The document discusses several studies related to monitoring surface and groundwater resources using remote sensing techniques.
1) One study compares soil moisture estimations from the Advanced Microwave Scanning Radiometer E (AMSR-E), ground-based measurements, and the Common Land Model (CLM). It finds that AMSR-E captures drying and wetting patterns but with lower variability than CLM or ground data.
2) Another evaluates global soil moisture from the ERS scatterometer and AMSR-E, finding general agreement except in deserts and dense vegetation due to limitations.
3) A third analyzes terrestrial water storage changes using GRACE satellite data and GLDAS land surface models,
3. Approaches to and Problems with
Measuring Soil Moisture
1. in situ field measurements
a.
b.
c.
d.
short duration
intensive field experiments
very sparse
field or regional mean soil moisture not properly represented
2. land surface models
a. limited measurements of model physical parameters
b. input data errors
3. remote sensing observations
a. shallow depth
b. scale is overly coarse
4. Remote Sensing Observations
Advanced Microwave Scanning Radiometer Earth Observing System
(AMSR-E) on the Aqua satellite
o May 2002
o modified passive microwave radiometer on Advanced Earth
Observing Satellite-II (ADEOS-II)
o measures brightness temperatures at 6 frequencies
o 6.9 GHz (C band) and 10.7 Ghz (X band)
o soil moisture algorithm uses a microwave transfer model to compare
observed and computed brightness temperature
o calculated by NSIDC and VUA-NASA (de Jeu et al., 2008)
6. Remote Sensing Observations (cont)
ERS scatterometer from the ERS-1 and ERS-2
o monitors wind speed and direction over the oceans
o configured a real aperture radar providing 2 radar images
o 50 km spatial resolution
o 500 km swath width
o active microwave sensors: sends out a signal and measures how
much of that signal returns after interacting with the target
o ERS-1 mission: 1991 to March 10, 2000
o ERS-2: 1995 to September 5, 2011
7. Remote Sensing Observations (cont)
SMOS
o carried on Proteus
o measures microwave radiation emitted from Earth’s surface
within the ‘L-band’ (around a frequency of 1.4 GHz)
o provide:
global maps of soil moisture every three days at a spatial
resolution of 50 km
global maps of sea-surface salinity down to 0.1 practical
salinity units for a 30-day average over an area of 200×200 km
9. AMSR-E (Aqua)
ERS
SMOS
6.9, 10.7, 18.7, 23.8, 36.5,
89.0
C-band (5.3 Ghz)
L-band (1.4 GHz)
varies from 5.4 km at 89 GHz
to 56 km at 6.9 GHz
50 km
50 km
Orbital
Near-circular, polar, Sunsynchronous
Syn-synchronous
Return Frequency
ascending (1:30pm) and
descending (1:30am) mode
35 days cycle
Global coverage every 3
days
Temporal Duration
2001
(no data from 89 Ghz after
2004; stopped spinning on
Oct 4)
EM frequencies (GHz)
Spatial resolutions
Orbital or Geostationary
1991 to March 10, 2000
(ERS-1)
1995 to September 5, 2011
(ERS-2)
2009 - present
10. “Remote sensing observatory validation of surface soil
moisture using Advanced Microwave Scanning
Radiomater E, Common Land Model, and ground based
data: Case study in SMEX03 Little River Region, Georgia,
U.S.”
Chou et al., 2008
11. Purpose
•
compare soil moisture estimations from:
o AMSR-E
o ground-based measurement
o Soil-Vegetation-Atmosphere Transfer (SVAT) model
combine land surface and atmosphere processes
modeling using both water and energy balances
• Common Land Model (CLM)
require model forcing data and certain parameters
12. Methods: AMSR-E
iterative multi-channel inversion procedure
o
microwave transfer model to compare observed
brightness temp (TB) and computed brightness
temp (TBP)
affected by soil volumetric water content, vegetation water
content (VWC), and surface temp (Ts)
14. Results
•
•
•
•
agreed well in drying and wetting
patterns
average soil moisture: 0.122 to 0.167
m3/m3
AMSR-E
o lower variability
o weak agreement with in situ &
CLM
o did not capture temporal
variability during SMEX03 period
CLM
o wetter than observed
o followed patterns during SMEX03
period
15. “Global Soil Moisture Patterns Observed by Space Borne
Microwave Radiometers and Scatterometers.”
de Jeu et al., 2008
16. Purpose
global evaluation of:
○
ERS scatterometer
■
○
obtained from 50 km scatterometer originally
designed for measuring winds over the oceans
AMSR-E soil moisture data
■
uses low frequency microwave brightness
temperatures to obtain soil moisture
17. Methods: AMSR-E Soil Moisture
•
contribution of the atmosphere to observed
brightness temperature
o
function of physical temperature of the radiating body
and its emissivity
18. Methods: AMSR-E Soil Moisture
•
contribution of the atmosphere to observed brightness
temperature
o function of physical temperature of the radiating body and its
emissivity
•
radiation from the land surface observed above canopy
19. (a) Comparison of smooth surface emissivity and the soil dielectric constant according to the Fresnel
relations with an incidence angle of 55 degrees. (b) Comparison of the soil dielectric constant and soil
moisture for typical sand, loam and clay soils.
20. Methods: SRS scatterometer
• soil moisture derived using retrieval method
proposed by Wagner et al. (1999, 2003)
change detection approach tracks relative soil
moisture changes rather than absolute
o dry and wet reference conditions identified
based on multi-year backscatter time series
o
21. Average soil moisture for 2006: (a) and (b) derived from ERS data, (c) and (d) from C-band
AMSR-E, and (e) and (f) from X-band AMSR-E.
22. Results
•
AMSR-E (X-band and C-band)
o
•
active radar instruments on the ground cause Radio
Frequency Interference (RFI) in C-band
ex) eastern part of the USA
ERS scatterometer
o
volume scattering in dry soil or reduced sensitivity of
dielectric constant
ex) wet region in northern Mexico
23. Results (cont)
•
Comparison between ERS scatterometer and AMSR-E
low and negative values found in deserts and more
densely vegetated regions
•
•
due to low sensitivity of dielectric constant in desert
effect of mountains
•
average correlation coefficient of 0.83
•
•
-0.08 and 0.33, respectively
explained by limited soil moisture retrieval capabilities
strong similarity in sparse to moderate vegetated
regions
low correlations in densely vegetated areas and
deserts
•
potential to combine both products
25. Purpose
● Spatial-temporal variations in Terrestrial Water Storage
Changes (TWSC)
● Compared results with those simulated Global Land Data
Assimilation Systems (GLDAS)
● Additionally, GLDAS simulated to infer TWSC partitioned in
snow, canopy water, and to understand how variations in the
hydrologic fluxes act to enhance or dissipate stores.
26. Methods
• To investigate water storage changes
o Groundwater storage monitoring
GRACE-driven TWSC
Global Land Data Assimilation System (GLDAS)
o Data frame
GRACE-driven data Center for Space Research RL01 (April 2002-July 2004, &
June, 2003).
collected corrected GRACE Stokes coefficients expanded to degree and order 60
smoothed with 1000 km half-width Gaussian averaging kernel to different gravity
estimates.
lesser degree, the degree two, and zero were not considered due to their
quantifiable errors.
smoothed spherical harmonics coefficients were transformed into 1 x 1 degree
gridded data, which represented vertically integrated water mass changes over
100 kilometers with accuracy of about 1.5 cm equivalent water thickness.
27. Methods (cont)
● PRIMARY LAND SURFACE FLUX DATA
○ NASA GLDAS
○ 1 degree, 3 hourly outputs from 1979 to present (Noah Land Surface ModelGLDAS).
○ Hydrologic fluxes and storages were gathered from January, 2002-December,
2004.
● GRACE-DRIVEN TWSC ESTIMATES
○ by differencing the monthly anomalies.
■ were derived from the mean gravity field from each monthly GRACE
resolutions.
■ estimates TWSC as average changes in TWS from one month to the other.
○ TWS (Total Soil Moisture, Snow Water Equivalent, & Canopy Water Storage).
28. Methods (cont)
1
Equation for the comparable
replication of GRACE
observations from GLDAS land
surface output:
Following the results of equation (1), estimates of TWSC from
GLDAS that closely approximate GRACE were computed; where,
the terms to the right of the equations are 15th day averages of
each calendar of the year with the assumption that average 15th
day can be representative of approximately 30 days average.
2
S, represents the average TWS for
the index day (i), and the subscripts
(i) and (N) represent day of month
and month respectively, and (t) is
time.
Source: Syed et al. 2008
Calculating TWSC using the monthly basin-scale terrestrial
water balance provides approximations; where, P
(Precipitation), R (Runoff), and E (Evapotranspiration)
3
29. Results
Terrestrial Water Storage
●
1a). TWS peaked during the NH Winter (DJF) with an amplitude of 0.6 cm/month
1b). Shows seasonal averages with strongest water storage change signals in a SH 0 to 30 degree S latitude band
with lesser peak in NH subtropics at 60 degree N.
1c). Shows associated peaks in amplitude of seasonal cycle in the zonally averaged absolute value of TWSC in
corresponding regions (two issues: 1st global TWSC data & TWS is predictable to precipitation & evaporation).
1a
1b
1c
Source: Syed et al. 2008
●
●
30. Source: Syed et al. 2008
Results (cont)
●
Increase in annual mean in Europe (0.32 cm/month), South America (0.30 cm/month, and Asia (0.08
cm/month), lesser depletion of total water storage in Australia (-0.14 cm/month, Africa (-0.02 cm/mont, & N.
America (-0.06 cm/month.
●
Tropical basins in NH gain water during JJA from precipitation, while basins in the SH tropics and those in NH
mid-to-high latitudes lose water.
●
Higher seasonal averaged amplitudes TWSC were noted in the tropics of the SH of latitudinal variability
compared to tropics of the NH.
31. Results (cont)
● GRACE-GLDAS Comparison
○
computed using equation 1 & 2
○
global model output from GLDAS
captures the magnitude and variations
of terrestrial hydrology.
○
good overall agreement between the
two estimates with RMSE ranging about
1 cm/month in JJA and about 0.7
cm/month in DJF.
32. Results (cont)
● TIME SERIES OF TWSC from
GRACE & GLDAS
○
GLDAS estimates agreed very well with
GRACE, with RMSE values of about 1.5
cm/month in the Mississippi and
Mackenzie River Basins
○
And about 2.5 cm/month in the Amazon
and Parana River Basins
○
Overall, Figures 4 & 5 show agreements
in spatial-temporal variability of TWSC
estimates from GRACE and GLDAS.
33. Conclusion
• The study characterized TWSC variations using GRACE and GLDAS:
o Global, zonal, and basin-scale estimates of GRACE-driven storage
changes indicate a wide range in variability and magnitude with
emphasis on the space-time heterogeneity in TWSC response.
o Continental and hemispheric differentiations in precipitation were noted.
o Averaged TWSC was found to have greatest amplitudes zonally in the
tropics of the SH (about 7 cm/month).
o At the river basin-scale, comparative analyses between GLDAS and
GRACE-driven estimates of TWSC agreed well.
o Noah Land Model used in the GLDAS simulations did not include surface
and groundwater stores because of the inability to quantify their
contribution to storage change.
34. Comparison and assimilation of global soil moisture
retrievals from the Advanced Microwave Scanning
Radiometer for the Earth Observing System (AMSRE) and the Scanning Multichannel Microwave
Radiometer (SMMR).”
“
Reichle et al., 2007
35. Purpose
•
•
to compare two satellite data sets of surface soil moisture
retrievals.
assimilate the preliminary products into NASA Catchment Land
Surface Model (CLSM) to determine retrieved soil moisture
using multiyear means and temporal variability as units to
determine the difference.
36. Methods
•
•
Global soil moisture retrievals
o NASA Catchment Land Surface Model (CLSM)
o Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E) on
the Aqua satellite
Scanning Multichannel Microwave Radiometer (SMMR).
o
Data frame
o Satellite-driven soil moisture retrievals
infer soil moisture from microwave signals
o Land model integrations
relates soil moisture to antecedent meteorological forcing
o Ground-based measurements
provides direct and accurate measurements of soil moisture
37. Methods (cont)
Satellite-driven soil moisture retrievals
o Using NASA Level-2B AMSR-E “AE_Land Product
o infer soil moisture from microwave signals
surface temperature inputs are from SMMR 37 GHz vertical polarization
channel (which are stored on 0.25 degree grid) their resolution is about
120 km based on footprint of 148 km by 965 km.
both day and night overpasses were used..
Quality control measures > used AMSR-E data points corresponding
flags for light vegetation, no rain, no snow, no frozen ground, and no
Radio Frequency Interference (RFI).
38. Methods (cont)
• Land integration Model
o obtained from the integration of NASA CLSM.
computational unit used is the hydrological catchment.
global land surface divided into catchment excluding inland water & ice-covered area.
in each catchment, vertical soil moisture profile were determined based.
it incorporates meteorological forcing inputs that rely on observed data
o 2002-2006 AMSR-E forcing data are from GLDAS project (3-hourly time step at 2 degree and
2.5 degree resolution in latitude and longitude corrected using CMAP.
based on global atmospheric data assimilation system at the NASA GMAO.
•
o 1979-1987 SMMR forcing data based on ECMWF 15 years reanalysis at 6-hourly time steps
corrected using the monthly mean observations (precipitation, radiation, temperature, and
humidity data).
Ground-based measurements (in situ measurements)
o
o
USDA Soil Climate Analysis Network (SCAN) were used to validate AMSR-E (2002-2006).
Global Soil Moisture Data Bank (GSMDB) were used to validate SMMR (1979-1989).
39. Methods (cont)
●
10 soil moisture retrievals per month were available due
to power constraints of platform and swath width.
●
50 AMSR-E soil moisture retrievals data were available.
●
Satellite soil moistures are available for low-latitude
regions with little vegetation (Northern & Southern Africa,
and Australia).
●
Freezing and snow cover limits data availability, which
impacts yearly averages.
●
Data are not available for densely forested ecoregions
(South America, East Asia and temperate and boreal
forest of NA and Euroasia).
Source: Reichle et al., 2007
40. Results
•
•
•
•
•
Validation against situ data indicates that for both data sets soil moisture fields from the
assimilation are superior to either satellite or model data.
Global analysis reveals how changes in the model and observations error parameters may
enhance filter performance in future experiments.
For surface soil moisture anomalies, both satellite data show similar skill in reproducing
the corresponding in situ data, with R = 0.38 for AMSR-E and R = 0.32 for SMMR (based on
different algorithms).
The model estimates agree somewhat better than the satellite data with the in situ data
and that recent AMSR-E years are superior to that of SMMR historic period.
Time series improvements reveal statistically significant correlation with CI exceeding
99.99% (AMSR-E) for surface and root zone soil moisture, and 99.9% for surface (root
zone) soil moisture (SMMR).
41.
42. Conclusion
•
•
•
•
demonstrated that the assimilation of surface soil moisture retrievals from AMSR-E
into NASA Catchment land surface model to provide estimates of surface and root
zones soil moisture validated with in situ data.
compared AMSR-E and SMMR soil moisture retrievals found significant difference in
their climatologies.
AMSR-E retrievals are considerably drier and show less temporal variability than the
SMMR data (Figure 3 and 4).
global analysis of model produced by the data assimilation system can add value to Lband retrievals of soil moisture from the planned SMOS and Aquarius missions.
43. “Estimating profile soil moisture and
groundwater variations using GRACE and
Oklahoma Mesonet soil moisture data.”
Swenson et al., 2008
44. Purpose
•
•
•
to estimate time series of regional groundwater anomalies by combining terrestrial
water storage estimates from GRACE with in situ soil moisture observations from the
Oklahoma Mesonet with supplementary data from DOE’s Atmospheric Radiation
Measurement Network (DOE ARM).
develop an empirical scaling factor to assess soil moisture variability within the top
75cm sampled sites.
to provide efficient and effective mechanism to monitor and assess groundwater
resources both above and below the surface.
45. Methods
•
Water balance approach
•
Oklahoma Mesonet
•
GRACE
o estimate variations in groundwater averaged over a region centered on of
OK.
o collected real-time hydrometeorological observations > 100 stations.
o in situ soil moisture measurements were conducted every 30 mins at depth
of 5cm, 25cm, 60cm, and 75cm.
o estimate variations and data sets from > 100 stations were combined with
total water estimates from GRACE using a water balance equation (not
provided in text).
o combine time series of spatially averaged groundwater storage variations.
46. Methods (cont)
o GRACE
used the Released 4 (RL04) data produced by Centered for Space Research (CSR).
employed the post-processing technique to produce water storage estimates averaged over a
region of 280,000 square km.
o OK Mesonet
soil moisture detection sensors were added to 60 sites (2,25,60, and 75cm depth) and 43
stites at (2 and 25cm depth).
volumetric water content is determined from a soil water retention curve.
automated algorithm assessed the quality of soil moisture data.
o DOE ARM Network
Soil Water and Temperature System (SWAT) 21 sites to collect hourly profiles of soil
temperature and water at eight depths (0.05 to 1.75m) below the surface.
Average inter-site distance was about 75 km. 10 sites spanned the period 2002 to present.
47. Results
OVERVIEW
Overall, results are comparatively observed with well level data from a
larger surrounding region and the data reveals consistent phase and
relative inter-annual variability in relations to soil moisture estimates.
groundwater storage estimated from approximately 40 USGS well levels
in the region around OK, scaled weight of the GRACE averaging kernel in
Figure 1.
Over 40% of the variability in unsaturated zone water storage occurs
below the deepest OM sensors.
•
•
•
48. Results (cont)
Figure 3 shows the time
series of soil moisture
expressed as monthly
anomalies of volumetric
water content at each four
depths at which OKM
sensors are located.
Figure 4 shows the monthly
averaged soil moisture
anomalies expressed as
volumetric water content
with increasing phase lag
with depth.
Source: Swenson et al. 2008
49. Results (cont)
●
Figure 10 shows the best results
for confirming the upper panel.
The results compared two
groundwater estimates.
●
The well level groundwater (dark
gray line) confirms the general
characteristics of the regional
groundwater signal estimated as
a residual from GRACE (light
gray line). Both results show
similar seasonal cycle, and the
phases of the time series agree
well.
Source: Swenson et al. 2008
50. Conclusion
•
•
•
•
•
in view of the discrepancies that exist in both spatial and temporal sampling between the data
used to create two groundwater estimates, the overall agreement is good.
both time series illustrate similar interannual variability:
o relatively dry 2004 preceded by much wetter 2005 and 2003 signal lying b/w the other years.
smaller amplitude of well level-derived time series is not surprising, where signals separated by
larger distances are likely to be less well correlated.
which indicates that variations in both soil moisture and groundwater are well correlated at scales.
The correlation between month-to-month changes in the two times series may also indicate that
the method for estimating GRACE is pessimistic.
51. Bibliography
• Choi, M., Jacobs, J.M., and Bosch, D.D. (2008). Remote Sensing Observatory Validation of Surface Soil
Moisture using Advanced Microwave Scanning Radiometer E, Common Land Model, and Ground-based
Data: Case Study in SMEX03 Little River Region, Georgia, U.S. Water Resources Research, Vol. 44, pg. 1-14.
• de Jeu, A.M., Wagner, W., Holmes, T.R.H., Dolman, A.J., van de Giesen, N.C., and Friesen, J. (2008). Global
Soil Moisture Patterns Observed by Space Borne Microwaves Radiometers and Scatterometers. Survey
Geophysics, Vol. 29, pg. 399-420.
• Reichle, R.H., Koster, R.D., Lui, P., Mahanama, S.P.P., Njoku, E.G., and Owe, M., (2007). Comparison and
Assimilation of Global Soil Moisture Retrievals from the Advanced Microwave Scanning Radiometer for the
Earth Observing System (AMSR-E) and the Scanning Multichannel Microwave Radiometer (SMMR). Journal of
Geophysical Research, Vol. 112, pg. 1-14.
• Swenson, S., Famiglietti, J., Basara, J., and Wahr, J. (2008). Estimating Profile Soil Moisture and Groundwater
Variations using Gravity Recovery and Climate Experiment (GRACE) and Oklahoma Mesonet Soil Moisture
Data. Water Resource Research, Vol. 44, pg. 1-12.
• Syed, T.H., Famiglietti, J.S., Rodell, M., Chen, J., and Wilson, C.R. (2008). Analysis of Terrestrial Water
Storage Changes from Gravity Recovery and Climate Experiment (GRACE) and Global Land Data Assimilation
System (GLDAS). Water Resources Research, Vol. 44, pg. 1-15.