Review Paper/

Potential for Satellite Remote Sensing of

Ground Water
by Matthew W. Becker

Abstract
Predicting hydrologic behavior at regional scales requires heterogeneous data that are often prohibitively
expensive to acquire on the ground. As a result, satellite-based remote sensing has become a powerful tool for sur-
face hydrology. Subsurface hydrology has yet to realize the benefits of remote sensing, even though surface ex-
pressions of ground water can be monitored from space. Remotely sensed indicators of ground water may provide
important data where practical alternatives are not available. The potential for remote sensing of ground water is
explored here in the context of active and planned satellite-based sensors. Satellite technology is reviewed with
respect to its ability to measure ground water potential, storage, and fluxes. It is argued here that satellite data can
be used if ancillary analysis is used to infer ground water behavior from surface expressions. Remotely sensed
data are most useful where they are combined with numerical modeling, geographic information systems, and
ground-based information.

Introduction source for ~50% of the drinking water in the United States
Ground water is the last component of the hydro- (Solley et al. 1998). Ground water pollution, once consid-
logic cycle to realize the benefits of remote sensing. ered a local issue (e.g., leaking fuel tanks and landfills), is
Ground water scientists have been late to embrace satel- increasingly being thought of as a regional phenomenon.
lite data for an obvious reason: ground water lies in the This is partially due to the overwhelming number of local
subsurface, and current air- and satellite-based radar problems that add up to create a regional problem. It has
and radiometers can normally penetrate only a few cen- been estimated that the United States as a whole will spend
timeters into the ground. In spite of this apparent road- 0.5 to 1 trillion dollars over the next 30 years cleaning up
block, remote sensing holds tremendous potential for 300,000 to 400,000 sites where ground water may be con-
regional ground water flow studies. Previous, current, taminated (National Research Council 1994).
and future remote sensing efforts that are applicable The move toward a big-picture view of ground water
to shallow ground water systems are reviewed in this is also necessitated by the monitoring and management
article. Ways in which remote sensing can be more of regional, continental, and global cycling of com-
effectively used for future ground water studies are pounds, both toxic and benign. The U.S. EPA is promot-
suggested. ing the use of ecosystem-scale multimedia (air, surface
The application of satellite data to ground water prob- water, ground water) transport and exposure models (U.S.
lems would support the general scientific trend toward EPA 2003) and has required that source water protection
a ‘‘big-picture’’ view of ground water issues. Ground water plans be developed at the State scale (1996 amendments
accounts for 26% of global renewable fresh water resour- to Section 1453 of the Safe Drinking Water Act). Concern
ces (Food and Agriculture Organization 2003) and is the over global warming has dramatically raised interest in
global water cycling. Although ground water accounts for
only 2% of the global water volume, it accounts for 80%
University at Buffalo, State University of New York, Buffalo, of the residence time of water in the global hydrologic
NY 14260; mwbecker@[Link]
Received December 2004, accepted March 2005. cycle (Chow et al. 1988). Any study of global cycling of
Copyright ª 2005 National Ground Water Association. carbon or other dissolved matter must ultimately take into
doi: 10.1111/j.1745-6584.2005.00123.x account the residence time in ground water.
306 Vol. 44, No. 2—GROUND WATER—March–April 2006 (pages 306–318)
 A key to the remote sensing of ground water is the previous reviews (Meijerink 1996; Waters et al. 1990), this
recognition that shallow ground water flow is usually article will focus on direct or implicit remote sensing of
driven by surface forcing and parameterized by geologic ~ and fluxes, q~.
ground water hydraulic potential, I,
properties that can be inferred from surface data. Tóth This review is presented from the perspective of
(1963) conceptualized ground water flow systems at mul- a ground water modeler. In a numerical ground water
tiple scales that depend upon topographic driving forces model, surface water is typically treated as boundary con-
(Figure 1). Ground water recharged in regionally maxi- ditions on the subsurface flow equations, which are based
mum elevations will tend to flow more deeply than upon Darcy’s law (Equation 1). We recognize mathemati-
ground water recharged in locally maximum elevations. cal boundary conditions that are of type I (hydraulic
Tóth’s topographically driven flow model suggests, there- head), type II (flux or discharge), or type III (mixed,
fore, that remote sensing will be more effective at pre- both head and discharge) (Anderson and Woessner
dicting local ground water flows, as it provides more 1992). A straightforward way to couple remote sensing
opportunity for detecting ground water. with ground water flow predictions, therefore, is to use
While the driving force of ground water flow is typi- remotely sensed imagery to define boundary conditions
cally surface recharge and discharge, the rate and behav- such as streams, lakes, wetlands, seepage areas, recharge
ior of flow is dictated by geology. The relationship zones, or evapotranspiration zones. As will be shown, it
between ground water flow rate, hydraulic driving forces, may also be possible to sense ‘‘monitoring’’ data such as
and geology is expressed by Darcy’s law: ground water heads. In keeping with the concepts of
ground water modeling, this review of remote sensing ap-
q~ ¼ K~~
I~ ð1Þ plications in ground water will be organized into the sens-
ing of hydraulic potential (heads) and hydraulic fluxes (or
where q~ is the specific discharge vector representing flux discharges).
of ground water (flow per unit area), K~~ is the hydraulic
conductivity tensor, which is a function of geology, and I~ Remote Sensing of Hydraulic Potential
is the hydraulic gradient, which is a function of surface
Hydraulic head is usually treated as the most impor-
forcing. There is a long history of the use of remote sens-
tant information for predicting the occurrence and move-
ing for inference of K~~ through geologic mapping. Geo- ment of ground water, because it provides an indication
logic maps derived from remotely sensed data have been of both ground water storage and hydraulic gradient. Visi-
used extensively for ground water prospecting (Waters ble, microwave, and gravity sensors intended for other
et al. 1990). Geologic maps provide information about the purposes may potentially be used to measure ground
hydraulic conductivity and water reserves of a water- water head.
bearing formation. A particularly successful application
of remote sensing to ground water has been the identifica- Surface Water Elevations
tion of lineaments that are thought to be related to faults From a hydrogeologic perspective, surface water rep-
and fracturing in hard-rock environments (e.g., Mabee and resents an outcropping of ground water, that is, a surface
Hardcastle 1997; Moore et al. 2002; Salama et al. 1994). manifestation of ground water (Winter et al. 1998).
As the application of geologic mapping to ground water Unless there is a severe change in hydraulic conductivity
prospecting has been well documented in the literature and at a stream or lake bed, surface water stage approximates

Figure 1. Tóth’s conceptual model of topographically driven ground water flow systems (after Fetter 2001).

M.W. Becker GROUND WATER 44, no. 2: 306–318 307

ground water head at the ground water/surface water water storage was limited primarily by the accuracy to
interface. The elevation at which the water table intersects which unsaturated water contents (soil moisture) could be
sloping topography generally coincides with the head estimated. The High-Plains Aquifer system is generally
water of first-ordered streams (de Vries 1995, 1994). At unconfined so that ground water storage changes would
a basin scale, therefore, remotely sensed positioning of imply changes in hydraulic head, as well. The utility of
stream head water can provide a dynamic monitoring of gravity surveys is clearly limited by the very coarse spa-
the water table. tial resolution of the measurements. Gravity-measuring
Surface water elevations are measured routinely in satellites present a trade-off between spatial resolution
the generation of topographic digital line graphs (DLGs) and measured mass accuracy. To offset observational er-
and digital elevation models (DEMs). Most high- rors and local atmospheric variability, accurate estimates
resolution DLGs and DEMs for land surfaces are created of mass can only be obtained for regions that are several
from stereographic aerial photos. During this process, hundred kilometers or more in scale (Swenson et al.
open water surfaces (i.e., lakes, seas, oceans) are forced 2003). Thus, estimating water storage changes in very
to be perfectly flat by manually corrected elevation to large aquifer systems is currently feasible. Estimating
ancillary stage data (USGS 1992). The measurement date ground water potential gradients is not currently feasible,
and accuracy of surface water elevations in a DEM are, as most aquifers are not continuous over the scale of hun-
therefore, difficult to establish. dreds of kilometers.
Remote altimetry provides the potential for a signifi- Plans for a follow-on mission to GRACE may result
cant improvement over published data sources. Satellite- in hydrologic measurements of greater relevance to typi-
based altimetry and interferometry have been used for the cal large aquifer systems (Watkins 2004). Rodell and
measurement of stage in large lakes and wetland systems. Famiglietti (2002) anticipate that error in ground water
Koblinsky et al. (1993) used the Geosat altimeter to mea- storage estimates by the current GRACE mission will
sure river stage in the Amazon Basin with an accuracy of increase rapidly in regions less than ~300,000 km2. Due
~70 cm. Birkett (1998; Birkett et al. 2002) used data from to technological improvements, a follow-on mission
the TOPEX/Poseidon satellite to measure stage in large might have a greater sensitivity that may allow accurate
lakes and rivers, with vertical accuracies of 11 to 60 cm. measurements over areas as small as 1000 to 20,000 km2.
Current satellite altimetry is constrained by its limited For reference, the USGS Principal Aquifers Map (Miller
horizontal resolution, however. TOPEX/Poseidon, for ex- 1998) lists extensive aquifer systems in the United States
ample, cannot measure stage in rivers that are narrower that produce >10 gallons per minute (Figure 6). There are
than ~1 km. Planned and potential satellite missions may ~3000 of these principal aquifers that cover 4.6 million
improve our ability to measure stage with better horizon- km2 of the conterminous United States. Only three of
tal and vertical resolution. Radar interferometry has been these aquifers have an area extent >300,000 km2, but 35
proposed as a tool for measuring stream stages in rivers as have an area >20,000 km2. The current GRACE mission
narrow as 30 m (Alsdorf et al. 2003). Laser altimetry has has limited applicability for monitoring aquifers in the
a much smaller ground footprint and vertical accuracy United States, but the follow-on mission, if launched,
than radar altimetry. The footprint of the GLAS laser on might be a practical method of monitoring water storage
the ICESAT satellite has a vertical resolution of 15 cm in U.S. aquifers.
and a footprint of ~70 m. Laser altimetry is limited for
some applications, such as flood monitoring, as the lasers Heat Capacity
used cannot penetrate most cloud cover. Saturated soils have a greater heat capacity than dry
soils, suggesting that remote thermal sensing might be
Water Column Mass used to estimate depth to water table. Cartwright (1968,
High-resolution gravitational surveys have been used 1970, 1974, 1979) was among the earliest to advocate the
to estimate water storage in the subsurface (Pool and use of ground temperature measurement to locate shallow
Eychaner 1995) and could potentially be used to estimate ground water. Soil temperature measurements showed
hydraulic head in aquifers. Gravitational instruments have that water tables within the depth of annual soil tempera-
no vertical resolving power, however, so that measure- ture variation constitute a heat sink in the summer and
ment of saturated water mass requires removing the influ- a heat source in the winter. At about the same time, re-
ence of all other sources of mass, including water stored searchers began to find evidence of shallow ground water
in the unsaturated zone (soils). For aerial gravity surveys, on thermal infrared imagery. Chase (1969) noted that cool
water stored in vegetation must be isolated. For satellite areas on thermal infrared imagery seemed to correspond
gravity surveys, water in the atmosphere and vegetation with shallow ground water. Heilman and Moore (1982)
must be isolated. used data from the Heat Capacity Mapping Mission satel-
The NASA Gravity Recovery and Climate Experi- lite to examine shallow ground water in this same region.
ment (GRACE) satellite has been cited as a potential tool The authors found that daytime thermal was not useful
for obtaining data regarding changes in ground water for predicting depth to water table, but found good agree-
storage. Rodell and Famiglietti (2002) used computer ment between cool areas in nighttime thermal images and
simulations and anticipated GRACE performance param- areas of known shallow ground water (Figure 2). A statis-
eters to estimate that ground water storage changes as tical correlation between radiometric temperature and
small as 8.7 mm could be measured in the U.S. High- water table depth was only found after correcting for veg-
Plains Aquifer system. Accuracy in estimates of ground etation effects (emissivity). Huntley (1978) used heat
308 M.W. Becker GROUND WATER 44, no. 2: 306–318
 Figure 3. December 7, 1972, Landsat 1 mosaic depicting
resistance to snow cover (darker areas) attributed to shallow
ground water in valley train deposits between Lake Simcoe
and Lake Ontario (Falconer et al. 1979).

Figure 2. August 29, 1978, nighttime HCMM Thermal

Infrared Image of Big Sioux Basin. Cooler areas (dark) are
provided sufficient heat to partially melt snow cover,
apparently related to shallow ground water (Heilman and thereby reducing its reflectance and emissivity. Areas of
Moore 1982). shallow water table derived from snow reflectance and
radiation correlated well with areas of lush vegetation, as
indicated by Band 5-7 scatter diagrams.
These studies demonstrate that ground water can
transport equations to determine that shallow ground influence surface temperatures, but it is not clear under
water has a minimal (<0.5
C) impact on diurnal tempera- what conditions these temperature influences can be
ture changes. Heat transport from the subsurface was remotely sensed. Where shallow ground water impacts
determined to be insignificant compared to the thermal the rate of snow melt, visible to thermal infrared imagery
inertia of the surface soils. The applicability of nighttime appears effective in discerning shallow water tables
thermal sensing of depth to water table has yet to be criti- (Bobba et al. 1992; Falconer et al. 1979). The timing of
cally evaluated. imagery used for this purpose is critical, however. Images
Snowpack provides an opportunity for amplifying cannot be collected after a fresh snowfall or too late in the
the heat signature of shallow ground water through sensi- season when most of the snowpack has already melted.
ble heat change in the snowpack or through the heat of Otherwise, the utility of daytime thermal imagery seems
fusion as snow is melted. Falconer et al. (1979) noted that
a December 7, 1972, composite Landsat image showed
a lack of ice snow and ice cover over valley train deposits
near Lake Erie, south of Lake Simcoe. The valley train
deposits are visible as a darker color on the image shown
as Figure 3. Falconer et al. (1979) surmised that shallow
ground water delayed snow and ice cover in lowlands rel-
ative to the surrounding highlands. No supporting infor-
mation such as water table measurements was provided
with this analysis, however.
Bobba et al. (1992) used a March 20, 1974, Landsat 1
image to detect shallow ground water in the Big Creek
and Big Otter Creek in Southern Ontario, ~150 km west
of Niagara Falls. They compared measurements of histor-
ical water table depth to Band 7 (near infrared) apparent
radiance in a snow-covered scene. Depth of water table
was linearly related to snow radiance up to a depth of ~4 m
(Figure 4). Images from a predawn airborne Daedalus
thermal infrared (8 to 12 lm) scanner taken in February
1976 showed similar agreement between radiance and
Figure 4. Water table depth related to Landsat Band 7 radi-
known areas of shallow water table (Figure 5). Bobba ance after Bobba et al. (1992).
et al. (1992) hypothesized that the shallow water table
M.W. Becker GROUND WATER 44, no. 2: 306–318 309
 humid regions microclimate may have a greater impact
than surface temperatures, as cool moist air settles into
low topography. If near-surface temperature drops below
the dew point, a layer of mist or fog may form in the early
hours that can obscure surface temperatures. The forma-
tion of dew on vegetation can also obscure surface tem-
peratures. In general, there has been little research into
the remote measurement of surface temperatures in pre-
dawn hours.

Land Subsidence
Ground water storage changes are related to land
subsidence phenomena. As an aquifer is depleted or re-
charged, the effective stress on the pore skeleton changes,
which may lead to significant inflation or compaction in
unconsolidated sediments, expressed as a surface eleva-
tion change. Subtle changes in land elevation may be
accurately measured using interferometric synthetic aper-
ture radar (InSAR). InSAR analysis involves the imaging
of the same point from two viewing angles at different
times. A sub-wavelength variation in the target distance
Figure 5. Warm areas in thermal image correspond to results in a detectable change in the phase of the reflected
known ground water discharge areas in Southern Ontario signal, from which elevation change can be calculated
after Bobba et al. (1992). Warmer areas are lighter in color (Galloway et al. 1998). InSAR measurements are affected
than cooler areas. by topographic effects and water vapor content in the
atmosphere. Accuracy of elevation measurements may be
on the order of 10 cm in very humid regions to on the
doubtful because the subtle effect of ground water will order of 1 mm in very dry regions (Galloway et al. 1998).
likely be masked by differential solar heating (Falconer Consequently, InSAR has been used most successfully to
et al. 1979; Huntley 1978). Predawn imagery seems to measure ground water storage changes in semiarid re-
hold promise for correlation of temperature to water table gions such as Southern California (Galloway et al. 1998;
depth. Pitfalls are abundant here as well, however. In Lu and Danskin 2001; Watson et al. 2002; Hoffmann

Figure 6. Area extent of principal aquifers of the United States, adapted from Miller (Miller 1998). Water storage in aquifers
>300,000 km2 may be monitored by the current GRACE mission, and those >20,000 km2 may be monitored by a follow-on
mission.

310 M.W. Becker GROUND WATER 44, no. 2: 306–318

et al. 2003; Schmidt and Burgmann 2003) and Southern sensors may be used to delineate shallow water tables by
Nevada (Hoffmann et al. 2001). vegetation stress or the proliferation of water-resistant
As pore pressures are decreased by ground water plant species (Bacchus et al. 2003). Active microwave
withdrawal, unconsolidated sediments may compact elas- (radar) sensors have been used to delineate inundated
tically or inelastically. Water level changes and land sub- areas for the purpose of flood prediction and monitoring
sidence are expected to be linearly related by the (Smith 1997; Townsend and Foster 2002; Mertes 2002).
compressibility (storativity) in elastic porous medium Jackson (2002) recently reviewed the application of
(Domenico and Schwartz 1998). Inelastic compression is microwave sensing for the estimation of ground water
caused by the rearrangement of grains and may result in recharge. Both passive and active microwave sensors can
apparent storativity hundreds of times greater than typical be used for this purpose. Passive sensors are hindered by
elastic storativity in similar media (Galloway et al. 1998). the necessity of footprints on the order of tens of kilo-
As a result, determination of storativity from surface meters but have been shown to provide quantitative mea-
deformation requires the use of ancillary data such as surements of depth to water table (Reutov and Shutko
geodetic controls from global positioning systems, strain 1992). Surface soil moisture contents can reflect the
measurements from extensiometers, water level measure- depth to water table, but one must be careful to assure
ments, and hydrologic fluxes to the study region. In aqui- that the soil profile is continuous. Surface drying may
fer systems that are known to behave within their elastic decouple surface and subsurface water contents (Jackson
limits, storativity may be obtained in a more straightfor- 2002). A NASA soil moisture sensing satellite to be
ward manner. For example, the ratio of land surface to called ‘‘Hydros’’ is in the formulation phase and is sched-
ground water–level rise in the Santa Ana River area of the uled for launch in 2010 (Entekhabi et al. 2004). This sat-
San Bernadino ground water basin was 0.0006 to 0.0008 ellite will combine active (L-band radar) and passive
over the 1992 to 1993 period, which corresponded to (radiometer) microwave radiation to estimate the mois-
a typical compressibility for similar alluvium (Lu and ture content of the shallow (~5 cm) soil. Under conditions
Danskin 2001). Where extensive pumping tests have been where the shallow soil surface is coupled with the subsoil,
conducted, it may be possible to obtain accurate measure- this soil may be used to indicate areas of near-inundation
ments of storativity from leaky confined aquifers with the by ground water.
aid of accurate land subsidence data (Burbey 2003). The application of remote measurement of soil mois-
The utility of InSAR is increased by its integration ture to ground water has been hampered by a lack of
with numerical models of ground water withdrawals at communication between soil scientists and hydrogeolo-
the basin scale (Galloway et al. 1998; Lu and Danskin gists. This schism is illustrated by the fact that hundreds
2001; Hoffmann et al. 2003; Watson et al. 2002). of soil moisture samples were collected to depths of 1 m
Hoffman et al. (2003) combined ground water models and during coordinated ground-space Soil Moisture Experi-
InSAR data, for example, to estimate spatially varying ment campaigns ([Link] but depth
compaction inelastic storage coefficients for the Mojave to water table was not monitored. Given the obvious link-
Desert, California. ERS (European Remote Sensing Satel- age between surface inundation and ground water drain-
lite) data from 1993 to 1995 were used to construct the age, greater effort should be made toward coupling soil
interferogram, and the parameter estimation program, water and ground water studies.
UCODE (Poeter and Hill 1998), was used to determine
the storage coefficients. The strength of InSAR is that
it offers greater spatial extent and resolution of land Remote Sensing of Ground Water Fluxes
subsidence than can be obtained through ground-based
Topographically driven ground water flow implies
measurements (Smith 2002). For example, ERS interfer-
that ground water will be recharged over broad upland
ometery was used to observe time-varying rates of sub-
areas and discharged at relatively focused lowlands as
sidence due to ground water withdrawals near Las Vegas,
surface water. In more arid climates, evapotranspiration
Nevada. The interferograms revealed that the spatial pat-
may cause broad areas of net ground water discharge.
tern of subsidence was influenced by Quaternary faults.
The contrast in spatial resolution of the ground water/sur-
In another study, InSAR data were used to look for rela-
face water and ground water/land surface interfaces sug-
tionships between ground water–induced subsidence and
gests that different remote sensing methods may be more
earthquakes in Southern California (Galloway et al.
effective at each interface. As a consequence, ground
1998). As more InSAR data become available, it is likely
water exchange with surface water and land surface will
that they will find further application for determination of
be considered separately here.
aquifer storage parameters in large basins.

Soil Moisture Ground Water/Surface Water

Remote sensing has been used to measure soil mois- Ground water and surface water are closely con-
ture for the purposes of predicting land/atmospheric nected in both arid and humid environments, but the
water exchange and predicting agricultural performance nature of that connection differs. In arid environments,
(Jackson 2002). Remote sensing techniques are available streams and lakes can be separated from the water table
for locating moistened or waterlogged soils that may indi- by a large vadose zone. During the wet season, ground
cate the presence of a shallow water table. Where plant water may become perched below surface water and then
roots becomes waterlogged, visible and near-infrared dissipate during the dry season. Because vegetation is
M.W. Becker GROUND WATER 44, no. 2: 306–318 311
typically water deprived in arid environments, an out- up by vegetation. All of these indicators of ground water
cropping of ground water is usually identifiable in fluxes can be sensed remotely; the difficulty lies in ob-
remotely sensed images through vegetation. The great taining some useful quantitative information from the
difficulty in arid environments, however, is that much of presence of ground water flow. Quantitative information
the ground water flow that appears at the surface is inter- lies in the spatial distribution of flows and the rate of
mittent. Time series ground data or multiple remote sens- flows. While the former has been obtained by remote
ing images (e.g., multiple Landsat scenes) are required to sensing for many years (e.g., land cover maps, flood
distinguish steady ground water behavior from storm- inundation maps, estuary discharge), the latter is still the
driven or seasonally limited behavior. subject of active research. Ground water generally dis-
In humid environments, shallow ground water sys- charges to surface water via (1) springs and seeps; (2)
tems are more prevalent and the surface water is likely an streambeds; or (3) lakes. The remote sensing technique
expression of the water table. Surface features are a more best applied to these discharge avenues depends upon the
reliable indicator of ground water conditions in these en- spatial scale and the indicator parameter one wishes to
vironments, provided they can be correctly measured and sense.
interpreted. Figure 7 provides a rough classification of
the interaction between ground water and surface water in
the continental United States. This map was created for Springs and Seeps
this review by averaging base flow index (BFI, the frac- Springs occur when ground water is confined by
tion of streamflow attributed to base flow) for each of the a low-permeability geologic unit or structure, which con-
physiographic sections of the United States (Fenneman strains water to a focused discharge point. Fractures in
and Johnson 1946) based upon data from the USGS for hard rock and confining clay units are typical conduits
14,177 streamgages (Wolock 2003). Figure 7 illustrates for spring flow (Fetter 2001). Springs may discharge at
the intimate connection between ground water and sur- the land surface or below surface water. Moore (1982)
face water and the importance of considering both in argued that the best technique for detecting springs
regional water assessments. In those regions where the remotely is infrared thermal analysis. Better results are
majority of streamflow is derived from ground water expected when the temperature of the discharging ground
(BFI > 0.5), it seems feasible that remote sensing of sur- water and the land surface temperature, or surface water
face water could provide important information about temperature, are in contrast. Shallow ground water tem-
ground water flow. peratures in the conterminous United States range from
Water discharging to the surface carries heat energy 7
C to 27
C and usually approach the yearly average tem-
and dissolved chemicals that may leave a sensible signa- perature for the region (Moore 1982). Thus, during sum-
ture. Water may pond at the surface, run off, or be taken mer and winter months the potential for discriminating

Figure 7. Average BFI (fraction of streamflow attributed to ground water. Map created from USGS data [Wolock 2003]) and
physiographic sections of the continental United States (Fenneman and Johnson 1946).

312 M.W. Becker GROUND WATER 44, no. 2: 306–318

springs and seeps from thermal data is greatest. In one estimations of stream discharge, or change in stream dis-
study, airborne thermal infrared sensing was used to charge, in the absence of a ground-measured rating curve
locate mine drainage to springs and seeps (Sams and (Goteti et al. 2004).
Veloski 2003). A single nighttime survey was performed The most successful application of remote sensing to
in March, when the ground water was significantly ground water–stream interactions has been in characteriz-
warmer than the surface water or land surface. Wood ing the spatial distribution of ground water discharge.
(1972) also used a predawn airborne thermal image to Ground water discharge to streams is expected to be dis-
locate streambank springs in the Lehigh River. tributed over the entire streambed, with higher flux near
Another indicator of springs and seeps is chemical the banks. Even in small streams, ground water discharge
alteration of the land surface. Ground water tends to be may be very heterogeneous. Meter-scale (Conant 2004)
saturated with dissolved solids. Upon contact with the and kilometer-scale (Becker et al. 2004; Silliman and
atmosphere, mineral precipitation can occur, leaving Booth 1993) measurements of streambed temperature
a chemical staining. Mineral deposition and staining is indicate that ground water inflow to streams can vary
a common and obvious indicator of ground water seepage widely in magnitude and even be reversed in stream sys-
from hard rocks and has been used to locate geothermal tems. In mountainous terrain, ground water–stream water
springs (Crowley and Hook 1996). Mineral loadings to interactions can be even more variable and are expected
the land surface, surface water, or the root zone may also to create a strong thermal signature (Constantz 1998).
lead to changes in vegetation species that can be detected.
The problem with chemical staining and vegetation in-
dicators, as well as thermal indicators, is that the impact Discharge to Lakes, Estuaries, and Oceans
is usually spatially limited. This implies that sensors must Quantifying ground water discharge to standing bod-
have resolution on the scale of meters rather than tens of ies of water is more challenging than quantifying dis-
meters to effectively locate a single spring or seep. If the charge to streams because water balance is not so easily
result is diffused at the exit point via dilution by surface obtained. Even performing a water balance on a small
water or spreading over the ground surface, courser reso- lake can be delicate because of the difficulty in constrain-
lution may be sufficient. Likewise, if the thermal contrast ing all surface water inlets and outlets, as well as direct
between the land surface and ground water is sufficient, precipitation and evaporation (Winter et al. 1998, 1989).
an elevated temperature may still be noticeable after the Remote sensing is useful, however, for quantifying the
entire thermal signature is averaged over a larger pixel spatial distribution of ground water discharge to lakes,
size (Gillespie 1992). estuaries, and oceans.
Where ground water enters a lake, for example, it
may produce a thermal, chemical, or vegetation signature
Base Flow to Streams that can be sensed remotely. Ground water discharge is
That component of streamflow which is independent not uniformly distributed in a lake. It is greatest near the
of short-term storm events, is termed base flow. It is usu- shoreline and decreases approximately exponentially
ally assumed that base flow is primarily supplied by away from the shoreline (Winter 1999). Rundquist et al.
ground water (Winter et al. 1998). To the extent that base (1985) detected ground water flow through lakes in the
flow can be estimated from remote sensing, therefore, the Nebraska Sand Hills using thermal remote sensing. Air-
flux of ground water exiting an aquifer can be estimated. borne Thermal Infrared Multispectral Scanner (TIMS)
The component of streamflow attributed to base flow is images collected in August showed ~2
C cooler temper-
generally determined by removing short-term (storm) atures in portions of the lake where ground water was
events from a stream hydrograph. Unless ancillary data thought to be entering. Similar airborne techniques have
are available, measuring total streamflow at any one point been used to study ground water discharging to estuaries.
in time can lead to a large error in the estimation of Banks et al. (1996) used March predawn and noontime
ground water discharge. TIMS data to locate focused and diffused ground water
At present, there is no remote sensing technology discharge to creeks on a peninsula in Chesapeake Bay.
that can quantify stream discharge directly. The NASA Portnoy et al. (1998) used airborne thermal imagery to
Surface Water Working Group proposes a satellite mis- characterize both concentrated and diffused ground water
sion that would use radar interferometry to estimate stage discharge into an estuary on Cape Cod (Figure 8).
and flow in rivers (Alsdorf et al. 2003). Measuring It is important to note that buoyancy can strongly
streamflow from space presents serious technical chal- influence the capacity of thermal sensors to detect ground
lenges in remote regions, however. The current practice is water discharge. Because fresh water is relatively buoy-
to convert stage measurements to stream discharge hydro- ant compared to the saline estuary water, thermal signa-
graphs using a ground-measured ‘‘stream rating curve.’’ If tures of ground water discharge are easier to detect in
a ground-measured rating curve is not available for a par- estuaries. At fresh water–fresh water interfaces, relatively
ticular stream, stream discharge must be estimated via cold water will not be detected immediately at the surface
theoretical relationships, which require the estimation of (Moore 1996). Based upon these encouraging airborne re-
either channel geometry or surface velocity (Bjerklie et sults, remote sensing may prove to be a valuable tool in
al. 2003). Virtual missions are currently being conducted understanding this flux, given the potential to detect
to determine whether these parameters can be measured ground water discharge via changes in temperature,
from space with sufficient accuracy to allow reasonable chemistry, and salinity.
M.W. Becker GROUND WATER 44, no. 2: 306–318 313
Figure 8. Aerial thermal image of Tone Cover, Nauset Marsh, Cape Cod, Massachusetts. Relatively cold ground water is visi-
ble as dark streaks emanating from the shore (after Portnoy et al. 1998).

Ground Water/Land Surface travel times (Batelaan and De Smedt 1998). Mapped
In humid climates where precipitation exceeds evap- phreatophytes were found to be useful for validating
oration, the hydraulic gradient is generally downward a ground water flow model that included ground water
except in lowland areas near points of surface water dis- discharge to the surface (Batelaan et al. 2003). Rosenberry
charge. The action of evaporation and transpiration in et al. (2000) found that marsh marigold (Caltha palustris)
these climates is that of reversing the flow of water from was a good indicator of spring seepage to a lake in
the surface to the ground water. From a ground water per- northern Minnesota. Marsh marigold and water-violet
spective, infiltration that makes its way to the water table (Hottonia palustris) were also found to be good indicators
is considered ground water recharge. Ground water and of upward ground water discharge in the Netherlands
dissolved nutrients may be taken up by plant roots, either (Klijn and Witte 1999).
directly from the water table (phreatophytes) or via the Classification of ground water interactions with the
unsaturated (vadose) zone. surface (i.e., recharge vs. discharge areas) using classifi-
cation of indicator vegetation is straightforward in con-
Vegetation Responses to Ground Water cept but can be tricky in practice. Vegetation species
The abundance, speciation, and growth of vegetation assemblages are interdependent, so that the presence of
are dependent upon the availability of nutrients and a dominant species in a certain location may be due to
water. As a result, classification of vegetation cover can soil chemistry (i.e., ground water) or the product of spe-
provide important clues to the occurrence of ground cies competition. Vegetation may respond slowly to
water. Such a link is relatively easy to make in arid envi- ground water flow changes and are constantly in a state
ronments, where ground water discharge or shallow water of transition. The relationship between water and ecology
tables may provide the only source of water for plants. is a burgeoning area of science called ‘‘ecohydrology.’’
The relationship between the presence of phreatophytes Ecohydrologists and wetland hydrologists have long rec-
and shallow water tables is particularly obvious in desert ognized the influence of ground water on land surface
environments (Nichols 1994). In humid environments, the ecology, although it has only been relatively recently that
relationship between plant distribution and ground water hydrogeology and ecohydrology have begun to overlap
is more complex and subtle. Plant communities are af- (Klijn and Witte 1999). Future development of vegetation
fected primarily by (1) moisture regime; (2) salinity; (3) indicators as a tool for studying ground water will depend
acidity; and (4) nutrients (Klijn and Witte 1999). Ground heavily on the collaboration between hydrogeologists and
water can certainly have an influence on these parameters, ecohydrologists.
but so can soil type, rainfall, and anthropogenic sources.
Batelaan and de Smedt (1998) used principal compo- Evapotranspiration
nent analysis to classify vegetation related to ground water Water, energy, and chemical exchange between the
discharge in a wetland preserve in Belgium. The histo- terrestrial and atmospheric components of the hydrologic
gram of the first principal component was composed of cycle has been a major focus of research within the
three Gaussian curves. Later, these remotely sensed data remote sensing community. The general motivation for
were combined with hydrochemistry and remote sensing this research is to quantify the exchange between the
in a GIS GRASS environment to estimate flow rates and atmospheric and terrestrial components of the hydrologic
314 M.W. Becker GROUND WATER 44, no. 2: 306–318
cycle and to build global climate models (National ground water is relatively longer than soil water or sur-
Research Council 1998). Such an effort requires dynamic face water. Even a small watershed, on the order of 2 km
observation of land surface and vegetation moisture between drainage divides, will take years to reach steady
fluxes at the continental scale, which can only be accom- state after a major recharge event (York et al. 2002). Due
plished through the coupling of satellite measurements to these scaling issues, coupling between ground water
with land surface hydrologic models. In particular, mod- and SVAT models has resulted in the simplification of one
els have been developed to model the soil, vegetation, and model or another. For example, the VIC-Philip2 model
atmosphere transfer (SVAT) of moisture. These models uses the sophisticated VIC SVAT model but does not pre-
have traditionally focused only on the shallow subsurface, dict ground water flow (Liang and Xie 2003), while the
with little consideration to ground water. CLASPII code couples a simple SVAT with the sophisti-
Recent studies, however, have shown that the cou- cated MODFLOW ground water flow model (York et al.
pling of ground water and SVAT models significantly im- 2002).
proves the representation of land-atmosphere fluxes.
Salvucci and Entekhabi (1995) coupled a SVAT model
with the ground water flow finite-difference model, Discussion and Conclusions
MODFLOW. A comparison of the SVAT model with There is a rich history of using satellite imagery to
and without ground water showed that the SVAT model aid hydrogeologic interpretation. Most of these uses have
overpredicted recharge at the expense of evaporation centered around mapping structural geology or bedrock
in the absence of ground water information. Levine and fractures for ground water resource evaluation (Waters
Salvucci (1999) coupled a water table–dependent vadose et al. 1990). More recently, satellite measurements of
zone model to MODFLOW to delineate recharge and dis- vegetation characteristics, topography, temperature, soil
charge zones in a Canadian prairie. Net soil water move- moisture, and gravity have been used to gather informa-
ment was predicted to be upward above shallow water tion about the presence of ground water. Remote sensing
tables and downward above deeper water tables. York has yet to become the quantitative tool that it has become
et al. (2002) coupled a simplified atmospheric model to for atmospheric, surface, and oceanic water, however.
MODFLOW to look at long-term interactions between There are few examples where important hydrogeologic
ground water and the atmosphere. At the study site in parameters have been measured directly from satellite
northeastern Kansas, 5% to 20% of evapotranspiration data. More typically, satellite imagery is one part of
was found to come directly from the aquifer. This is the a larger investigation that combines a variety of data col-
only published model to date that allows dynamic water lected on the ground or from the air.
table distribution coupled to a SVAT. Simulations demon- Tóth’s conceptual model of topographically driven
strated that, especially during times of drought, the water ground water flow (Figure 1) has an important implica-
table cannot be assumed to mimic topography. Over tion for remote sensing: the shallower the ground water
a period of months, the water table becomes flatter, thus flow system, the smaller the scale over which the surface
becoming deeper below uplands. conditions change. It follows that the shallower the
Ground water conditions change over longer time ground water system of interest, the higher the resolution
scales than atmospheric, surface, or even soil water con- of the sensing device must become. Interestingly, this
ditions. While soil water is subject to differential drying restriction may currently be more severe than the limited
and evapotranspiration at the scale of meters, the water penetration of radiation through the subsurface. In humid
table tends to distribute itself relatively smoothly over environments, shallow ground water systems may be
kilometers in areas of low relief. The residence time of influenced by surface water features that vary over the

Table 1
A Selected List of Active Space-Based Sensors That Report Data of Potential Use for
Investigations of Ground Water

Launch Ground Surface Soil Water Snow Land

Sensor Year Resolution (m) Precipitation Temperature Moisture Storage Water Cover Topography

AMSR-E 2002 5400–56,000 3 3 3 3

ASTER 1999 15, 30, 90 3 3 3
AVHRR 1991–2003 1100 3 3 3 3
GRACE 2002 300,000 3
ENVISAT-RA2 2002 ~1000 3 3
Landsat-7 1999 30, 60 3 3
MODIS 1999 250, 500, 1000 3 3
OrbView-2 1997 1100 3 3
OrbView-3 2003 1, 4 3 3 3
RADARSAT-1 1995 8–100 3
SRTM 2000 30, 90 3

M.W. Becker GROUND WATER 44, no. 2: 306–318 315

scale of meters (e.g., stream head waters, springs), obtaining spatially distributed data in remote regions or
whereas commonly used satellite-based multispectral vis- undeveloped countries. Satellite data may be the only
ible and near-infrared sensors (e.g., Landsat and ASTER) resource for predicting aquifer vulnerability, finding
have pixel resolutions on the scale of tens of meters ground water resources, or tracking aquifer depletion in
(Table 1). Satellite sensors are collecting data at a spatial such areas. Satellites are also the only practical way of
scale that is 1 or 2 orders of magnitude larger than neces- collecting hydrologic data on a global scale. Global water
sary to resolve important surface water features. As is so and energy balances are key to understanding climate
often the case, a clear trade-off exists between temporal change. At present, ground water is usually quantified in
and spatial resolution of sensors. these calculations as an abstraction from surface runoff.
Ground water hydraulic potential may be measured If the global water balance is to be closed with any cer-
directly from space, in some cases. Heads are apparent tainty, ground water fluxes and storage will need to be
where surface and ground water are well connected, such included. Combining satellite measurements with physi-
that the surface water stages are close to immediate cally based models may be the only practical approach
ground water heads. Surface water stages may be mea- for understanding the role of ground water in the global
sured through orthoimagery (e.g., ASTER), radar al- water cycle.
timetry (e.g., TOPEX/Poseidon), interferometry (e.g.,
SRTM), or lidar ranging (e.g., ICESAT). None of these
current data sources are ideal for accurately measuring Acknowledgments
surface water elevations but may be important sources This work was supported by the NASA New Investi-
of information where no other water elevations are avail- gator Program (NAG5-10608). The author is grateful to
able. Shallow water tables may also be expressed the National Research Council and the NASA Terrestrial
through snow melt, saturated soils, or surface tempera- Hydrology Program for supporting his sabbatical research
ture changes. at the NASA Goddard Space Flight Center. Review com-
Ground water fluxes can be detected from space but ments provided by Okke Batelaan and Todd Halihan sig-
cannot be measured directly via satellite. Ground water nificantly improved the manuscript and were greatly
fluxes may be detected where ground water discharges to appreciated.
the surface. Ground water discharge zones may be indi-
cated by open water, saturated soil thermal anomalies,
vegetation changes, or chemical weathering or residues.
Researchers are currently working on satellite technology References
that will allow measurement of river stage with sufficient Alsdorf, D.E., D.P. Lettenmaier, C. Vörösmarty, and NASA
accuracy to measure streamflow. Measurement of base Surface Water Working Group. 2003. The need for global,
flow from space would vastly improve on current capa- satellite-based observations of terrestrial surface waters.
EOS Transactions, AGU 84, no. 29: 269.
bilities for performing aquifer water balance in remote Anderson, M.P., and W.W. Woessner. 1992. Applied Ground Water
regions. Modeling: Simulation of Flow and Advective Transport,
Given that subsurface ground water cannot be mea- 381 pp. Academic Press, San Diego, California.
sured directly from space, the immediate future of Bacchus, S.T., D.D. Archibald, G.A. Brook, K.O. Britton, B.L.
remote sensing of ground water will certainly depend Haines, S.L. Rathbun, and M. Madden. 2003. Near-infrared
spectroscopy of a hydroecological indicator: New tool for
upon the efficient integration of data and theoretical determining sustainable yield for Floridan aquifer system.
models. There is precedence for this approach in the Hydrological Processes 17, no. 9: 1785–1809.
study of land–surface atmosphere interactions. Evapo- Banks, W.S.L., R.L. Paylor, and W.B. Hughes. 1996. Using
transpiration, for example, is currently measured through thermal-infrared imagery to delineate ground water dis-
the combination of surface radiation in the visible, ther- charge. Ground Water 34, no. 3: 434–443.
Batelaan, O., F. De Smedt, P. de Becker, and W. Huybrechts.
mal, and radar ranges. These data are rarely used alone, 1998. Characterization of a regional ground water discharge
however. Predictive models of evapotranspiration and area by combined analysis of hydrochemistry, remote sens-
soil moisture must be calibrated to historical trends ing and groundwater modelling. In Shallow Groundwater
through the process of ‘‘data assimilation’’ (National Systems, ed. P. Dillon and I. Simmers, pp. 75–86. Rotter-
Research Council 1991). Even though evaporation and dam, The Netherlands: A.A. Balkema.
Batelaan, O., F. De Smedt, and [Link]. 2003. Regional ground-
soil moisture can be observed by satellites directly, meas- water discharge: Phreatophyte mapping, groundwater mod-
urements must be made through application of physical elling and impact analysis of land-use change. Journal of
models and incorporation of multiple data types. Applica- Hydrology 275, no. 1–2: 86–108.
tion of remote sensing to ground water flow will require Becker, M.W., T. Georgian, H. Ambrose, J. Sinscalchi, and K.C.
an analogous assimilation of satellite data into numerical Fredrick. 2004. Estimating ground-water discharge using
stream temperature and velocity. Journal of Hydrology 296,
ground water flow models. no. 1–4: 221–233.
Satellite-based remote sensing of ground water is Birkett, C.M. 1998. Contribution of the TOPEX NASA radar
clearly in its infancy. At present, the best that can be altimeter to the global monitoring of large rivers and wet-
achieved with satellite sensors is to determine spatial dis- lands. Water Resources Research 34, no. 5: 1223–1239.
tribution of ground water discharge and recharge areas, Birkett, C.M., L.A.K. Mertes, T. Dunne, M.H. Costa, and M.J.
Jasinski. 2002. Surface water dynamics in the Amazon
storage changes over vast areas, or measurement of sur- Basin: Application of satellite radar altimetry. Journal of
face water heads in large water bodies. Although this Geophysical Research—Atmospheres 107, no. D20, doi:
may not seem promising, there are few alternatives for 10.1029/2001JD000609.

316 M.W. Becker GROUND WATER 44, no. 2: 306–318

Bjerklie, D.M., S.L. Dingman, C.J. Vorosmarty, C.H. Bolster, of aquifer system compaction and land subsidence using
and R.G. Congalton. 2003. Evaluating the potential for interferometric synthetic aperture radar, Antelope Valley,
measuring river discharge from space. Journal of Hydrol- Mojave Desert, California. Water Resources Research 34,
ogy 278, no. 1–4: 17–38. no. 10: 2573–2585.
Bobba, A.G., R.P. Bukata, and J.H. Jerome. 1992. Digitally pro- Gillespie, A.R. 1992. Spectral mixture analysis of multispectral
cessed satellite data as a tool in detecting potential ground- thermal infrared images. Remote Sensing of Environment
water flow systems. Journal of Hydrology 131, no. 1–4: 42, no. 2: 137–145.
25–62. Goteti, G., J.S. Famiglietti, D.E. Alsdorf, P. Bates, E. Clark, D.P.
Burbey, T.J. 2003. Use of time-subsidence data during pumping Lettenmaier, D. Moller, and M.D. Wilson. 2004. A virtual
to characterize specific storage and hydraulic conductivity surface water satellite mission: Identifying key science is-
of semi-confining units. Journal of Hydrology 281, no. 1–2: sues. EOS Transactions, AGU 85, no. 47: Fall Meeting
3–22. Supplement, Abstract H22C-07.
Cartwright, K. 1979. Measurement of fluid velocity using tem- Heilman, J.L., and D.G. Moore. 1982. Evaluating depth to shal-
perature profiles: Experimental verification. Journal of low groundwater using Heat-Capacity Mapping Mission
Hydrology 43, no. 1–4: 185–194. (HCMM) data. Photogrammetric Engineering and Remote
Cartwright, K. 1974. Tracing shallow groundwater systems by Sensing 48, no. 12: 1903–1906.
soil temperatures. Water Resources Research 10, no. 4: Hoffmann, J., D.L. Galloway, and H.A. Zebker. 2003. Inverse
847–855. modeling of interbed storage parameters using land sub-
Cartwright, K. 1970. Groundwater discharge in the Illinois basin sidence observations, Antelope Valley, California. Water
as suggested by temperature anomalies. Water Resources Resources Research 39, no. 2: 13 p.
Research 6, no. 3: 912–918. Hoffmann, J., H.A. Zebker, D.L. Galloway, and F. Amelung.
Cartwright, K. 1968. Temperature prospecting for shallow gla- 2001. Seasonal subsidence and rebound in Las Vegas
cial and alluvial aquifers in Illinois. Illinois State Geo- Valley, Nevada, observed by synthetic aperture radar inter-
logical Survey Circular 433. Urbana, Illinois. ferometry. Water Resources Research 37, no. 6: 1551–
Chase, M.E. 1969. Airborne remote sensing for ground water 1566.
studies in a prairie environment. Canadian Journal of Huntley, D. 1978. Detection of shallow aquifers using thermal
Earth Sciences 6, 737–741. infrared imagery. Water Resources Research 14, no. 6:
Chow, V.T., D.R. Maidment, and L.W. Mays. 1988. Applied 1075–1083.
Hydrology. New York: McGraw-Hill. Jackson, T.J. 2002. Remote sensing of soil moisture: Im-
Conant, B. 2004. Delineating and quantifying ground water dis- plications for groundwater recharge. Hydrogeology Journal
charge zones using streambed temperatures. Ground Water 10, no. 1: 40–51.
42, no. 2: 243–257. Klijn, F., and J.P.M. Witte. 1999. Eco-hydrology: Groundwater
Constantz, J.E. 1998. Interaction between stream temperature, flow and site factors in plant ecology. Hydrogeology Jour-
streamflow, and groundwater exchanges in alpine streams. nal 7, no. 1: 65–77.
Water Resources Research 34, no. 7: 1609–1615. Koblinsky, C.J., R.T. Clarke, A.C. Brenner, and H. Frey. 1993.
Crowley, J.K., and S.J. Hook. 1996. Mapping playa evaporite min- Measurement of river level variations with satellite altime-
erals and associated sediments in Death Valley, California, try. Water Resources Research 29, no. 6: 1839–1848.
with multispectral thermal infrared images. Journal of Geo- Levine, J.B., and G.D. Salvucci. 1999. Equilibrium analysis
physical Research—Solid Earth 101, no. B1: 643–660. of groundwater–vadose zone interactions and the re-
de Vries, J.J. 1995. Seasonal expansion and contraction of sulting spatial distribution of hydrologic fluxes across
stream networks in shallow groundwater systems. Journal a Canadian prairie. Water Resources Research 35, no. 5:
of Hydrology 170, no. 1–4: 15–26. 1369–1383.
de Vries, J.J. 1994. Dynamics of the interface between streams Liang, X., and Z.H. Xie. 2003. Important factors in land-
and groundwater systems in lowland areas, with reference atmosphere interactions: Surface runoff generations and in-
to stream net evolution. Journal of Hydrology 155, no. 1–2: teractions between surface and groundwater. Global and
39–56. Planetary Change 38, no. 1–2: 101–114.
Domenico, P.A., and F.W. Schwartz. 1998. Physical and Chem- Lu, Z., and W.R. Danskin. 2001. InSAR analysis of natural
ical Hydrogeology, 2nd ed. New York: John Wiley & Sons. recharge to define structure of a ground-water basin, San
Entekhabi, D., E.G. Njoku, P. Houser, M. Spencer, T. Doiron, Bernardino, California. Geophysical Research Letters 28,
Y. Kim, J. Smith, R. Girard, S. Belair, W. Crow, T.J. Jackson, no. 13: 2661–2664.
Y.H. Kerr, J.S. Kimball, R. Koster, K.C. McDonald, Mabee, S.B., and K.C. Hardcastle. 1997. Analyzing outcrop-
P.E. O’Neill, T. Pultz, S.W. Running, J. Shi, E. Wood, and scale fracture features to supplement investigations of bed-
J. van Zyl. 2004. The hydrosphere state (hydros) satellite rock aquifers. Hydrogeology Journal 5, no. 4: 21–36.
mission: An earth system pathfinder for global mapping of Meijerink, A.M.J. 1996. Remote sensing applications to hydrol-
soil moisture and land freeze/thaw. IEEE Transactions on ogy: Groundwater. Hydrological Sciences Journal—Journal
Geoscience and Remote Sensing 42, no. 10: 2184–2195. Des Sciences Hydrologiques 41, no. 4: 549–561.
Falconer, A., L. Myers, and M.A. Deutsch, ed. 1979. Ob- Mertes, L.A.K. 2002. Remote sensing of riverine landscapes.
servations on Lake Ontario Basin hydrogeology from Freshwater Biology 47, no. 4: 799–816.
optical enhancements of landsat imagery. In Satellite Miller, J.A. 1998. Principal Aquifers of the 48 Contiguous
Hydrology: Fifth Annual William T. Pecora Memorial Sym- United States, 1:5,000,000-scale Map. U.S. Geological
posium on Remote Sensing, ed. M.D.W. Deutsch and Survey, Reston, Virginia.
A. Rango pp 427–436. American Water Resources Associ- Moore, G.K. 1996. Using thermal-infrared imagery to delineate
ation, Minneapolis, Minnesota Sioux Falls, South Dakota. ground-water discharge—Discussion. Ground Water 34,
Fenneman, N.M., and D.W. Johnson. 1946. Physical Divisions no. 6: 962–962.
of the United States, 1:7,000,000 Map. U.S. Geological Moore, G.K. 1982. Ground-water applications of remote sens-
Survey, Reston, Virginia. ing. USGS Open File Report 82-240. U.S. Geological
Fetter, C.W. 2001. Applied Hydrogeology, 4th ed. Upper Saddle Survey, Reston, Virginia.
River, New Jersey: Prentice Hall. Moore, R.B., G.E. Schwartz, S.F. Clark Jr., G.J. Walsh, and J.R.
Food and Agriculture Organization. 2003. Review of World Degnan. 2002. Factors related to well yield in the fractured-
Water Resources by Country. Rome, Italy: Food and Agri- bedrock aquifer of New Hampshire. USGS Professional
culture Organization of the United Nations. Paper. U.S. Geological Survey, Reston, Virginia.
Galloway, D.L., K.W. Hudnut, S.E. Ingebritsen, S.P. Phillips, National Research Council. 1998. Global Energy and Water
G. Peltzer, F. Rogez, and P.A. Rosen. 1998. Detection Cycle Experiment (GEWEX) Continental-Scale Interna-

M.W. Becker GROUND WATER 44, no. 2: 306–318 317

 tional Project: A Review of Progress and Opportunities, Smith, L.C. 2002. Emerging applications of interferometric syn-
Global Energy and Water Cycle Experiment (GEWEX) thetic aperture radar (InSAR) in geomorphology and
Panel, Commission on Geosciences, Environment and Re- hydrology. Annals of the Association of American Geogra-
sources. Washington, D.C.: National Academy Press. phers 92, no. 3: 385–398.
National Research Council. 1994. Alternatives for Ground Water Smith, L.C. 1997. Satellite remote sensing of river inundation
Cleanup. Washington, D.C.: National Academy Press. area, stage, and discharge: A review. Hydrological Pro-
National Research Council. 1991. Four-Dimensional Model cesses 11, no. 10: 1427–1439.
Assimilation of Data: A Strategy for the Earth System Sci- Solley, W.B., R.R. Pierce, and H.A. Perlman. 1998. Estimated
ences, Commission on Geosciences Environment and Re- use of water in the United States in 1995. U.S. Geological
sources. Washington, D.C.: The National Academies Press. Survey, Reston, Virginia.
Nichols, W.D. 1994. Groundwater discharge by phreatophyte Swenson, S., J. Wahr, and P.C.D. Milly. 2003. Estimated accura-
shrubs in the Great Basin as related to depth to ground- cies of regional water storage variations inferred from the
water. Water Resources Research 30, no. 12: 3265–3274. Gravity Recovery and Climate Experiment (GRACE).
Poeter, E.P., and M.C. Hill. 1998. Documentation of UCODE, Water Resources Research 39, no. 8: 1223.
a computer code for universal inverse modeling. Paper read Tóth, J. 1963. A theoretical analysis of groundwater flow in
at Water-Resources Investigations, U.S. Geological Survey, small drainage basins. Journal of Geophysical Research 68,
Reston, Virginia. 4795–4812.
Pool, D.R., and J.H. Eychaner. 1995. Measurements of aquifer- Townsend, P.A., and J.R. Foster. 2002. A synthetic aperture radar-
storage change and specific yield using gravity surveys. based model to assess historical changes in lowland flood-
Ground Water 33, no. 3: 425–432. plain hydroperiod. Water Resources Research 38, no. 7.
Portnoy, J.W., B.L. Nowicki, C.T. Roman, and D.W. Urish. doi:10.1029/2001WR001046.
1998. The discharge of nitrate-contaminated groundwater U.S. EPA. 2003. Multimedia, multipathway, and multirecep-
from developed shoreline to marsh-fringed estuary. Water tor risk assessment (3MRA) modeling system. U.S. EPA
Resources Research 34, no. 11: 3095–3104. Washington D.C.
Reutov, E.A., and A.M. Shutko. 1992. Estimation of the depth to USGS. 1992. Standards for digital elevation models: Part 3 qual-
a shallow water-table using microwave radiometry. Inter- ity control. In National Mapping Program Technical In-
national Journal of Remote Sensing 13, no. 12: 2223–2232. structions. Reston, Virginia: USGS.
Rodell, M., and J.S. Famiglietti. 2002. The potential for satellite- Waters, P., D. Greenbaum, P. Smart, and H. Osmaston. 1990.
based monitoring of groundwater storage changes using Applications of remote sensing to groundwater hydrology.
GRACE: The High Plains aquifer, Central US. Journal of Remote Sensing Reviews 4, no. 2: 223–264.
Hydrology 263, no. 1–4: 245–256. Watkins, M.M. 2004. Bowie lecture: Time variable gravity
Rosenberry, D.O., R.G. Striegl, and D.C. Hudson. 2000. Plants measurements come of age. EOS Transactions, AGU 85,
as indicators of focused ground water discharge to a north- no. 47: Fall Meeting Supplement, Abstract G24A-01.
ern Minnesota lake. Ground Water 38, no. 2: 296–303. Watson, K.M., Y. Bock, and D.T. Sandwell. 2002. Satellite inter-
Rundquist, D., G. Murray, and L. Queen. 1985. Airborne ther- ferometric observations of displacements associated with
mal mapping of a flow-through lake in the Nebraska sand- seasonal groundwater in the Los Angeles basin. Journal
hills. Water Resources Bulletin 21, no. 6: 989–994. of Geophysical Research—Solid Earth 107, no. B4.
Salama, R.B., I. Tapley, T. Ishii, and G. Hawkes. 1994. Identifi- doi:10.1029/2001JB000470.
cation of areas of recharge and discharge using landsat-tm Winter, T.C. 1999. Relation of streams, lakes, and wetlands to
satellite imagery and aerial-photography mapping techni- groundwater flow systems. Hydrogeology Journal 7, no. 1:
ques. Journal of Hydrology 162, no. 1–2: 119–141. 28–45.
Salvucci, G.D., and D. Entekhabi. 1995. Hillslope and climatic Winter, T.C., J.S. Eaton, and G.E. Likens. 1989. Evaluation of
controls on hydrologic fluxes. Water Resources Research inflow to Mirror Lake, New Hampshire. Water Resources
31, no. 7: 1725–1739. Bulletin 25, no. 5: 991–1008.
Sams, J.I. III, and G.A. Veloski. 2003. Evaluation of airborne Winter, T.C., J.W. Harvey, Q.L. Franke, and W.M. Alley. 1998.
thermal infrared imagery for locating mine drainage sites Ground water and surface water: A single resource. USGS
in the Lower Kettle Creek and Cooks Run Basins, Pennsyl- circular 1139. Denver, Colorado: USGS.
vania, USA. Mine Water and the Environment 22, no. 2: Wolock, D. 2003. Flow characteristics at U.S. Geological Sur-
85–93. vey streamgages in the conterminous United States. USGS
Schmidt, D.A., and R. Burgmann. 2003. Time-dependent land Open-File Report 03-146. U.S. Geological Survey, Reston,
uplift and subsidence in the Santa Clara valley, California, Virginia.
from a large interferometric synthetic aperture radar data Wood, C.R. 1972. Ground-water flow. Photogrammetric Engi-
set. Journal of Geophysical Research—Solid Earth 108, neering 38, no. 4: 347–352.
no. B9. doi:10.1029/2002JB002267. York, J.P., M. Person, W.J. Gutowski, and T.C. Winter. 2002.
Silliman, S.E., and D.F. Booth. 1993. Analysis of time-series Putting aquifers into atmospheric simulation models:
measurements of sediment temperature for identification of An example from the Mill Creek Watershed, northeast-
gaining vs. losing portions of Juday Creek, Indiana. Jour- ern Kansas. Advances in Water Resources 25, no. 2:
nal of Hydrology 146, no. 1–4: 131–148. 221–238.

318 M.W. Becker GROUND WATER 44, no. 2: 306–318