh human activities and slr on swi yucatan peninsula

Nat Hazards
DOI 10.1007/s11069-016-2621-5
ORIGINAL PAPER

The integrated impacts of human activities and rising sea
level on the saltwater intrusion in the east coast
of the Yucatan Peninsula, Mexico
Yujun Deng1 • Caitlin Young2 • Xinyu Fu1 • Jie Song1
Zhong-Ren Peng1

•

Received: 4 February 2016 / Accepted: 6 October 2016
Ó Springer Science+Business Media Dordrecht 2016

Abstract Saltwater intrusion is a major hazard to coastal communities as it causes
degradation of fresh water resources. The impact of rising sea level on the saltwater
intrusion into coastal aquifers has been studied for decades, but how human activities affect
the extent of saltwater intrusion is poorly understood. Human activities are known to
influence groundwater availability indirectly by affecting precipitation patterns and
directly by extracting groundwater and reducing recharge. In this paper the authors
investigated the integrated impacts of human activities and rising sea level on aquifer
recharge in Quintana Roo, Mexico, by incorporating anthropogenic impacts on groundwater recharge into an analytical saltwater intrusion model. The impact of human activities
on groundwater extraction was firstly calculated; then, the resulting groundwater recharge
was used in a Ghyben–Herzberg analytical model to determine the inland distance of
saltwater intrusion. The analytical model tested six scenarios stemming from different
combinations of human development patterns, hydrological settings, hydraulic conditions
and rising sea level to obtain the range of possible inland movement of saltwater intrusion.
Our results indicate that the groundwater recharge will decrease to 32.6 mm year-1 if
human activities increase by 50 % more. With 1-m sea level rise, inland saltwater intrusion
distance is estimated to be up to 150 and 1 km under head-controlled and flux-controlled

scenarios, respectively. A sensitivity analysis of the model reveals that the large hydraulic
conductivity of the Quintana Roo aquifer (0.26–68.8 m s-1) is the most important factor in
determining saltwater intrusion distance. Therefore, in this aquifer, the response to human
activities is greatly exceeded by natural hydrogeological conditions.
Keywords Rising sea level  Saltwater intrusion  Human activities  Coastal water
resources

& Zhong-Ren Peng

1

Department of Urban and Regional Planning, University of Florida, Gainesville, FL 32603, USA

2

Department of Geological Sciences, University of Florida, Gainesville, FL 32603, USA

123


Nat Hazards

1 Introduction
Saltwater intrusion is the invasion of saltwater into fresh groundwater in coastal aquifers
due to changes in hydraulic gradient at the coastline (Bruington 1972; Barlow and
Reichard 2010). It is known to contaminate potable water wells and even jeopardize water
supply in vulnerable communities that rely on sole source aquifers (Vengosh and Rosenthal
1994; Feseker 2007; Germain et al. 2008; Barlow and Reichard 2010; Liu and Dai 2012;
Cai et al. 2015). Whereas some communities are more resilient to the saltwater intrusion
due to local environmental characteristics such as abundant precipitation, low evapotranspiration or low hydraulic conductivity, all of which can mitigate the effects of saltwater

intrusion (Feseker 2007; Qi and Qiu 2011). However, the local environment can be altered
by human activities through changing land use, increasing the amount of impervious
surface as well as increasing groundwater extraction, which in turn alter fresh water stores
in the local aquifer (Nicholls and Cazenave 2010; Beck and Bernauer 2011; Goudie 2013).
Thus, it is essential to consider the impacts of human activities to obtain an accurate
estimation of saltwater intrusion in association with rising sea level. The overexploitation
of groundwater in Taiwan and coastal Tunisia both altered local groundwater hydrology,
which allowed for extensive saltwater intrusion and led to eliminate many potable water
wells (Willis and Finney 1988; Narayan et al. 2003; El Ayni et al. 2013). As a consequence, local residents may continue to consume water from salt-contaminated wells under
clear health risks. In the coastal communities of Bangladesh that experiences severe
saltwater intrusion, people consume much more salt compared to people inland, thereby
putting them at risk of hypertension (Rasheed et al. 2014).
In response to the potential hazards, the shrinkage of local fresh water reservoirs caused
by saltwater intrusion necessitates an effective community planning, which requires an
understanding of what drives saltwater intrusion in the local aquifer. Rising sea level is
recognized as one important factor increasing saltwater intrusion, but the extent of intrusion is also related to the aquifer composition and local hydraulic conductivity (Werner and
Simmons 2009; Chang et al. 2011; Guha and Panday 2012; Carretero et al. 2013; Werner
et al. 2013). Current estimates of sea level rise predict rates will exceed 3.0 mm year-1,
while this number could go even higher depending on the stability of the West Antarctic
ice sheet (Church and White 2006; Rahmstorf 2007; Nicholls and Cazenave 2010; Church
et al. 2013). The fifth assessment report of Intergovernmental Panel on Climate Change
(IPCC) points out that sea level will continue to rise more than 1 m by 2100 in the worst
case (Pachauri et al. 2014). Thus, coastal communities will be faced with the intensified
saltwater intrusion and potential water shortage with sea level rise (Werner and Simmons
2009; Chang et al. 2011; Guha and Panday 2012; Carretero et al. 2013). Nonetheless, it is
possible to decrease the effects of saltwater intrusion by implementing sustainable water
management policies. Many pumping strategies have been proposed to manage saltwater
intrusion, but these strategies are site specific (Willis and Finney 1988; Hallaji and
Yazicigil 1996; Cheng et al. 2000; Mantoglou 2003; Sreekanth and Datta 2013; Cai et al.
2015). In addition to site-specific pumping strategy, both current and future development

projections should be included in the water management plan to sustain local water
resources (Alley et al. 1999). Moreover, many researchers have pointed out that human
activities impose compounding impacts on environment including alteration in temperature
and precipitation, which results in changes in evapotranspiration (Santer et al. 1996;
Vitousek et al. 1997; Tett et al. 1999; Mitchell et al. 2001; Thorne et al. 2002; Hegerl et al.
2004; Sun et al. 2008; Min et al. 2011). Therefore, an effective water management plan

123


Nat Hazards

must start with an understanding of the synergy between local hydrogeology and human
activities in order to determine where and how much human activities actually control the
extent of saltwater intrusion.
The increases in human activities, e.g., population, vehicle and agricultural land, would
contribute to the growth in water consumption and subsequently the groundwater extraction (Gleick 1996; Peters and Meybeck 2000; Voăroăsmarty et al. 2000; Schnoor
2015; Mehdizadeh et al. 2015). Groundwater recharge can be viewed as the net of precipitation minus evapotranspiration, overland flow and groundwater extraction. Therefore,
managing human activities might be the only approach to sustain groundwater recharge as
precipitation, evaporation and hydraulic condition are inherent properties of the climate
and aquifer (Carretero et al. 2013). In this study we incorporate human activities into the
groundwater recharge term of a Ghyben–Herzberg and the Dupuit-Forchheimer solution
analytical hydrological model in order to determine how humans will alter the hydrological
balance in a karst aquifer undergoing sea level rise.
Including human activities in estimating saltwater intrusion should provide better
understanding that how human impacts will alter the hydrogeological balance during saltwater intrusion which will lead to more effective water management planning. The authors
employ a partial least square regression (PLSR) model to determine the impact of human
activities on aquifer recharge and extraction (Q). PLSR was developed as an econometric
tool, but it is also widely used in chemometrics, chemical engineering and monitoring
industrial processes. This method can construct a dependable mathematical model with

adequate predictive power even when factors are many and highly collinear (Tobias 1995;
Martens and Dardenne 1998; Helland 2006; Rosipal and Kraămer 2006; Jie et al. 2007; Mevik
and Wehrens 2007; Carrascal et al. 2009; Abdi 2010). In this study the authors use PLSR to
predict the impact of varying extent of human activities on groundwater extraction and
groundwater recharge. These values are then incorporated into a saltwater intrusion model to
estimate the inland distance of saltwater intrusion under a scenario of up to 1-m sea level rise.
To illustrate the integrated impacts of human activities and rising sea level on the
saltwater intrusion, six scenarios resulted from the combination of four human development patterns, three hydraulic conditions and up to 1 m of sea level rise, were tested under
two hydrological settings: (1) flux-controlled scenarios, where the groundwater discharge
rate will stay still despite the rising sea level, and (2) head-controlled scenarios, where the
inland aquifer head will not rise despite the rising sea level. In any particular place, it is
possible to have either flux-controlled or head-controlled scenario or the combination of
them (Carretero et al. 2013). Thus it is imperative to assess both models to provide
information for water management. In this study, we quantify the migration of toe of
saltwater intrusion as a function of sea level rise, groundwater recharge, ground fresh water
discharge, hydraulic conductivity and water table under sea level.

2 Study area
The state of Quintana Roo is located on the east coast of the Yucatan Peninsula, Mexico.
The aquifer is comprised of a large carbonate platform with high hydraulic conductivity,
with macroscale connections (conduits) between the fresh water aquifer and the coastal
ocean. The coastal zone has undergone rapid population and tourism growth from 1980 to
2012. According to the database of Instituto Nacional de Estadı´stica y Geografı´a (INEGI,
the population was 220,000 in 1980 and increased to 1,300,000

123


Nat Hazards

in 2012 (INEGI Information Databank; INEGI ‘‘Population and Housing Census 2010’’),
simultaneously annual tourist influx increased from one million to nearly nine million
(INEGI Information Databank). Therefore, growth in human activities has resulted in a
heavily impacted coastal aquifer in Quintana Roo which is in need of a long-term water
management plan. This plan should account for the relationship of human activities with
groundwater extraction and recharge. Groundwater extraction in Quintana Roo has
increased from 29.11 million cubic meters of water to 873 million cubic meters of water
between 1980 and 2012 (Comisio´n Nacional del Agua). Quintana Roo has undergone
increases in tourism for three decades, and projected annual tourism rate indicates accelerated rate of growth for the next decade. Thus groundwater extraction, due to both
expanding population and tourism, is likely to increase in the future.
Quintana Roo is comprised of a karst aquifer, which is a sole source aquifer for drinking
water. The minimum air temperature and the median air temperature both have increased
1–2 °C since 2004 (Comisio´n Nacional del Agua, ‘‘Servicio Meteorolo´gico Nacional’’
2014). The average air temperature increased from 25.7 °C in 2004 to 27 °C in 2014,
which leads to an increase in the evaporation rates. The annual report of Comisio´n
Nacional del Agua points out that the annual evaporation is larger than the annual precipitation. Hence the aquifer recharge primarily occurs during the wet season when precipitation is greater than evaporation. Average precipitation and evaporation in the wet
season from 1980 to 2012 were 777.67 and 719.52 mm year-1, respectively. Due to the
hydraulic conductivity, overland flow is negligible as water rapidly infiltrates the highly
fractured carbonate rocks (Comisio´n Nacional Del Agua [National Water Commission of
Mexico] 2010, 2011, 2012, 2015) (Fig. 1).
Table 1 gives the aquifer parameters used to estimate the migration of saltwater intrusion;
these values were obtained from field experiments and derived from the literature. A Nortek
Vector Acoustic Doppler Velocimeter (ADV) was placed at an offshore spring vent. Velocity
components (u, v, w), temperature and pressure data were collected at 64 Hz in 10-min bursts
every 30 min. Data beyond three standard deviations from the mean were considered spurious and were replaced by the average of the nearest neighbors (Monismith et al. 2010). The
conduit outflow data were multiplied by the area of the vent opening (0.62 m2) to estimate
volumetric flow rate into and out of the mouth, q (m3 s-1). As the offshore spring cycled
between discharge and backflow conditions, the average groundwater discharge rate was
calculated by averaging the flow rate only during times of discharge over the two-week
period of the field experiment (Fig. 2). Hydraulic conductivity (K) for Quintana Roo is

highly variable, with values ranging from 0.26 to 68.84 m s-1 (Bauer-Gottwein et al. 2011).
We used these maximum and minimum values to model the length of saltwater intrusion in
addition to a K value calculated from data gathered during this study. Two CTDs continuously recorded pressure measurements, one located at Pargos spring and the other at a Cenote
located 9 km inland. The difference between tidal amplitude was used to calculate K, following a method suitable for karst aquifers (Martin et al. 2012).

3 Methodology
3.1 Partial least square regression
As discussed, PLSR can construct predictive model even when predictors are highly
correlated. PLSR aims to extract the underlying or latent factors that account for most of

123


Nat Hazards

Fig. 1 The state of Quintana Roo on the Yucatan Peninsula (inset) with study area outlined. Enlarged study
area shows Pargos, an offshore spring used to determine q0 and UNAM well, used to confirm values for
hydraulic conductivity

Table 1 Aquifer parameters used in saltwater intrusion modeling
Parameter

Value

Source

Groundwater density ratio

40

Carretero et al. (2013)

Aquifer depth (m)

40

Gondwe et al. (2010)

Horizontal water discharge (q0) (m3 s-1)

0.48

Field data

Hydraulic conductivity (m s-1)

68.8

Bauer-Gottwein et al. (2011)

Hydraulic conductivity (m s-1)

0.26

Bauer-Gottwein et al. (2011)

Hydraulic conductivity (m s-1)

0.66

Field data

123


Nat Hazards

Fig. 2 Discharge volume of water from Pargos spring during March–April 2014. Positive values indicate
times when the aquifer was discharging to the lagoon, while negative values indicate times when the lagoon
was backflooding water into the aquifer. Horizontal fresh groundwater flow to the sea per unit length of
shoreline was calculated by averaging the positive discharge values and then multiplying this average by the
average fresh fraction of the discharging water. The average fresh fraction was determined from the average
salinity of the discharging water

the variation in the response although there are many predictors. Figure 3 indicates the
schematic outline of PLSR, whose ultimate goal is to predict the response given by the
predictors in the population. To achieve this goal, the X score, t, and Y score, u, will be
firstly extracted from the predictors (represented by matrix X) and responses (represented
by matrix Y) in the sample. Then scores t are used to predict u which will subsequently be

Sample
T

Regression

Extract
Latent
Variables

U

Extract
Latent
Variables

Predictor

Regression

Response

Population
Predict

Predictor

Fig. 3 The schematic outline of PLSR

123

Response


Nat Hazards

used to predict the responses. Generally, PLSR is a process of indirect modeling, constructing the prediction for responses given by the predictors.
To be more specific, PLSR iteratively obtained the components, called latent variables
or principal components. It starts with the singular value decomposition of the crossproduct matrix S ¼ X T Y, extracting information regarding the variation and correlation of
X and Y. The first left and right singular vectors, w and q, are the weights for X and Y,
respectively. To obtain X score and Y score:

t ẳ Xw ẳ Ew

1ị

u ¼ Yq ¼ Fq

ð2Þ

where E and F are initialized as X and Y. The scores t and u are often normalized:
t
t ẳ p
tT t

3ị

u
u ẳ p
uT u

4ị

p ẳ ET t

5ị

q ¼ FT t

ð6Þ

To obtain the factor loadings:

Finally, the information regarding this principal component must be subtracted from
current data matrices E and F.
Enỵ1 ẳ En  tpT

7ị

Fnỵ1 ẳ Fn  tqT

8ị

Therefore, the next component must be estimated from the singular value decomposition of
the cross-product matrix En?1Fn?1. The vectors w, t, p and q are stored in the corresponding matrices W, T, P and Q, respectively, after each iteration. Then scores T can be
used to calculate the regression coefficients, and later the normalized variables are converted back:

1
9ị
R ẳ W PT W

1
B ẳ R T T T T T Y ẳ RQT

10ị

The sum of squares of X explained by the principal components is calculated by PPT, and
the proportion of explained variance is obtained by dividing the sum of explained variance
by the corresponding sum of total variance. The optimal subset of principal components is
determined using cross-validation.
The authors developed a predictive model to reveal the impacts of human activities on
groundwater extraction. To be accurate, this study selected the predictors from the most

common human activities in the following categories: industry, transportation, agriculture,
livestock, economy, demographics and human waste. Table 2 shows the predictors
selected to represent the human activities in different fields and the time variable, trend.
The variable, trend, was a dummy variable representing the effects of other factors

123


Nat Hazards
Table 2 Description of predictors
Category

Indicator

Description

Industry

Volume of
timber

The timber is the primary industrial produce in Quintana Roo

Automotive

The increase in vehicles and road reveals the urban expansion and
economic growth

Transportation

Motorcycle
Length of road
Agriculture
Livestock

Total area
sown

The increase in agricultural area indicates the growth in irrigation

Cattle

Cattle and poultry are the primary livestock in Quintana Roo

Poultry
Economy

Tourists

Tourism is the driving force of economic growth in Quintana Roo

Lodging room
Hotel
Demographics

Population

Population growth increases the water demand

Human waste

Garbage

Increasing garbage reflects the growth in human activities

Time

Trend

Reduce the dynamics of data along time

excluding human activities. Then this model can to the best reduce the disturbance from
other factors.
Due to the high collinearity, the authors applied PLSR to the predictors to extract the
principal components, finding the largest variation in these predictors. The resulting
variation represented the hidden information behind the predictors, and this information
should be able to explain the original data. Afterward, these principal components
extracted from predictors were used to construct a model with the principal components
extracted from the response, groundwater extraction. Subsequently, the regression between
them later was converted back to the original predictors and responses. To have a validly
predictive model, the principal components should be able to explain 80 % or more
variation in original data, and the cross-validation indices should be larger than 0.0975 as
well.

3.2 Saltwater intrusion length estimation
In this study, both the Ghyben–Hertzberg relationship and the Dupuit–Forchheimer
approximation were adopted to estimate the integrated impact of rising sea level and
human activities on the saltwater intrusion for a homogeneous, isotropic, unconfined
aquifer subjecting to a constant recharge and steady-state conditions. The horizontal flux of
fresh water at position (xi) was computed using the following equations (Falkland 1991):

qi xi ị ẳ q0  Wxi ẳ K h ỵ ahịdh=dxị

11ị

which was integrated and transferred into (Custodio and Bruggeman 1987)
h2i ¼

2q0 xi  Wx2i
K 1 ỵ aị

12ị

where xi (m) is the distance from inland to coastal shore, and hi (m) is the elevation of the
water table at inland position xi (m), W (mm year-1) is the net recharge, q0 (m3 s-1) is the

123


Nat Hazards

Fig. 4 Illustration of saltwater intrusion model variables used for the coastal aquifer in Quintana Roo, MX.
Groundwater recharge (W) varied according to human activities. The new position of the saltwater toe
(x) was modeled, and the movement of the salt water–fresh water interface was described as Xt. Aquifer
depth (z), horizontal movement of groundwater (q0) were determined from literature and field data. The
system exhibits multiple types of permeability leading to a range of hydraulic conductivity, from low values
associated with matrix permeability to pipe-flow permeability associated with conduit flow

horizontal fresh groundwater flow to the sea per unit length of shoreline, and K is the
hydraulic conductivity (Table 1). The net recharge is calculated from the following
equation:

W ¼ precipitation  evapotranspiration  overland flow  Q

ð13Þ

where precipitation and evapotranspiration stemmed from historical average value, overland flow is negligible given few surface water bodies are in Quintana Roo and Q is
groundwater extraction which was calculated with PLSR model. The ground water density
ratio, a, is equal to qf/(qs - qf), where qf is the density of fresh groundwater and qs is the
density of saltwater and a is assumed to be 40. In all the scenarios, the toe of the saltwater
wedge is designated as xt (m), which is calculated as (Custodio and Bruggeman 1987):
r
q0
q20 K 1 ỵ aịz20

xt ẳ 
14ị
W
W2
Wa2
where z0 (m) is the depth of the aquifer below the mean sea level. The rising sea level will
change z0 (m). In saltwater intrusion predictions, the water table elevation (hi) is required
and it can be determined by a simple steady-state mass balance and Darcy’s law to be:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi


2
W
hðxi Þ ẳ
15ị
xi  xtị q0  xi ỵ xT ị ỵ hT ỵ z0 ị2
k

2
where hT (m) is the water table height at the toe of sea water wedge, and xT (m) is the
distance from the toe of salt water–fresh water interface to the inland.

123


Nat Hazards

In the flux-controlled scenarios, the horizontal discharge of fresh groundwater to the
sea, q0 (m3 s-1), is constant despite the sea level changes. However, the water table elevation, hi (m), will rise at the same rate of sea level rises to maintain a constant flow. On
the contrary, in the head-controlled scenarios the inland head is consistent regardless of the

Table 3 Modeling scenarios
Scenarios

Increase in human
activities (%)

Hydrological
settings

Hydraulic conductivity
(m s-1)

A

0–50

Flux controlled

68.8

0–1

B

0–50

Head controlled

68.8

0–1

C

0–50

Flux controlled

0.26

D

0–50

Head controlled

0.26

0–1

E

0–50

Flux controlled

0.66

0–1

F

0–50

Head controlled

0.66

0–1

Sea level rise
(m)

0–1

Groundwater recharge was computed based on the increase in human activities which are stated in Sect. 4.2.
Then the results were used to estimate saltwater intrusion length in association with hydrological settings,

parameter sets and sea level rise

Table 4 R-squared coefficient of groundwater extraction model
t1

t2

t3

0.2319

0.0036

R-squared coefficients for each component
Extraction model

0.7534

Table 5 Explained variance of
each variable in PLSR model

Extraction model
t1

t2

t3

Explained variance of X–Y by T

123

Automotive

0.8919

0.9953

0.9956

Motorcycle

0.7585

0.9945

0.9973

Tourists

0.9773

0.9833

0.9909

Lodging room

0.9723

0.9943

0.9973

Population

0.9904

0.9928

0.9939

Garbage

0.9670

0.9850

0.9857

Total area sown

0.7761

0.9931

0.9953

Length of road

0.7282

0.9282

0.9997

Stocks of cattle

0.8010

0.9673

0.9815

Trend

0.9779

0.9951

0.9951

Use

0.7534

0.9853

0.9889

Nat Hazards
Table 6 Validation of water consumption model
PRESS

RSS

Q2

LimQ2

Q2cum

0.3789

0.4561

0.1694

0.0975

0.9851

Cross-validation indices
Usage model

Table 7 Regression coefficients
of water consumption model

Extraction model

Regression regular coefficients
Intercept

9.270E?01

Automotive

1.003E-03

Motorcycle

4.138E-03

Tourists

1.327E-05

Lodging room

4.416E-04

Population

5.580E-05

Garbage
Total area sown

1.944E-01
-1.278E-03

Length of road

6.682E-03

Stocks of cattle

-1.972E-03

Trend

1.410E?00

rising sea level. We calculate the distance of salt water–fresh water interface, xt(m), for
both head- and flux-controlled scenarios (Fig. 4).

3.3 Application of test scenarios
To precisely reveal the impact of human activities on groundwater extraction and saltwater
intrusion, this study tested a variety of combinations of development patterns, hydrological
settings and hydraulic conditions. Though the development patterns are represented by four
different levels in response to 0 % growth, 10 % growth, 30 % growth and 50 % growth,
respectively, they will be integrated together with different hydrological setting and
hydraulic conditions in estimating saltwater intrusion. In other words, the human activities
that induced groundwater extraction and sea level rise were control variable in the tests and
their effects under different scenarios would be obtained through sensitivity analysis. The
sea level is assumed to rise 1 m, which is very likely to happen before twenty-second
century on the basis of IPCC reports and other researchers’ studies. Three different values
of K and two different hydrological settings were used to investigate the model’s sensitivity to both hydraulic conditions and human activities (Table 3).

4 Results

4.1 Groundwater extraction model
Table 4 gives the R-squared coefficients of groundwater extraction model by each component. It fully summarized the information of all variables in the model owing to an R-

123


Nat Hazards

square value of 0.003571 at the third component. Table 5 shows the extent to which the
information of each variable is captured by the components. Almost all the variables were
captured well by the first component, which was able to explain at least 73 % variation for
both predictors and response. Then the leftover information in the data was obtained by the
second and third principal component, which extracted about 99 % information from
original data in the end, enabling the authors to construct a validly predictive model.
The cross-validation results of this model at third component are shown in Table 6. The
h
indices were computed using equation, Q2h ¼ 1  PRESS
RSSh1 , where RSSh-1 was the squared
sum of residuals using the th-1 component, and PRESSh was the prediction error sum of
squares using the th component. A component th was considered to be significant if Q2h was

Fig. 5 Comparison between origin and prediction in water consumption model

Impacts of Human Activties on Groundwater

45.00
40.00
35.00
30.00
25.00

20.00
15.00
10.00
5.00
0.00

Zero Development
(0%)

Low Development
(10%)

Groundwater Recharge W (mm/year)

Median Development
(30%)

High Development
(50%)

Groundwater Abstraction Q (mm/year)

Fig. 6 Impacts of human activities for different development patterns ranging from 0 to 50 % increase in
human activities

123


Nat Hazards
30% Percent of q0 is fresh water

a 1000

ΔT (m)

800
600
400
200
0

1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.042

0.04

0.038

Water Recharge (m/year)

70% Percent of q0 is fresh water

b 500

ΔT (m)

400
300
200
100
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

100% Percent of q0 is fresh water

c

300

ΔT (m)

250
200
150
100
50
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

Fig. 7 Saltwater intrusion length for scenario A: a 30 % of q0 is fresh water; b 70 % of q0 is fresh water;
c 100 % of q0 is fresh water

123


Nat Hazards
Fig. 8 Saltwater intrusion length for scenario B: a 30 % of q0 is fresh water; b 70 % of q0 is fresh water; c
c 100 % of q0 is fresh water. In response to the decreasing water recharge and other parameters in scenario
B, the qmin ranged from 60,329 to 53,852 m2 s-1

greater than or equal to LimQ2 = 0.0975. In other words, the model had significant
predictive power at third component.
Table 7 shows the regression coefficients of variables selected in the model. Except for
agriculture and livestock, the other human activities had positive coefficients, which collectively indicated that groundwater extraction will increase with the growth in human
activities. Specifically, the growth in transportation, tourism and demographics all contributed to the growth in groundwater extraction. The transportation was more than a proxy
to the local economy, the increase in transportation facilitated the delivery of bottled water,
which has become the primary source of drinking water for the majority Mexicans (Green
and MacQuarrie 2014). Though the per capita bottled water consumption in Mexico was
already the greatest in the world, 69.8 gallons per capita in 2014, it is very likely that the
number will continue increasing due to the deficient infrastructure of public supply system
(International Bottled Water Association n.d.). Accordingly, the transportation will also
develop to satisfy the demand of delivering the bottled water, which is extracted from more
pristine aquifers away from population centers.

Figure 5 shows the comparison between original data and prediction for groundwater
extraction. The goodness of fit for this model was acceptable given the prediction fitted the
original data very well and the value of root-mean-square error (RMSE) was 25.46. The
small value of RMSE indicated the small differences between original data and prediction.
More than that, the figure also revealed the trend that the groundwater extraction increased
very quickly in recent years.

4.2 Scenario analysis
The critical variables tested in the scenario analysis include groundwater recharge (W),
hydraulic conductivity and flux- versus head-controlled aquifer systems. Groundwater
recharge (W) was determined by using the PLSR model to determine how the growth in
human activities will increase the amount of groundwater extraction (Q) and alter recharge
due to changes in human activities, according to Eq. 13. The results of this analysis are
shown in Fig. 6. Starting with the current groundwater recharge and Q values of 40.1 and
17.3 mm year-1, respectively, we see marked decreases in both values with increasing
human activities. The maximum amount of increase calculated, 50 % more than current
activities, will decrease the annual recharge to 32.6 mm year-1 and simultaneously
increase Q to 25.5 mm year-1. The resulting W used in scenario analyses ranges from
0.032 m year-1 (0 % development) to 0.042 m year-1 (50 % development).
Three values of hydraulic conductivity were used to explore the importance of human
activities on saltwater intrusion. The hydraulic conductivities were selected based on their
representation in the literature and field-determined values. The hydraulic conductivities
used were 68.8 m s-1 for scenario A and B, 0.26 m s-1 for scenario C and D and
0.66 m s-1 for scenario E and F (Table 3). The maximum value was 68.8 m s-1, calculated by Bauer-Gottwein in 2011. Field-derived data were used to calculate the hydraulic
conductivity between an inland Cenote and Pargos spring, which was found to be
0.66 m s-1. Finally, a minimal value 0.26 m s-1 was chosen as calculated by BauerGottwein.

123

Nat Hazards

30% Percent of q0 is fresh water

a

104

15

ΔT (m)

10

5

0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

70% Percent of q0 is fresh water

b

104

8

ΔT (m)

6
4
2
0
1
0.5

Sea Level Rise (m)

0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

100% Percent of q0 is fresh water

c

104

5

ΔT (m)

4
3
2
1
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

123


Nat Hazards
Fig. 9 Saltwater intrusion length for scenario C: a 30 % of q0 is fresh water; b 70 % of q0 is fresh water; c
c 100 % of q0 is fresh water

The maximum groundwater discharge (q0) measured was 0.48 m3 s-1 in study area
(Fig. 2). Three cases representing 30, 70 and 100 % fresh water discharge during offshore
discharge were tested in each scenario. The recorded salinity values for Pargos spring
ranged between 22 and 35. Therefore a 30 % fresh water discharge scenario is closest to
current conditions at this offshore spring. Other offshore springs sampled as part of a larger
study in Quintana Roo ranged in salinity from 14 to 27, so the scenarios encompass the

range of observed fresh water fractions in discharging offshore springs (Null et al. 2014).
In flux-controlled scenarios, the groundwater discharge would not change despite the
sea level rises and the head in aquifers would rise at the same rate as sea level rises. In
contrast, for head-controlled scenarios, the hydraulic gradient will shallow and groundwater discharge will decrease. For head-controlled scenarios the minimal groundwater
discharge in study area was calculated according to Eq. 16.
r
WK 1 ỵ aịz20
16ị
qmin ẳ
a2
where W is recharge, K is hydraulic conductivity, z0 is depth of aquifer, and a is
groundwater density ratio as described above. The groundwater discharge would at most
decrease to qmin in head-controlled scenarios when sea level rises.
Flux- and head-controlled scenarios were tested using three different hydraulic conductivity values. Assuming a hydraulic conductivity of 68.8 m s-1 and a 30 % fresh
discharge (q0), the maximum inland penetration of saltwater is 1000 and 15,000 m under
head- and flux-controlled scenarios, respectively (Figs. 7, 8). In both cases an increase in
human activities, and the consequent decrease in W, caused the saltwater to penetrate
further inland. Under a flux-controlled scenario (Fig. 7) saltwater intrusion length would
increase with the rising sea level no matter how much fresh water was discharged from
aquifer to ocean. However, as the fraction of fresh water in the discharging water
decreased, the more saltwater intrusion penetrated inland. Saltwater penetrated 1000 m
inland under a 30 % fresh water scenario, but this distance decreased to 300 m when
100 % of discharge was fresh water. The qmin ranged from 60,329 to 53,852 m2 s-1 here
(Fig. 8).
Scenarios C and D use a hydraulic conductivity 0.26 m s-1, as which is a value
observed in other studies of Yucatan karst aquifer. The lower hydraulic conductivity
resulted in less landward penetration of the saltwater toe (Figs. 9, 10). In a flux-controlled
scenario the maximum inland penetration of saltwater is 4 m, which is significantly less
than the maximum observed in scenario A, which reached a maximum of 1000 m inland.
Under head-controlled conditions when the discharge is 30 % fresh water the saltwater

intrusion length is 700 m inland, and qmin ranges 3707–3515 m2 s-1.
The final scenarios, E and F, utilized a hydraulic conductivity of 0.66 m s-1, which was
determined from field data gathered in this study. Flux-controlled scenario finds that when
discharging groundwater (q0) is 30 % fresh saltwater will penetrate 10 m inland, but this
value decreases to 3 m inland when discharging groundwater is 100 % fresh (Fig. 11).
Under head-controlled conditions, scenario F, saltwater will penetrate 500–1600 m inland,
depending on the fresh water composition of the discharging groundwater (Fig. 12). In this
scenario qmin ranges 5907–5207 m2 s-1.

123


Nat Hazards

123


Nat Hazards
Fig. 10 Saltwater intrusion length for scenario D: a 30 % of q0 is fresh water; b 70 % of q0 is fresh water; c
c 100 % of q0 is fresh water. In response to the decreasing water recharge and other parameters in scenario
D, the qmin ranged from 3707 to 3515 m2 s-1

5 Discussion
Results from this study found that human activities impacted both the amount of
groundwater extraction (Q) and the amount of aquifer recharge. A 50 % increase in human
activities over today’s current level would result in a decrease of *10 mm year-1 in
groundwater recharge and an equivalent increase in groundwater usage (Fig. 6). The PLSR
modeling revealed that transportation has a large impact on the amount of groundwater
usage and recharge (Table 5). This information can be utilized by local governments to
effectively manage water resources and develop strategies to offset local water usage as

population expands. Mahesha and Lakshmikant (2014) hypothesized that one effective
way to minimize the saltwater intrusion is by identifying the optimal location and rate of
fresh water pumping. Our results confirmed that the rate of fresh water pumping is critical
in determining the extent of saltwater intrusion.
Although human activities do impact the extent of saltwater intrusion in Quintana Roo,
the magnitude of their impact was small compared to how hydrogeological conditions
impact saltwater intrusion. For example, in scenario F, which utilized field-derived
hydraulic conductivity value of 0.66 m s-1, differences in water recharge attributed to
human impacts alter the inland salt migration distance by less than 100 m (Fig. 12). In
comparison, the fraction of fresh water discharging from the aquifer altered the inland
distance of saltwater intrusion by more than 1 km (Fig. 12). Thus, the fraction of discharging fresh water played a key role in determining the length of saltwater intrusion.
Despite this, there are little data regarding the range of discharging water composition in
Quintana Roo. Measured salinity values in the Puerto Morelos lagoon area in this study
range from 4.2 in groundwater discharging from the beach to 27.8 in water discharging
from offshore springs (Null et al. 2014). Therefore, uncertainty regarding the composition
of discharging groundwater makes determining the exact extent of saltwater intrusion
challenging.
Another critical factor affecting saltwater intrusion length is the hydraulic conductivity
in Quintana Roo. The saltwater intrusion length decreased with less hydraulic conductivity
for both flow-controlled and head-controlled settings. In the extreme case, when the
hydraulic conductivity was 68.84 m s-1, the saltwater intrusion was able to penetrate
150 km, which would contaminate about 1860 billion cubic fresh water. However, the loss
would dramatically decrease to about 2640 million cubic fresh water if the hydraulic
conductivity reduced to 0.26 m s-1. The dynamics of hydraulic conductivity in Quintana
Roo makes the aquifer vulnerable to the sea level rise, whose impacts on saltwater
intrusion in Quintana Roo were larger than in other well-studied coastal aquifers, where
rising sea level was found to be the least significant factor in determining saltwater
intrusion (Green and MacQuarrie 2014).
Though the hydraulic conductivity and sea level rise to a great extent determined how
saltwater intrusion affects local aquifers in Quintana Roo, the varied hydrological settings

also played an important role. As shown in Fig. 4 the aquifer is comprised of both matrix
and conduit permeability. Therefore it is likely that our predictions of saltwater intrusion
are accurate at different scales. Conduits allow for the rapid flow of water exchange
between the coast and the aquifer. A hydraulic conductivity of 68.8 m s-1 is reasonable in

123


Nat Hazards

123


Nat Hazards
Fig. 11 Saltwater intrusion length for scenario E: a 30 % of q0 is fresh water; b 70 % of q0 is fresh water; c
c 100 % of q0 is fresh water

a conduit setting. Thus within conduits it is likely that saltwater intrusion will extend 15 to
20 km, as results from scenario B suggest (Fig. 8). The Quintana Roo aquifer also contains
low permeability matrix rock, as modeled using a conductivity of 0.26 m s-1. Results from
head-controlled modeling of this hydraulic conductivity (scenario F) indicated that saltwater penetration would reach a maximum of 1.5 km, an order of magnitude lower than
penetration within conduits (Fig. 12). Thus, the aquifer will likely have multiple salt
water–fresh water interfaces that affect potable water quality for residents living in the
coastal zone.
Compared with head-controlled settings, in which the saltwater intrusion length
increased exponentially, the flux-controlled settings were much less vulnerable given the
saltwater intrusion could penetrate much less than in head-controlled settings (Werner and
Simmons 2009). However, the places where the flux-controlled setting dominates can be
turned into head-controlled setting once the groundwater extraction exceeds its recharge,
due to lowering of the water table (Carretero et al. 2013). In order to maintain sufficient

recharge for a flux-controlled system, water management agencies will need to actively
control human activities, thereby preventing human alteration of the hydrological balance.
Currently, sufficient precipitation in Quintana Roo sustains groundwater recharge,
whereas the karstic islands in Bahamas experience water scarcity although with abundant
precipitation (Jones 2014; World Bank 2015). Presently evaporation rates do not outstrip
precipitation during certain portions of the year, but this could change due to changes in air
temperature associated with climate change (Roderick and Farquhar 2002). A combination
of increased air temperature associated with increased evaporation would decrease
groundwater recharge in Quintana Roo, moving the system toward water scarcity as
observed on small karst islands. A perturbation in the fresh water–salt water balance may
take years to re-establish. During this time local inhabitants and tourism operations will
need to invest in alternative methods to obtain fresh water. These methods such as shipping
or piping water from inland areas and desalinization will increase human activities, thereby
creating a negative feedback loop that will continually deplete coastal fresh water
resources. Poor water resources management has caused many health-related problems in
developing countries, and public involvement is essential to avoid such a result (Falkenmark and Widstrand 1992; Dungumaro and Madulu 2003).
Our results indicate a clear and urgent need to understand the role of anthropogenic
effects in saltwater intrusion due to the varied effects in different natural hydrogeological
conditions, which necessitates the local study of anthropogenic effects. In this paper the
authors focused on how human activities alter water quantity, but water quality is equally
important in karst aquifers with high hydraulic conductivity. Moreover, a density-dependent numeric model would more accurately represent this hydrologically complex system,
providing a more refined estimation of saltwater intrusion. Our study demonstrates an
increase in human activities over the next decade, which will lead to an increase in
anthropogenic impacts on the aquifer. The fresh water lens of the aquifer in Quintana Roo
are already heavily contaminated by human sewage as evidenced by high levels of nitrogen
and fecal coliform bacteria (Herna´ndez-Terrones et al. 2011). Anthropogenic nutrient
loading impacts seagrass meadows around offshore spring discharge points by introducing
phosphorous after rain events and increasing nitrogen availability in coastal hotel zones
(Carruthers et al. 2005). A profound understanding regarding the impacts of human

123


Nat Hazards

30% Percent of q0 is fresh water

a 10
ΔT (m)

8
6
4
2
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

70% Percent of q0 is fresh water

b

5

ΔT (m)

4
3
2
1
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

100% Percent of q0 is fresh water

c

3

ΔT (m)

2.5
2
1.5
1
0.5
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

123


Nat Hazards
30% Percent of q0 is fresh water

a

2000

ΔT (m)

1500
1000
500
0
1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

70% Percent of q0 is fresh water

b

800

ΔT (m)

600
400
200
0

1
0.5

Sea Level Rise (m)
0

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)

100% Percent of q0 is fresh water

c

600
500

ΔT (m)

400

300
200
100
0
1
0.5

Sea Level Rise (m)
0

123

0.032

0.034

0.036

0.038

0.04

0.042

Water Recharge (m/year)


Nat Hazards
b Fig. 12 Saltwater intrusion length for scenario F: a 30 % of q0 is fresh water; b 70 % of q0 is fresh water;
c 100 % of q0 is fresh water. In response to the decreasing water recharge and other parameters in scenario

F, the qmin ranged from 5907 to 5207 m2 s-1

activities on saltwater intrusion is indispensable for developing an efficient water use
management given only human activities are manageable.

6 Conclusions
The impact of sea level rise on the saltwater intrusion in karst aquifers was substantially
different between flux-controlled and head-controlled scenarios. Flux-controlled scenarios
reduced the extent of saltwater intrusion substantially, independent of the hydraulic conductivity and composition of discharging groundwater. In Quintana Roo hydraulic conductivity was the most significant factor in determining the extent of saltwater intrusion. As
aquifer permeability ranged from low permeable matrix rock to pipe-flow permeability of
conduits, the extent of saltwater intrusion would not be uniform. Instead, saltwater
intrusion may extend for kilometers within conduits but only hundreds of meters in matrix
rock. Human activities alter the amount of recharge to the aquifer, which can alleviate the
saltwater intrusion, but this effect is minor in comparison with the importance of discharging water composition and flux- versus head-controlled system. Instead, the importance of human activities lies in the amount of aquifer recharge, which controls whether the
aquifer is head or flux controlled. Increasing human activities will lower aquifer recharge,
thereby reducing inland hydraulic head and switching the system from flux to head control.
This switch dramatically increases the modeled extent of saltwater intrusion. Future
integrated modeling of hydraulic conditions and human activities should include water
quality impacts in addition to water quantity. Furthermore, the potential anthropogenic
effects on hydraulic conditions, such as aquifer pumping and injection of wastewater, must
be evaluated in order to fully understand the total impact of human activities on water
resources. These findings are applicable to other karst aquifers worldwide as they have
similar hydraulic conditions, particularly high hydraulic conductivities and conduit flow
connections to the coastal ocean.
Acknowledgments This study is funded by NSF Project 12-594 [Coastal SEES (Track 1): Planning for
hydrological and ecological impacts of sea level rise on sustainability of coastal water resources]. The
authors appreciate the help of Professor Jonathan B Martin, Professor Andrew V Ogram and Professor
Arnoldo Valle-Levinson from University of Florida, Daniel Miret and Dr. Ismael Tapia Marin˜o from the
Center for Research and Advanced Studies of the National Polytechnic Institute and other colleagues in this
project.

References
Abdi H (2010) Partial least squares regression and projection on latent structure regression (PLS regression).
Wiley Interdiscip Rev Comput Stat 2(1):97–106
Alley WM, Reilly TE, Franke OL (1999) Sustainability of ground-water resources, vol 1186. US Department of the Interior, US Geological Survey
Barlow PM, Reichard EG (2010) Saltwater intrusion in coastal regions of North America. Hydrogeol J
18(1):247–260
Bauer-Gottwein P, Gondwe BR, Charvet G, Marı´n LE, Rebolledo-Vieyra M, Merediz-Alonso G (2011)
Review: the Yucata´n Peninsula karst aquifer, Mexico. Hydrogeol J 19(3):507–524

123


Nat Hazards
Beck L, Bernauer T (2011) How will combined changes in water demand and climate affect water availability in the Zambezi river basin? Glob Environ Change 21(3):1061–1072
Bruington AE (1972) Saltwater intrusion into aquifers1. JAWRA J Am Water Resour Assoc 8(1):150–160.
doi:10.1111/j.1752-1688.2012.00659.x
Cai J, Taute T, Schneider M (2015) Recommendations of controlling saltwater intrusion in an inland aquifer
for drinking-water supply at a certain waterworks site in Berlin (Germany). Water Resour Manag
29(7):2221–2232
Carrascal LM, Galva´n I, Gordo O (2009) Partial least squares regression as an alternative to current
regression methods used in ecology. Oikos 118(5):681–690
Carretero S, Rapaglia J, Bokuniewicz H, Kruse E (2013) Impact of sea-level rise on saltwater intrusion
length into the coastal aquifer, Partido de La Costa, Argentina. Cont Shelf Res 61:62–70
Carruthers TJB, Van Tussenbroek BI, Dennison WC (2005) Influence of submarine springs and wastewater
on nutrient dynamics of Caribbean seagrass meadows. Estuar Coast Shelf Sci 64(2):191–199
Chang SW, Clement TP, Simpson MJ, Lee KK (2011) Does sea-level rise have an impact on saltwater
intrusion? Adv Water Resour 34(10):1283–1291
Cheng AD, Halhal D, Naji A, Ouazar D (2000) Pumping optimization in saltwater-intruded coastal aquifers.
Water Resour Res 36(8):2155–2165

Church JA, White NJ (2006) A 20th century acceleration in global sea-level rise. Geophys Res Lett
33:L01602. doi:10.1029/2005GL024826
Church JA, Clark PU, Cazenave A, Gregory JM, Jevrejeva S, Levermann A et al (2013) Sea level change. In
Stocker TF et al (eds) Climate change 2013: the physical science basis. Contribution of working group
I to the fifth assessment report of the intergovernmental panel on climate change. Cambridge
University Press, Cambridge, pp 1137–1216
Comisio´n Nacional Del Agua [National Water Commission of Mexico] (2010) Statistics on water in
Mexico, 2010 Edition. National Water Commission (CONAGUA). Web. 13 Aug 2015
Comisio´n Nacional Del Agua [National Water Commission of Mexico] (2011) Statistics on Water in
Mexico, 2011 Edition. National Water Commission (CONAGUA). Web. 13 Aug 2015
Comisio´n Nacional Del Agua [National Water Commission of Mexico] (2012) Atlas del agua en Me´xico
2012 [Atlas of water in Mexico 2012]. National Water Commission (CONAGUA). Web. 13 Aug 2015
Comisio´n Nacional Del Agua [National Water Commission of Mexico] (2014) Atlas del agua en Me´xico
2013 [Atlas of water in Mexico 2013]. National Water Commission (CONAGUA). Web. 13 Aug 2015
Comisio´n Nacional Del Agua [National Water Commission of Mexico] (2015) Atlas del agua en Me´xico
2014 [Atlas of water in Mexico 2014]. National Water Commission (CONAGUA). Web. 13 Aug 2015
Custodio E, Bruggeman G (1987) Groundwater problems in coastal areas: a contribution to the International
Hydrological Programme. UNESCO, Paris
Dungumaro EW, Madulu NF (2003) Public participation in integrated water resources management: the
case of Tanzania. Physics Chem Earth Parts A/B/C 28(20):1009–1014
El Ayni F, Manoli E, Cherif S, Jrad A, Assimacopoulos D, Trabelsi-Ayadi M (2013) Deterioration of a
Tunisian coastal aquifer due to agricultural activities and possible approaches for better water management. Water Environ J 27(3):348–361
Falkenmark M, Widstrand C (1992) Population and water resources: a delicate balance. Popul Bull
47(3):1–36
Falkland A (1991) Hydrology and water resources of small islands: a practical guide. UNESCO Press, Paris
Feseker T (2007) Numerical studies on saltwater intrusion in a coastal aquifer in northwestern Germany.
Hydrogeol J 15(2):267–279
Germain DJS, Cohen DK, Frederick JJ (2008) A retrospective look at the water resource management
policies in Nassau County, Long Island, New York. J Am Water Resour As 44(5):1337–1346
Gleick PH (1996) Basic water requirements for human activities: meeting basic needs. Water Int

21(2):83–92
Gondwe BRN, Lerer S, Stisen S, Marı´n L, Rebolledo-Vieyra M, Merediz-Alonso G et al (2010) Hydrogeology of the south-eastern Yucatan peninsula: new insights from water level measurements, geochemistry, geophysics and remote sensing. J Hydrol 389(1–2):1–17
Goudie AS (2013) The human impact on the natural environment: past, present, and future. Wiley, New
York
Green NR, MacQuarrie KTB (2014) An evaluation of the relative importance of the effects of climate
change and groundwater extraction on seawater intrusion in coastal aquifers in Atlantic Canada.
Hydrogeol J 22(3):609–623
Guha H, Panday S (2012) Impact of sea level rise on groundwater salinity in a coastal community of South
Florida1. JAWRA J Am Water Resour As 48(3):510–529

123


Nat Hazards
Hallaji K, Yazicigil H (1996) Optimal management of a coastal aquifer in southern Turkey. J Water Resour
Plan Manag 122(4):233–244
Hegerl GC, Zwiers FW, Stott PA, Kharin VV (2004) Detectability of anthropogenic changes in annual
temperature and precipitation extremes. J Clim 17(19):3683–3700
Helland I (2006) Partial least squares regression. Encyclopedia of statistical sciences
Herna´ndez-Terrones L, Rebolledo-Vieyra M, Merino-Ibarra M, Soto M, Le-Cossec A, Monroy-Rı´os E
(2011) Groundwater pollution in a karstic region (NE Yucatan): baseline nutrient content and flux to
coastal ecosystems. Water Air Soil Pollut 218(1–4):517–528
INEGI Information Databank, INEGI (n.d) Instituto Nacional De Estadı´stica Y Geografı´a, Mexico. Web. 12
Oct 2014
International Bottled Water Association (n.d.) Per capita consumption of bottled water worldwide in 2009
and 2014, by leading countries (in gallons). InStatista—The Statistics Portal. Retrieved 26 June 2016,
from />Jie Y, Jun F, Dexiu H, Jianwen Y (2007) Application of partial least-squares regression to safety monitoring
of water conservancy projects. Trans Chin Soc Agric Eng 2007. doi:10.3969/j.issn.1002-6819.2007.3.
028
Jones JAA (2014) Water sustainability: a global perspective. Routledge, London

Liu CCK, Dai JJ (2012) Seawater intrusion and sustainable yield of basal aquifers1. JAWRA J Am Water
Resour Assoc 48:861–870. doi:10.1111/j.1752-1688.2012.00659.x
Mahesha A, Lakshmikant P (2014) Saltwater intrusion in coastal aquifers subjected to freshwater pumping.
J Hydrol Eng 19(2):448–456
Mantoglou A (2003) Pumping management of coastal aquifers using analytical models of saltwater intrusion. Water Resour Res 39(12):1335. doi:10.1029/2002WR001891
Martens HA, Dardenne P (1998) Validation and verification of regression in small data sets. Chemometr
Intell Lab Syst 44(1):99–121
Martin JB, Gulley J, Spellman P (2012) Tidal pumping of water between Bahamian blue holes, aquifers, and
the ocean. J Hydrol 416:28–38
Mehdizadeh SS, Vafaie F, Abolghasemi H (2015) Assessment of sharp-interface approach for saltwater
intrusion prediction in an unconfined coastal aquifer exposed to pumping. Environ Earth Sci
73(12):8345–8355
Mevik BH, Wehrens R (2007) The pls package: principal component and partial least squares regression in
R. J Stat Softw 18(2):1–24
Min SK, Zhang X, Zwiers FW, Hegerl GC (2011) Human contribution to more-intense precipitation
extremes. Nature 470(7334):378–381
Mitchell JFB, Karoly DJ, Hegerl GC, Zwiers FW, Allen MR, Marengo J (2001) Detection of climate change
and attribution of causes. In: Houghton JT et al (eds) Climate change (2001): the scientific basis.
Cambridge University Press, Cambridge, pp 638–695
Monismith SG, Davis KA, Shellenbarger GG, Hench JL, Nidzieko NJ, Santoro AE, Reidenbach MA,
Rosman JH, Holtzman R, Martens CS, Lindquist NL, Southwell MW, Geninf A (2010) Flow effects on
benthic grazing on phytoplankton by a Caribbean reef. Limnol Oceanogr 55(5):1881–1892
Narayan KA, Schleeberger C, Charlesworth PB, Bistrow KL (2003) Effects of groundwater pumping on
saltwater intrusion in the lower Burdekin Delta, North Queensland. In: MODSIM 2003 International
Congress on modelling and simulation, vol 2, pp 212–217
Nicholls RJ, Cazenave A (2010) Sea-level rise and its impact on coastal zones. Science
328(5985):1517–1520
Null KA, Knee KL, Crook ED, de Sieyes NR, Rebolledo-Vieyra M, Herna´ndez-Terrones L, Paytan A (2014)
Composition and fluxes of submarine groundwater along the Caribbean coast of the Yucatan Peninsula.
Cont Shelf Res 77:38–50

Pachauri RK, Allen MR, Barros VR, Broome J, Cramer W, Christ R, et al (2014) Climate change 2014:
synthesis report. Contribution of Working Groups I, II and III to the fifth assessment report of the
Intergovernmental Panel on Climate Change
Peters NE, Meybeck M (2000) Water quality degradation effects on freshwater availability: impacts of
human activities. Water Int 25(2):185–193
Population and Housing Census 2010 (n.d.) Retrieved 8 September 2014, from />contenidos/proyectos/ccpv/cpv2010/Default.aspx
Qi SZ, Qiu QL (2011) Environmental hazard from saltwater intrusion in the Laizhou Gulf, Shandong
Province of China. Nat Hazards 56(3):563–566
Rahmstorf S (2007) A semi-empirical approach to projecting future sea-level rise. Science
315(5810):368–370

123