Space geodetic imaging of rapid ground subsidence in Mexico City

Enrique Cabral-Cano†
Departamento de Geomagnetismo y Exploración, Instituto de Geofísica, Universidad Nacional Autónoma de México, Ciudad Universitaria,
México D.F. 04510, Mexico
Timothy H. Dixon
Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, 4600 Rickenbacker Causeway, Miami, Florida 33149, USA
Fernando Miralles-Wilhelm
Department of Civil and Environmental Engineering, Florida International University, 10555 West Flagler Street, Miami, Florida 33174, USA
Oscar Díaz-Molina
Departamento de Geomagnetismo y Exploración, Instituto de Geofísica, Universidad Nacional 10 Autónoma de México, Ciudad
Universitaria, México D.F. 04510, Mexico
Osvaldo Sánchez-Zamora
Departamento de Sismología, Instituto de Geofísica, Universidad Nacional Autónoma de México, Ciudad Universitaria, México D.F.
04510, Mexico
Richard E. Carande
Neva Ridge Technologies, 4750 Walnut Street, Suite 205, Boulder, Colorado 80301, USA

ABSTRACT INTRODUCTION in the Mexico City metropolitan area due to

Since the late 1950s, several areas of Mex-
ico City have undergone accelerated ground
subsidence and have developed associated
fracturing and faulting. New interferometric
synthetic aperture radar (InSAR) and global
positioning system (GPS) data indicate that
rates of current land subsidence in Mexico
City exceed 350 mm/yr. These rates are close
to historical maximum levels of the mid-twen-
tieth century, when mitigation efforts were
first undertaken to reduce damage to urban
infrastructure. The locus of maximum subsid-
ence has shifted from its historical location in
the old city center to the east. Correlation of
our InSAR results with seismically mapped
stratigraphic units suggests that subsidence is
primarily controlled by compaction of Quater-
nary lacustrine clays and silts. We also evalu-
ate spatial gradients in subsidence and suggest
that this, rather than subsidence magnitude, is
the key factor in risk assessment. Subsidence
represents a major geologic risk for Mexico
City and imposes serious constraints to any
further urban development.
Keywords: subsidence, interferometry, GPS,
SAR, Mexico Basin.
†
E-mail: ecabral@geofisica.unam.mx.
Many of Earth’s urban and suburban areas
are subsiding due to excess withdrawal of flu-
ids, principally water, but also petroleum, natu-
ral gas, and geothermal fluids (Poland, 1984).
While most subsidence rates are relatively low
(<10 mm/yr) and local (<100 km2), much higher
rates over larger areas are possible, increasing
the risk of flooding, damage to infrastructure
from differential subsidence, and damage to the
fluid reservoirs by overpumping and permanent
porosity loss.
Since the late 1950s, several areas of Mexico
City have undergone accelerated ground subsi-
dence and associated shallow fracturing and
faulting. These faults have mainly developed on
the piedmont and talus deposits of older Qua-
ternary volcanoes and other volcanic structures
and have continuously damaged housing, utility
works, and other urban infrastructure. The inte-
grated economic damages of this process are
large, rivaling those of a strong earthquake, but
they have received less attention because of the
longer time frame. The economic consequences
of subsidence, while large, are generally fac-
tored into yearly maintenance budgets rather
than accounted for as unique natural disasters
at a single point in time. As these integrated
costs grow, it becomes increasingly important
to assess the extent and magnitude of damage
ground subsidence. Monitoring of the spatial
and temporal patterns of surface deformation
associated with fluid withdrawal is an important
first step, and it is the focus of this paper.
No current single technique gives complete
temporal and spatial sampling of subsidence.
Here, we describe the recent subsidence of
Mexico City due to groundwater withdrawal
using a combination of interferometric synthetic
aperture radar (InSAR) for high spatial resolu-
tion and global positioning system (GPS) data
for improved temporal information and cali-
bration. We use a remote-sensing approach to
define regions where large differential subsid-
ence results in large strain gradients, which thus
require closer monitoring.
GEOLOGIC AND HYDROLOGIC
BACKGROUND
The southern portion of the Basin of Mexico
(Fig. 1) includes a low-relief lacustrine plain,
formerly covered by shallow water bodies and
wetlands, commonly referred to as the Val-
ley of Mexico. This area at present has several
small lakes, including Texcoco, Zumpango, and
Chalco; the latter was completely drained at the
turn of the twentieth century. These lakes, along
with the Xochimilco canal system, are remnants
of a large water body that encompassed about

GSA Bulletin; November/December 2008; v. 120; no. 11/12; p. 1556–1566; doi: 10.1130/B26001.1; 10 figures.

For permission to copy, contact editing@geosociety.org

1556
© 2008 Geological Society of America
 Space geodetic imaging of rapid ground subsidence in Mexico City
one-fourth of the total surface of the basin sev-
eral thousand years ago.
The Mexico City metropolitan area, located
in the southern section of the Mexico Basin, is
a heavily populated urban area with ~17 million
inhabitants (INEGI, 2000). Originally named
Tenochtitlán, the capital of the Aztec empire, it
was built over the former Lake Texcoco, parts of
which survive east of the Mexico City metropol-
itan area, in a high (2200 m elevation), closed
basin ringed by mountains that can exceed 5000
m elevation (Fig. 1) and that provide natural
recharge of basin groundwater (Ortega and
Farvolden, 1989). The unusual location poses
technical challenges for hydraulic management.
Flooding in the sixteenth and seventeenth cen-
turies led to artificial opening of the basin and
construction of other hydraulic works in the
late 1700s to divert flood water. Since then, a
major hydraulic management network has been
built and periodically upgraded, maintaining the
flood-control function but also drastically reduc-
ing natural groundwater recharge.
Mexico Basin stratigraphy is well described
(Schlaepfer, 1968; Mooser, 1975; Vázquez-
Sánchez and Jaimes-Palomera, 1989). GODF
(2004) presented the most recent geotechnical
classification of the main surface and near-sur-
face units: a hard rock unit, a transitional unit,
and a lacustrine unit. The hard rock unit (Unit I
in Fig. 2) corresponds to the slopes of surround-
ing mountain ranges and includes basaltic lava
flows and pumiceous tuffs and sandy/silty beds
with a high percentage of gravel. The transi-
tional unit (Unit II, Fig. 2) is a slope deposit; it
represents the transition between the lacustrine
beds and rock outcrops. It consists of progres-
sively thicker sedimentary deposits overlying
the uppermost clay-rich lacustrine beds with
interbedded lacustrine and alluvial deposits. The
lacustrine unit (Unit III, Fig. 2) includes depos-
its from former Lake Texcoco, mainly soft and
compressible silts and clays with relatively low
permeability. A large percentage of the modern
city is built over these beds, reflecting the his-
tory of urban development since the Spanish
conquest. Unit III ranges up to 80 m in thick-
ness and overlies coarser, more permeable beds
that comprise the main aquifer, mainly alluvial
sands and gravels, as well as Pleistocene-Recent
volcanic rocks in the depth range 100–400 m.
The Mexico City metropolitan area con-
sumes over 65 m3/s of water (JACMCW, 1995),
and more than 70% of it comes from the aqui-
fer beneath the city through a system of more
than 380 water wells. The larger basin has more
than 630 wells. In a typical year, consumption
exceeds recharge, lowering the water table by
0.1–1.5 m/yr, reducing pore fluid pressure in
the aquifer and overlying aquitard, and leading
Geological Society of America Bulletin, November/December 2008
to compaction of lacustrine shales and surface
subsidence. Drilling for groundwater started in
the 1850s. Subsidence was eventually recog-
nized as a serious problem (Gayol, 1925), but
the link between groundwater extraction and
clay compaction was only recognized later
(Carrillo, 1948). By 1952, total subsidence
(1891–1952) had reached 6.0 m in the down-
town area (CHCVM, 1953). More recent sur-
veys show up to 2.5 m of additional subsidence
between 1952 and 1973. Other studies show an
average subsidence rate of 90 mm/yr for the
20 yr period 1965–1985 in the downtown area
(CAVM, 1975; Figueroa-Vega, 1984; Ortega et
al., 1993). The decrease in subsidence rates after
1965 reflects conservation measures instituted
in the 1950s and 1960s, which included capping
wells near the city center.
Consequences of the subsidence process are
costly. Water sewage works must be constantly
upgraded due to loss of gradient, and transi-
tional areas between lacustrine beds and slope
deposits are prone to severe differential subsid-
ence, damaging housing and urban infrastruc-
ture. However, the regional extent and spatial
variation of subsidence, and seasonal and lon-
ger-term variations, are not well monitored or
understood, hampering effective mitigation.
DATA PROCESSING
InSAR has been used to study a variety of
surface deformation processes, including sub-
sidence from groundwater withdrawal, and the
technique is well described (e.g., Massonnet et
al., 1997; Galloway et al., 1998, 1999; Galloway
and Hoffman, 2006; Fielding et al., 1998; Ame-
lung et al., 1999). We used Synthetic Aperture
Radar data from the European Remote Sensing
Satellite (ERS) 1 and 2 (pre-2001) and from the
Advanced Synthetic Aperture Radar (ASAR)
onboard the Environment Satellite (ENVISAT;
2003 and later). ERS-1/2 data collected prior to
2001 was used, but many interferometric pairs
yielded poor coherence, in some cases, due to
the long time span between passes. Best results
were obtained with image pairs spanning rela-
tively short time spans. The following discus-
sion is based on SAR images acquired in 1996
(1 and 2 February and 16 May), 1999, 2000 (7
January and 17 March), 2003 (10 October and 31
December), and 2005 (15 April and 24 June).
Topography data from the Shuttle Radar
Topography Mission (SRTM) was used for the
topographic correction. We assumed a constant
rate of surface change to make a first-order cor-
rection for this effect and used a phase unwrap-
ping algorithm to convert ambiguous fractional-
phase measurements to continuous phase
corresponding to range change (Goldstein et al.,
1988; Ghiglia and Pritt, 1998). In the interfero-
grams (Fig. 2), one color cycle represents 28 mm
of range change (one half the SAR wavelength)
in the line of sight direction between the satel-
lite and ground (23° from vertical in the case of
ERS1/2 and 15°–45° for ENVISAT_ASAR).
Although this range change is usually interpreted
as vertical motion when considering fluid with-
drawal, reservoir contraction may induce hori-
zontal motions as well. If the motion is purely
vertical, 28 mm of range change corresponds to
a true vertical motion of 30.4 mm. We then reg-
istered the SAR interferogram to high-resolution
optical image data from the Advanced Space-
borne Thermal Emission and Reflection Radi-
ometer (ASTER) to improve georeferencing
and facilitate feature matching, for example, the
location of InSAR fringes with respect to water-
well locations or major street intersections.
GPS analysis and error estimation proce-
dures follow Dixon et al. (2000) and Sella et al.
(2002). Permanent station UIGF (Ciudad Uni-
versitaria) on the southwestern margin of the
Mexico City metropolitan area has been occu-
pied since 1997. Station AIBJ (Mexico City
International Airport) was occupied for a total
of 10 twenty-four hour sessions at the end of
the dry season (May–June) between 1995 and
2001. A permanent GPS station located on the
center of the historic downtown was installed
April 2004, and more recent permanent GPS
stations (MRRA, MPAA, and MOCS) have
continuously recorded data since early 2005 in
order to monitor subsidence with high temporal
resolution. We fit a weighted least-squares line
to the GPS position data for each site to derive
the average velocity and uncertainty over the
entire observation period, and also considered
subsidence over shorter intervals.
GPS data also allow calibration of InSAR
measured subsidence. For example, the unad-
justed 1996 interferogram agrees well with both
GPS sites, implying minimal orbit error and/or
atmospheric delay effects in this data set. On
the other hand, the 1999–2000 interferogram
predicts subsidence at UIGF 55 mm below that
indicated by the GPS analysis at this site. The
InSAR-derived profile in Figure 3 is adjusted
based on the UIGF data, allowing an InSAR-
based estimate of subsidence within the study
area relative to UIGF.
SUBSIDENCE ANALYSIS
The InSAR data (Fig. 2) suggest significant
range change across most of the Mexico City
metropolitan area in the 1996, 1999–2000,
2003, and 2005 data sets. However, assuming
these changes represent real surface displace-
ment, do they indicate purely vertical motion, or
1557
 Cabral-Cano et al.

20°00′ N

30 N

20

-110 -100 -90

Sierra de Guadalupe

Figure 1. Location map and

shaded digital elevation model
of study area in central Mexico.
Rectangle shows the coverage of
Sierra de Las Cruces Mexico City images in Figure 2.
Sierra Sta. Catarina

Sierra Nevada

0 8 16
Popocatpetl volcano
4 12 20 km
19°00′
99°40′ 98°20″
5400 0m

19°15′ N
N
Lake
Texcoco N 1996 N 2000
Tlalnepantla Unit I

Unit II Unit III

Naucalpan MRRA
AIBJ
Downtown Airport

UPEC MOCS MPAA

Cuajimalpa
Nezahualcóyotl

Unit I
Unit II
UIGF Unit I

0 4 8 0 4 8 0 4 8
2 6 10
0 km 2 6 10 km 2 6 10 km
19°35′
99°15′ W 99°00′

Figure 2. Left: Study area shaded digital elevation model with geotechnical subsoil classification (white lines; GODF, 2004). Red and yellow
lines show the leveling transects described in the text and Figure 5. Global positioning system (GPS) sites referenced in the text are shown as
blue triangles. Center: Interferometric synthetic aperture radar (InSAR) fringe maps of Mexico City metropolitan area for 1 February to 16
May 1996. Right: InSAR fringe maps of Mexico City metropolitan area for 16 July 1999 to 7 January 2000. Each color cycle phase represents
28 mm distance change between sensor and ground. The digital elevation and fringe images have been merged and registered with a high-
resolution ASTER band 2 image.

1558 Geological Society of America Bulletin, November/December 2008

 Space geodetic imaging of rapid ground subsidence in Mexico City

0.05 sides at an average annual rate of 115 mm/year,

while the current GPS rate is −84 mm/yr. These
0 rates are less than the historical maximum at this
Elevation change (m)

location (Fig. 4), consistent with the capping of

UIGF wells in the 1950s.
-0.05 Figure 4 shows a multitechnique compos-
ite plot of historical subsidence in the down-
-0.1 town area. Pre-1985 leveling data (Mazari and
InSAR Jul 1999–Mar 2000 (175 days) AIBJ Alberro, 1991) were collected at selected city
InSAR Feb–Mar 1996 (105 days) landmarks. The 1985–2002 data were derived
-0.15 GPS sites Jul 1999–May 2000 from leveling of a modern benchmark network
GPS sites Feb–May 1996 encompassing most of the city, surveyed at
-0.2 ~2 yr intervals by the former Dirección General
0 5 10 15 20 de Construcción y Operación Hidráulica (1993;
Distance (km) now Sistema de Aguas de la Ciudad de Mexico).
Although benchmarks used in Figure 4 are not
Figure 3. Elevation change versus distance (profile UIGF to AIBJ, see Fig. 2 for the same on the pre-1985 and post-1985 surveys,
location) for global positioning system (GPS) (triangles) and interferometric the closest modern benchmark to the location of
synthetic aperture radar (InSAR) (lines) for time periods of 1996 and 1999–2000 the historical landmarks was selected, typically
SAR interferograms. GPS displacement was calculated assuming constant rate within just a few hundred meters distance. Con-
for period 1995–2001, interpolated to time span of interferogram. The 1996 tinuous GPS data (2004–2007) from the current
interferogram is unadjusted for orbit error; the 1999–2000 interferogram shown permanent network (map on Fig. 2) are also dis-
is adjusted to match GPS data at UIGF near the southwest edge of basin. played in Figure 4.
Total subsidence of the downtown area
(Alameda park) between the end of the nine-
a mix of vertical and horizontal motion? A SAR estimate at AIBJ is 100 mm over 105 d (1996) teenth century, when artesian flow from the
interferogram from one look direction, as in our and 148 mm over 175 d (1999–2000), equivalent local water springs ceased, and spanning the
case, measures only the scalar length change in to average annual rates of 347 mm/yr (1996) and interval between the times first water wells
the satellite line of site direction and does not 309 mm/yr (2000), compared to the 291 mm/yr were drilled in the basin and present time is
resolve the three orthogonal components of the derived from the GPS campaigns. shown in Figure 4. Between 1895 and 2002, a
displacement vector. Also, the imaged changes The agreement between the different data total of 9.7 m subsidence occurred. The rates
may include short-term (e.g., seasonal) changes, sets with different time spans suggests several of subsidence since 1985 show sharp differ-
or could reflect longer-term trends. The avail- important points: ences with other areas east of downtown. Sub-
able SAR data do not address the issue of high- (1) Most of the InSAR recorded ground sidence rates of ~−57 mm/yr (1960–1985) and
resolution temporal variability. motion is vertical. Independent analysis of the −112 mm/yr (1985–1992) are comparable with
We recorded negligible GPS vertical veloc- horizontal component GPS data at MRRA, the current rate of −84 mm/yr measured by
ity at UIGF, outside the subsidence affected MPAA, UPEC, and MOCS confirms that hori- GPS techniques at UPEC.
zone, and high and constant subsidence at zontal motion at these sites is small. Subsidence rates of other eastern sites such as
AIBJ, at an average rate of −291 mm/yr for the (2) Short-term subsidence rates measured by Airport SW (benchmark M[S01E03]05), which
1995–2001 period using campaign data. More SAR in 1996 and 1999–2000 are similar to the lies close to GPS site AIBJ, show similar rates to
recently installed GPS sites at MOCS, MPAA, average rate for the period 1995–2001, the time the GPS-derived rates of subsidence: −215 mm/
and MRRA (Fig. 4) located on the high-sub- span of campaign GPS measurements at AIBJ. yr at Airport SW (leveling) compared to −211
sidence region show rates that range from −168 (3) It confirms that extraordinarily high rates mm/yr, −255 mm/yr, and −287 mm/yr for the
to −255 mm/yr and display linear trends with of surface subsidence are occurring within the GPS sites MPAA, MRRA, and AIBJ, respec-
no or very little annual variation. On the other Mexico City metropolitan area. tively. The daily sampled GPS data, and general
hand, the UPEC site, located further to the west The general agreement of the various subsid- agreement with rates derived from less frequent
(Fig. 4) where lacustrine sediments are thinner, ence estimates for different time intervals and leveling data, suggest essentially constant sub-
shows a lower subsidence rate (−84 mm/yr) but independent techniques (Fig. 4) indicates that sidence with little seasonal fluctuations.
displays small seasonal variations. the short-term SAR-based measured subsidence We also compared several techniques to
Figure 3 plots InSAR displacement, assuming process does not have a significant seasonal better characterize the post-1985 subsidence
only vertical motion, and the vertical component bias. The eastern part of Mexico City has been period. We used space geodetic data and com-
of GPS displacement for a transect across the subsiding at a fast and essentially constant rate pared them to the two main north-south leveling
basin that intersects UIGF and AIBJ GPS sites, for at least the past 10 yr. While short-term GPS- transects that run across the city (red and yel-
assuming that the longer-term average veloc- based rates (better shown on UPEC site; Fig. 4) low lines in Fig. 2). These are tied to reference
ity at the GPS sites (representing data spanning indicate fluctuations spanning a few week’s benchmarks on rock outcrops and are assumed
more than 6 yr at AIBJ) is representative of the period, these short-term fluctuations are small to be devoid of regional subsidence effects. A
average velocity over the 175 d (1999–2000) or compared to the long-term signal and probably temporal comparison of these leveling transects
105 d (1996) InSAR period. The three subsi- account for the small discrepancies between the (Fig. 5, top panels) indicates that except for a
dence estimates at AIBJ (two InSAR, one GPS) InSAR and GPS based estimates. For example, few benchmarks that exhibit anomalous behav-
agree fairly well. The InSAR-based subsidence InSAR data indicate that the old city center sub- ior, the rate remains essentially constant over

Geological Society of America Bulletin, November/December 2008 1559

 Cabral-Cano et al.

Mexico City Historical Subsidence GPS Vertical component

0 -100
Cathedral
-2 Palacio Minería -300

(mm)
Carlos IV
Relative subsidence (m)

-4 Monument -500
Downtown MRRA site rate: -255.8+/- 4.6 mm/yr

(M[S01E01]01) 0
2005 2006 2007

-6
Alameda Park
−200

-8 0

(mm)
−400
Airport NE (AIBJ)
-10 -2 −600 MPAA site rate: −211.8+/− 2.1 mm/yr
2005 2006 2007 2008
Airport SW (M[S01E03]05)
-12 -4 0

−100
-14 -6 (m)

(mm)
1900 1920 1940 1960 1980 2000 −200

Years UPEC site rate: −84.1+/−

r 6.2 mm/yr
2004 2005 2006 2007 2008
0

19°26′15″ N −100
Alameda UPEC
✪ ✪ ✪
✪
−200
(mm)
Carlos IV
P. Minería
✪ −300
Catedral −400 MOCS site rate: −168.6+/− 2.6 mm/yr
M[S01E01]01 2005 2006 2007 2008

✪
0 1 km
N 19°25′18″
99°09′08″ W 99°07′30″
Figure 4. Left: Multitechnique composite plot of the subsidence in the Mexico City downtown area since 1895. Pre-1985 leveling data were
collected at selected city landmarks, whereas 1985–2002 data were derived from leveling of a modern benchmark network. See text for further
explanation. Right: Vertical component time series for global positioning system (GPS) sites (red triangles) within the high subsidence region;
see Figure 2 for their location. Map inset shows location of leveling benchmarks (blue circles) and GPS site UPEC (red circle).

time. We then constructed a relative subsidence
plot following the same locations of benchmark
transects using InSAR-derived subsidence mag-
nitude maps (Fig. 5, bottom panels). The most
relevant observations from this comparison are
as follows:
(1) InSAR-based relative subsidence transects
show ~8× better spatial resolution compared to
leveling, and they are capable of resolving ver-
tical motion for areas less than 100 × 100 m,
well within average city block dimensions in a
medium to high population density zone.
(2) Each technique has a characteristic time
interval that needs to be considered in the inter-
pretation, especially if subsidence has a time-
varying rate, e.g., seasonal fluctuations. The
InSAR-derived transects represent an integrated
measurement over the time span of the interfero-
metric pair. The continuous GPS data give daily
measurements. Most of the continuous GPS
sites show more or less continuous subsidence
at an essentially constant rate; hence, the InSAR
and GPS rates may be usefully compared, even
if they were acquired at different times. The
conventional leveling transects are carried out
every 2 yr over a 2–4 wk period during normal
working hours; the methodology includes a ref-
erence benchmark of known (static?) elevation
and assumes a static reference frame during the
period of the survey. Therefore, any leveling sur-
vey that is performed over high subsidence rate
areas, such as the eastern part of Mexico City,
with rates over 250 mm/yr, may be biased by up
to ~9.5 mm in a typical 3 wk survey (differential
subsidence between benchmarks at beginning
and end of survey).
(3) Correspondence between the leveling and
InSAR rates is better displayed on the eastern
transect (Fig. 5, right middle and bottom pan-
els) than on the western transect (Fig. 5, left
middle and bottom panels). This is evidenced
by the overall correspondence in magnitude and
location of high and low values of both eastern
leveling and InSAR plots (Fig. 5, right middle
and bottom panels). This may be a consequence
of the magnitude of the subsidence rate, which
is higher on the western transect, and is located
mostly over lacustrine clays (Unit III on Fig. 2),
than on the eastern transect, located along the
transitional zone (Unit II on Fig. 2).
Current maximum subsidence for the Mex-
ico City metropolitan area (Figs. 6 and 7) is
localized at Ciudad Nezahualcóyotl (on the
eastern side of the Mexico City metropolitan

1560 Geological Society of America Bulletin, November/December 2008

 Space geodetic imaging of rapid ground subsidence in Mexico City

Subsidence Rate Western Transect Subsidence Rate Eastern Transect

500 100

0
300
-100

(mm/yr)
(mm/yr)

Rate 1998–2000
100 Rate 1998–2000 Rate 2000–2002
0 Rate 2000–2002 Rate 1998–2002
Rate 1998–2002 -300
-100

-300
-500
0 10.0 20.0 30.0
-500
Benchmark Distance (km)

Subsidence Rate Western Transect (1998–2002) Subsidence Rate Eastern Transect (1998–2002)

20 50

0
(mm/yr)

(mm/yr)
0 -50

-40 -150

-80 -250
Benchmark Benchmark

Susidence Rate InSAR Western Transect Subsidence Rate InSAR Eastern Transect
150
Subsidence rate (mm/yr)

Subsidence rate (mm/yr)

50

50 0

0 -50

-50

-150

-150
Pixel Number
-250
Pixel Number
Figure 5. Comparison of two north-south–trending leveling transects (top and middle panels) and the corresponding interferometric synthetic
aperture radar (InSAR)–derived relative subsidence (bottom panels) along the same transects (location shown in Fig. 2). InSAR-based relative
subsidence transects show ~8× better spatial resolution compared to leveling. Subsidence rate magnitude is higher on the western transect,
which is located mostly over lacustrine clays (Unit III on Fig. 2), than on the eastern transect, which is over coarser-grain alluvial-fan deposits
(Unit II on Fig. 2). See text for further details.

area), southeast of the historical maximum sub-
sidence area. This area registered an average
annual rate of 378 mm/yr, close to the highest
annual subsidence rate in the downtown area
recorded in the mid-twentieth century (~400
mm/yr; see Fig. 4). This shift is important and
suggests that water extraction has not declined,
but rather moved eastward. Compaction may
now be affecting deeper units near the center
of the basin.
Groundwater overdraft in the Iztapalapa-
Nezahualcóyotl region is acute; there has been
a sustained static piezometric level drop of
−1.4 m/yr averaged over the past 20 yr (Lesser
y Asociados, 2003; Ortega, 1999). The region
where current subsidence rates exceed a few
mm/yr corresponds closely to the lacustrine
unit (Unit III on Fig. 2). In contrast, the western
part of the city, mostly built over alluvial-fan
deposits and/or volcanic tephra, tuffs, and lava
flows (Unit I), shows negligible motion. The
high subsidence region corresponds closely to
the boundary of old Lake Texcoco just prior to
Spanish settlement, when a major change in
agricultural practices and hydraulic manage-
ment was initiated. The outer InSAR fringe in
Figure 2 (2000 data panel) maps the location
of the lake boundary at the time of Spanish
conquest of the Aztec empire, when Mexico
City was built over the ruins of old Tenochti-
tlán, and the lake was drained to build the new
city, irrevocably changing the ecosystem and
hydrologic balance.
Further evidence supporting the strong cor-
relation between subsidence and thickness of
clay-rich units is shown in Figure 8, where
we superimpose the seismically derived depth
of the Quaternary lacustrine clay unit (Perez-
Cruz, 1988) and the subsidence magnitude
estimated for 2000. This clay unit is thickest in
the high subsidence region east of the Mexico
City metropolitan area. This unit is 350 m thick
in the Texcoco-1 deep well (Fig. 8) and can be
seismically correlated under most of the Ciu-
dad Nezahualcóyotl neighborhood.

Geological Society of America Bulletin, November/December 2008 1561

 Cabral-Cano et al.

1996 2000 2003

N N N

0 4 8 0 4 8 0 4 8
2 6 10 km 19°15′ 2 6 10 km 2 6 10 km
99°15′ W 99°00′
0 Subsidence rate -400 mm/yr

Figure 6. Examples of annual interferometric synthetic aperture radar (InSAR)–derived subsidence maps for Mexico City for 1996, 2000,
and 2003. White line shows the Distrito Federal political boundary.

2263 19°26′45″ N
2038 2624
AIBJ
1978 2380 N mm/yr
160
1903
-374
1258 514 Airport
1911 2666

689 165 1439

2
380
27
196 28
190 Ciudad Nezahualcoyotl
2681
Viaducto 44 38 942
14 45
13 2106 -187
rubusco

1757 Ca
2679 2221 lza
199
160 da
105 Za
128 rag
Tlalpan

267 oz
Rio Chu

1078 7 8 4 a
6 9
170 107
5
193 TEC-2 110
138
198
Div.

185
2259 162 2239 150 0
del N

53 243
119 244
192 50 151 123-126 0 2 4
1857 164 59
orte

167 152
2075 61
2161 2294 1 3 5 km 19°21′20″
99°09′45″ W 99°00′30″

Figure 7. Location of pilot wells used in the analytical subsidence calculation superimposed onto the 2003 interferometric syn-
thetic aperture radar (InSAR) subsidence map. Major streets are show as white lines.

1562 Geological Society of America Bulletin, November/December 2008

 Space geodetic imaging of rapid ground subsidence in Mexico City

19°30′07″ N
9
14

0
24
149
Texcoco

342
Figure 8. Seismically derived depth
Roma (black contours) of the Quaternary
lacustrine clay unit from Perez-
Cruz (1988) superimposed onto the
Mixhuca interferometric synthetic aperture
radar (InSAR)–measured subsid-
ence magnitude for 2000. Stars show
location of PEMEX deep wells.
240
149

149
14

Copilco 24
9

0
342 Deep stratigraphic borehole
Seismic depth contour (m)
Municipal boundary
24

Tulyehualco
0

454 34
0 2 4 2
0 Subsidence rate -400 mm/yr

1 3 5 km

19°14′00″
99°11′51″ W 98°59′00″

SUBSIDENCE GRADIENT
We computed the horizontal gradients of
subsidence rate from the subsidence maps to
investigate possible correlation with damage to
infrastructure. Figure 9 shows the magnitude of
maximum horizontal gradient, computed from
the October–December 2003 ENVISAT-ASAR
image pair. While there are minor differences
between gradients computed from the various
interferograms, all show four regions of large
horizontal subsidence gradient:
(1) southern slopes of the Sierra de Guadal-
upe, north of Mexico City;
(2) Peñón de Los Baños, immediately north
of Mexico City International Airport;
(3) the Zaragoza corridor, which has a
NW-SE feature running parallel to this major
avenue between the Agricola Oriental and Aca-
titla neighborhoods, including the Peñón del
Marqués area, on the eastern part of the Mexico
City metropolitan area; and
(4) a NE-SW corridor located immediately to
the SE of Canal de Garay Avenue and into the
Santa Cruz Meyehualco neighborhood, north of
Calzada Ermita Iztapalapa.
All of these regions coincide with Quaternary
volcanic features in close proximity to lacus-
trine clay-rich sediments. These high-gradient
zones mark the location of abrupt transitions
between continuous subsidence of the lacus-
trine beds and stable volcanic outcrops (Fig. 2).
These areas are known for extensive damage to
housing and large civil engineering structures
such as subways and large hydraulic infrastruc-
ture. Detailed information on the location and
extent of these zones from the InSAR-derived
gradient maps provides a new and valuable tool
to include in urban land use and mitigation of
subsidence hazard.
ANALYSIS OF SUBSIDENCE DATA
The consolidation analysis (Terzaghi and
Peck, 1967) establishes a relationship between
those changes in effective stress caused by
extraction pumping in an aquifer and the result-
ing deformation of its porous matrix, as follows:

Geological Society of America Bulletin, November/December 2008 1563

 Cabral-Cano et al.

19°30′00″ N

0.04

0.0
0 4 8

2 6 10 km 19°14′00″

19°17′00″ W 99°00′00″
Figure 9. Horizontal subsidence gradient for the Mexico City metropolitan area calculated from the 2003 subsidence magnitude
map. High gradient (nondimensional) values depict areas where structural damage risk to housing and other civil engineering
structures is higher due to intense surface fracture and faulting. These areas coincide with transitional piedmont zones between
Cenozoic volcanic structures and clay-rich Quaternary lacustrine deposits.

dV
= α γ dh , (1)
V the land subsidence (b0 – b), where b0 and h0 are
where V is the aquifer porous matrix bulk vol-
ume, α is the porous media compressibility, γ is
the specific weight of water, and h is the piezo-
metric head (groundwater table elevation) in
the aquifer. Assuming that deformation of the
porous matrix occurs predominantly in the verti-
cal direction and solving for the thickness of the
aquifer as a function of the change in piezometric
head, this equation can be rewritten in terms of
the reference (datum) conditions, i.e., b(h0) = b0.
This equation assumes that the surface responds
instantly to changes in piezometric head.
b0 − b
= 1 − exp [ − α γ (h0 − h) ]. (2)
b0
Figure 10 compares the InSAR-measured
subsidence values to those analytically derived
subsidence values following Equation 2 at those
water-well locations shown in Figure 7, using
compressibility (α) values that correspond to
clay, silt, and sand soils (Freeze and Cherry,
1979). We used a value for the specific weight
of water of γ = 9800 N/m3 and a reference aqui-
fer thickness of b0 = 80 m (Ortega et al., 1993).
This analysis suggests that the land subsidence
observed in the vicinity of these wells can be
represented with soil parameters that correspond
to a spatial composite of silt and clay. The offset

1564 Geological Society of America Bulletin, November/December 2008

 Space geodetic imaging of rapid ground subsidence in Mexico City
0.2
α = 10–6 Pa–1 (clay)
α = 10–7 Pa–1 (silt)
Analytical subsidence (m)
0.15
α = 10–8 Pa–1 (sand)
0.1
0.05
0 ence and aquifer-system deformation: Geology, v. 27,
0 0.05 0.1 0.15 0.2
Observed InSAR subsidence (m)
Figure 10. Plot of interferometric synthetic aperture radar (InSAR)–mea-
sured land subsidence versus analytically derived subsidence values based
on Terzaghi and Peck (1967) for water wells shown on Figure 7. We applied
compressibility values that correspond to clay, silt, and sand soils (Freeze and
Cherry, 1979), a specific weight value of water of γ = 9800 N/m3, and a refer-
ence aquifer thickness of b0 = 80 m (Ortega et al., 1993).
between observed and modeled values (e.g., 5 data also suggest that subsidence is not primar-
cm offset between zero calculated subsidence ily seasonal. Long-term, inelastic compaction
and 5 cm observed subsidence) may reflect the and shrinkage of the lacustrine unit are continu-
influence of fine-grained clays, which retard the ing at high rates, close to maximum historical
migration of water and consequent attainment values. This impermeable layer currently limits
of equilibrium. Hence, areas with no current surface pollutants from reaching the underlying
change in water levels still experience subsi- aquifer. However, as this layer shrinks, fractur-
dence due to past overdraft. Furthermore, the ing and faulting may occur, enabling pollutants
estimated order of magnitude of aquifer com- to percolate down and contaminate the underly-
pressibility compares favorably with the histori- ing aquifer (Rudolph et al., 1991).
cal observations of land subsidence reported by Economic assessment of damage to urban
Ortega et al. (1993) for a total land subsidence of infrastructure due to subsidence will benefit from
7.5 m in the old downtown area of Mexico City detailed mapping of the horizontal gradient pre-
during the period 1940–1985, and to the 9.7 m sented here. Continued monitoring of the spatial
(this work) for the period 1940–2007. and temporal patterns of surface deformation
within the Mexico City metropolitan area by the
DISCUSSION AND CONCLUSIONS techniques outlined here can lead to the imple-
mentation of stronger mitigation actions, which
The spatial correspondence between subsi- are necessary to preserve the aquifer beneath the
dence and the location of young lacustrine beds Mexico City metropolitan area.
identified here, combined with the high and
ACKNOWLEDGMENTS
essentially constant subsidence rate, implies that
subsidence is due mainly to pressure loss in the
This work was funded by the Office of Naval
shallow aquitard (clay-rich lake sediments) asso- Research (ONR), the National Aeronautics and
ciated with groundwater overdraft (withdrawal in Space Administration (NASA), Universidad Nacio-
excess of recharge). This poses important impli- nal Autónoma de Mexico (UNAM) Projects Papiit
IN-121515 and IN-114907, and Geofisica-Cardi.
cations for water management in the Mexico European Remote Sensing satellites (ERS)-1, 2, and
City metropolitan area because compaction of Envisat data were provided by the European Space
Agency (ESA) Projects AO-3 441 and CAT-1 1409.
clay-rich aquitards is often associated with per-
manent loss of porosity and reservoir capacity NASA’s Earth Observing System provided Advanced
Spaceborne Thermal Emission and Reflection Radi-
(Holzer, 1984). Our data suggest that mitigation ometer (ASTER) imagery through the Unites States
activities have not had a significant effect on the Geological Survey Earth Resources Observation and
long-term compaction of the lacustrine beds. Our Science Data Center Land Processes Distributed
Geological Society of America Bulletin, November/December 2008
Active Archive Center (EDC-DAAC). We thank
Francisco Correa-Mora, Gerardo Cifuentes-Nava,
Esteban Hernández-Quintero, Teodoro Hernandez-
Treviño, and Mario Mártinez-Yáñez for field sup-
port, and F. Amelung, D. Galloway, T. Holzer, D.
Eaton, and other anonymous reviewers for their com-
ments, which improved this paper. This paper is pub-
lication 12 from the Center for Southeastern Tropical
Advanced Remote Sensing (CSTARS).
REFERENCES CITED
Amelung, F., Galloway, D.L., Bell, J.W., Zebker, H.A., and
Laczniak, R.J., 1999, Sensing the ups and downs of Las
Vegas: InSAR reveals structural control of land subsid-
p. 483–486, doi: 10.1130/0091-7613(1999)027<0483:
STUADO>2.3.CO;2.
Carrillo, N., 1948, Influence of artesian wells on the sinking
of México City, in Proceedings of the Second Inter-
national Conference on Soils Mechanics: Rotterdam,
Holland, International Society for Soil Mechanics and
Geotechnical Engineering, vol. VII.
Comisión de Aguas del Valle de México (CAVM), 1975,
Boletín de Mecánica de Suelos, v. 7, México City.
Comisión Hidrológica de la Cuenca del Valle de Mexico
(CHCVM), 1953, Boletín de Mecánica de Suelos, v. 1,
México City.
DGCOH (Dirección General de Construcción y Operación
Hidráulica), 1993, Fichas de nivelación de banco de
nivel en la Cd. de México y área metropolitana con
información de croquis de localización, fotografías,
estadísticas de nivelación y curvas de hundimiento:
Mexico City, Dirección General de Construcción y
Operación Hidráulica Technical Report, 410 p.
Dixon, T.H., Miller, M., Farina, F., Wang, H., and Johnson,
D., 2000, Present-day motion of the Sierra Nevada
block and some tectonic implications for the Basin and
Range Province: North American Cordillera: Tecton-
ics, v. 19, p. 1–24, doi: 10.1029/1998TC001088.
Fielding, E.J., Blom, R.G., and Goldstein, R.M., 1998,
Rapid subsidence over oil fields measured by SAR
interferometry: Geophysical Research Letters, v. 25,
p. 3215, doi: 10.1029/98GL52260.
Figueroa-Vega, F., 1984, Case history no. 9.8, Mexico, in
Poland, J.F., ed., Guidebook to Studies of Land Subsid-
ence due to Ground-Water Withdrawal: Paris, France,
United Nations Educational Scientific and Cultural
Organization, p. 217–232.
Freeze, R.A., and Cherry, J.C., 1979, Groundwater: Engle-
wood Cliffs, New Jersey, Prentice Hall, 604 p.
Galloway, D.L., and Hoffman, J., 2006, The application of
satellite differential SAR interferometry-derived ground
displacements in hydrogeology: Hydrology Journal,
v. 15, p. 133–154, doi: 10.1007/s10040-00-60121-5.
Galloway, D.L., Hudnut, K.W., Ingebritsen, S.E., Phillips,
S.P., Peltzer, G., Rogez, F., and Rosen, P.A., 1998,
InSAR detection of aquifer system compaction and
land subsidence, Antelope Valley, Mojave Desert, Cali-
fornia: Water Resources Research, v. 34, p. 2573–2585,
doi: 10.1029/98WR01285.
Galloway, D.L., Jones, D.R., and Ingebritsen, S.E., 1999,
Land Subsidence in the United States: U.S. Geological
Survey Circular 1182, 177 p.
Gayol, R., 1925, Estudio de las perturbaciones que en el
fondo de la Ciudad de México ha producido el drenaje
de las aguas del subsuelo, por las obras del desague
y rectificación de los errores a que ha dado lugar una
incorrecta interpretación de los efectos producidos:
Revista Mexicana de Ingeniería y Arquitectura, v. III,
p. 96–132.
Ghiglia, D.C., and Pritt, M.C., 1998, Two Dimensional Phase
Unwrapping: New York, John Wiley and Sons, 103 p.
Gobierno del Distrito Federal, 2004, Normas técnicas
complementarias para diseño y construcción de cimen-
taciones: Gaceta Oficial del Distrito Federal, v. II,
103-BIS, p. 11–39.
Goldstein, R.M., Zebker, H.A., and Werner, C.L., 1988,
Satellite radar interferometry: Two dimensional phase
unwrapping: Radio Science, v. 23, p. 713, doi: 10.1029/
RS023i004p00713.
1565

Holzer, T.L., 1984, Ground failure induced by ground-water
withdrawal from unconsolidated sediment: Reviews in
Engineering Geology, v. 6, p. 67–105.
Instituto Nacional de Estadistica Geografia e Informatica
(INEGI), 2000, Cuaderno Estadístico de la Zona
Metropolitana de la Ciudad de México: México City,
México, INEGI, 254 p.
Joint Academies Committee on The Mexico City Water Sup-
ply, 1995, Mexico City’s Water Supply Improving the
Outlook for Sustainability. Washington, D.C., National
Academy Press, 256 p.
Lesser y Asociados, 2003. Piezometria del Valle de Mexico,
Año 2003, Sistema de Aguas de la Ciudad de Mexico,
Technical Report, 21 p.
Massonnet, D., Holzer, T., and Vadon, H., 1997, Land sub-
sidence caused by the East Mesa geothermal field,
California, observed using SAR interferometry: Geo-
physical Research Letters, v. 24, p. 901–904, doi:
10.1029/97GL00817.
Mazari, M., and Alberro, J., 1991, Hundimiento de la Ciu-
dad de México, in Mazari, M., and Kumate, J., eds.,
Los Problemas de la Cuenca de México: Mexico City,
Mexico, El Colegio Nacional, p. 83–114.
Mooser, F., 1975, Historia geológica de la Cuenca de Méx-
ico, in Memoria de las Obras del Sistema de Drenaje
1566 Geological Society of America Bulletin, November/December 2008
Cabral-Cano et al.
Profundo del Distrito Federal: Mexico, DF, Departa-
mento del Distrito Federal, Tome 1, p. 7–38.
Ortega, A., 1999, Estudio Geologico, Hidrogeológico y
Geotécnico del Fracturamiento en la Unidad Habita-
cional Ejército de Oriente, Sección Peñon, Delegación
Iztapalapa: México, D.F., Instituto de Geología, Uni-
versidad Nacional Autónoma de México Technical
Report, 50 p.
Ortega, A., and Farvolden, R.N., 1989, Computer analysis
of regional groundwater flow and boundary conditions
in the Basin of Mexico: Journal of Hydrology, v. 110,
p. 271–294, doi: 10.1016/0022-1694(89)90192-3.
Ortega, A., Cherry, J.A., and Rudolph, D.L., 1993, Large-
scale aquitard consolidation near Mexico City: Ground
Water, v. 31, p. 708–718, doi: 10.1111/j.1745-6584.1993.
tb00841.x.
Perez-Cruz, G.A., 1988, Estudio Sismológico de Reflex-
ión del Subsuelo de la Ciudad de Mexico [M.I. the-
sis]: México Distrito Federal, Universidad Nacional
Autónoma de México, 62 p.
Poland, J.F., ed., 1984, Guidebook to Studies of Land Sub-
sidence due to Ground-Water Withdrawal: UNESCO
Studies and Reports in Hydrology, Volume 40: Paris,
France, United Nations Educational, Scientific and
Cultural Organization, 305 p., 5 appendixes.
Rudolph, D.L., Cherry, J.A., and Farvolden, R.N., 1991,
Groundwater flow and solute transport in fractured
lacustrine clay near Mexico City: Water Resources
Research, v. 27, p. 2187–2201.
Schlaepfer, J. C, 1968, Resumen de la Geología de la Hoja
México, Distrito Federal y Estados de México y More-
los. Hoja México 14 Q-h (5): México, D.F., Instituto de
Geologia, Universidad Nacional Autónoma de México,
scale 1:250,000, 1 map with text.
Sella, G.F., Dixon, T.H., and Mao, A., 2002, REVEL: A
model for recent plate velocities from space geodesy:
Journal of Geophysical Research, v. 107, no. B4, 2081,
doi: 10.1029/2000JB000033.
Terzaghi, K., and Peck, R.B., 1967, Soil Mechanics in Engi-
neering Practices: New York, Wiley, 752 p.
Vázquez-Sánchez, E., and Jaimes-Palomera, R., 1989,
Geología de la Cuenca de México: Geofísica Interna-
cional, v. 28, p. 133–190.
MANUSCRIPT RECEIVED 9 MARCH 2006
REVISED MANUSCRIPT RECEIVED 22 OCTOBER 2007
MANUSCRIPT ACCEPTED 2 DECEMBER 2007
Printed in the USA