Comparison of Seasonal Terrestrial Water Storage Variations from GRACE with in situ Measurements from the High Plains Aquifer

1
Comparison of Seasonal Terrestrial Water Storage
Variations from GRACE with in situ Measurements
from the High Plains Aquifer

• Gil Strassberg1, Bridget Scanlon1, Matthew Rodell2

Bureau of Economic Geology, March 2007
1 Bureau of Economic Geology 2 Hydrological
Sciences Branch, NASA Goddard Space Flight Center
2
Outline

1. GRACE
2. High Plains Aquifer
3. Data and Methods
4. Results

3
Gravity Recovery and Climate Experiment (GRACE)

• Two satellites distance 220km
• Satellites detect changes in Earth's gravity
  field by monitoring the changes in distance
  between the satellites as they orbit Earth.

Launch March 2002
Figures from CSR website http//www.csr.utexas.e
du/grace/
Distance 220 km
4
How it works
The positions of the two GRACE satellites change
in response to variations in Earth's gravity field
Lead satellite pulls away due to gravity change
High accuracy device can detect changes within 10
µm
Figures from CSR website http//www.csr.utexas.e
du/grace/
5
Can we monitor water storage changes from space?

• Evaluate the use of GRACE data for estimating
  water storage changes
• Use the High Plains Aquifer as a case study

changes in water mass should be readily
detectable in a region the size of the High
Plains aquifer
Satellite Gravity and the Geosphere
Contributions to the Study of the Solid Earth and
Its Fluid Envelopes (1997).
6
From gravity to water mass

• GRACE provides time varying global gravity fields
  (30 days)
• Observed monthly changes in gravity are caused by
  monthly changes in mass
• Mass changes are mainly attributed to changes in
  the distribution of water mass in
• Hydrologic reservoirs (soil, groundwater,
  waterbodies, snow)
• Oceans
• Atmosphere
• Cryosphere (frozen)

Figure from GRACE Tellus website
http//gracetellus.jpl.nasa.gov/
7
Terrestrial water storage (TWS)
Atmospheric mass
Total mass
Terrestrial mass
Total mass TWS Atmospheric mass

• Terrestrial Water Storage includes
• Water in reservoirs and rivers
• Soil moisture
• Groundwater
• Snow and Ice
• Biomass

Estimated from climate models European Center for
Medium-Range Weather Forecasts (ECMWF)
8
Estimating Storage Changes

• GRACE satellites measure monthly gravity changes
• Gravity changes ? Terrestrial Water Storage
  changes
• Storage changes in TWS components can be
  estimated when combined with other data (modeled
  or measured)

Groundwater storage
Snow water equivalent
Biomass
Terrestrial water storage
Soil moisture
Surface water
9
Previous studies - Illinois
Yeh et al (2006) compared GRACE TWS changes in
Illinois (200,000 km2) with monitored soil
moisture and groundwater
GRACE TWS
SM GW IZ
TWS - SM
GW IZ
10
Previous studies Greenland Ice Sheet
Chen et al. (2006) showed melting of the ice
sheet in East Greenland
Mass changes between April 2002 and November 2005
11
Previous studies Land Surface Models

• TWS data is useful for calibration and validation
  of land surface models
• Niu et al. (2006) combined GRACE TWS with modeled
  subsurface water to compare with modeled snow
  water equivalent in 4 arctic river basins

• AMSR Advanced Microwave Scanning Radiometer
• Model NCAR Community Land Surface Model (CLM)

12
Other studies

• GRACE data is used to help monitor and model many
  large scale processes
• Estimation of ET and P-ET over large river basins
• Continental water cycle and continental ocean
  discharge
• Sea level change
• Water balance in large lakes (e.g. Three-Gorges
  Reservoir, lake Chad)
• Climate change
• Earthquake effects

GRACE data capturing changes in the Earths
gravity field due to the December 2004 Sumatra
earthquake (CRS website http//www.csr.utexas.edu
/GRACE/publications/press/ )
13
Outline

1. GRACE
2. High Plains Aquifer
3. Data and Methods
4. Results

14
High Plains Aquifer

• Underlies 8 states in the U.S. ( 450,000 km2)
• One of the principal aquifers in the U.S. and one
  of the major agriculture areas of the world (
  175,000 km2 cultivated)
• 27 of irrigated land in the US
• 30 of the groundwater used for irrigation in the
  US
• Highly monitored by Federal and State agencies

15
Irrigation

• Semi arid conditions (P 400-700 mm and E 1500
  - 2700 mm)
• 50,000 km2 of irrigated area, 30 of cultivated
  area)
• 23.5 km3 (19M acre-feet) of water pumped from the
  aquifer in 2000, with about 95 withdrawn for
  irrigation
• Equivalent to 52 mm of water over the entire
  aquifer area

16
Irrigation and Water Levels

• Water levels and storage have been declining due
  to irrigation (average 3.8 m and maximum of 68 m)
• USGS surveys since 1988 only in winter
• Good estimate of groundwater storage (9200 wells
  monitored in 2003)

Groundwater storage from predevelopment to 2003
million-acre feet
Figure from McGuire (2004) http//pubs.usgs.gov/f
s/2004/3097/
17
Expected seasonal groundwater signal

• 90 of groundwater pumped during summer
  (May-August)
• Water levels recover to a static level during
  winter
• Expect a seasonal signal in groundwater storage
• Long term trend of decreasing water storage

Winter
Winter
Groundwater storage
Summer
Summer
18
Outline

1. GRACE
2. High Plains Aquifer
3. Data and Methods
4. Results

19
Estimating Groundwater Storage Changes

• Combine GRACE derived Terrestrial Water Storage
  (TWS) and modeled Soil Moisture (SM) from a land
  surface model to estimate groundwater changes

Groundwater storage
Snow water equivalent
Biomass
Terrestrial water storage
Soil moisture
Surface water
20
Terrestrial Water Storage

• TWS is derived from 3 datasets processed by the
  GRACE partners.
• Data represent mass anomalies after taking out
  atmospheric mass.
• Data is smoothed with a 400-km Gaussian smoother
  to isolate the signal.

Maximum TWS anomaly 80 mm
Missing data
mass anomaly (mm)
21
Soil Moisture and TWS

• Soil moisture was estimated from the Noah land
  surface model driven by the North American Land
  Data Assimilation System (NLDAS).
• 1/8 degree grid over North America
• High resolution parameters including hourly
  observation-based precipitation and solar
  radiation, and detailed soil and vegetation.

SWE is negligible
mm water
22
Is SWE really negligible?
Drilling near Amarillo
23
Soil Moisture

• Soil moisture from 86 sites
• Seasonal SM changes were calculated for each site
• Data were averaged on a 1x1 degree grid
• Estimated overall SM changes
• Compared to modeled SM anomalies

SM anomaly (mm)
24
Surface water

• Flat terrain, no major reservoirs
• Internally drained into ephemeral lakes (playas)
• 53,000 playas covering 0.5 of the land surface
  area

http//www.rw.ttu.edu/torrence/
25
Surface water

• Use 1/3 second (10m) DEM to estimate playa
  depths
• Sample of 130 playas
• Average depth 1.06 m

Playa area
0.5 1.06 m 5.3 mm
Sample area
1 m
Depth hmax hmin
hmax
hmin
Elevation from 1/3 arc second (10m)
26
Biomass

• Rodell et al. (2005) estimated biomass changes
  from remotely sensed Leaf Area Index (LAI)
• Seasonal variations were lt 5 mm

Figure form Rodell et al (2005)
27
Estimating Groundwater Storage

• Assimilated water levels from groundwater
  databases USGS NWIS, Texas Water Development
  Board (TWDB), and Kansas WIZARD groundwater
  database

2,719 wells with seasonal water level changes
Water level change between 4-6 2003 and 7-9 2003
Water level changes are calculated per well as
differences between seasonal averages
28
Spatial Aggregation of Water Level Changes

• Water level changes were averaged over a 1x1
  degree mesh
• Based on wells with daily water levels we filter
  out changes gt 4.57 m (15 feet)

Water level change Between 4-6 2003 and 7-9 2003
Spatially aggregated water level changes
Water level change (m)
29
Variations in Groundwater Storage

• Convert water level changes to storage changes
  (Sy 0.15)
• Seasonal cycle with an amplitude of 75 mm
• Compare to changes (winter to winter) from USGS

Anomalies of groundwater storage
Period (winter-winter) Our Analysis USGS
(1) 2003-2004 -45 mm -37 mm
(2) 2004-2005 -9 mm -9 mm
1
2
30
Outline

1. GRACE
2. High Plains Aquifer
3. Data and Methods
4. Results

31
Groundwater Storage and Soil Moisture
Changes of groundwater and soil moisture are same
order of magnitude
Seasonal variation in SM and GWS
32
TWS vs. Groundwater Storage Soil Moisture

• Seasonal TWS change ranged 8-93 mm, mean 43 mm
• Changes winter/spring summer/fall range 40-100,
  mean 75 mm

Estimated the TWS uncertainty between 32-37 mm,
mean 34 mm
33
Groundwater Storage vs. TWS - Soil Moisture

• Seasonal TWS-SM change ranged 1-63 mm, mean 26
  mm
• Changes winter/spring summer/fall range 30-100,
  mean 57 mm

Close to estimated discharges of 52
34
Detectability of changes
Mean uncertainty in TWS-SM changes 49 mm
Detectible Uncertainty 49 lt change
Uncertainty estimate might be too conservative
and will probably be modified

• Winter/spring Summer/fall TWS changes detected
  4 out of 5 periods
• Maximum TWS-SM were detected 3 out of 5 periods

35
Summary

• Seasonal changes in TWS in the High Plains
  aquifer area are due to changes in soil moisture
  and groundwater storage (surface water, snow
  water equivalent, and biomass are negligible)
• GRACE-derived TWS agreed well with SM GWS (same
  magnitude, trends, R 0.82)
• Variations in groundwater storage from water
  level measurements showed a fair agreement with
  TWS SM (same trend and magnitude, R 0.58)
• TWS changes were detectible in 4 out of 5 periods
  (winter/spring-summer/fall)
• TWS-SM changes were detectible in 3 out of 5
  periods

36
Future Work

• Add data for 2006
• Include new release of GRACE data
• Add new soil moisture data
• Spatial variations
• Improve uncertainty estimate
• See how this can be applied in other regions

Winter-Spring 2005 SM change
Seasonal SM change (mm)
37
Questions?