measuring soil surface water content using full- wave ... · data inversion presents considerable...
TRANSCRIPT
Measuring soil surface water content using full- wave inversion of off-ground GPR data
Sébastien Lambot(1,2), Evert Slob(3)
Diana Chavarro(4), Maciek Lubczynski(4)
(1)Université
catholique de Louvain (Belgium)(2)Forschungszentrum Jülich (Germany)
(3)Delft
University of Technology (The Netherlands)(4)ITC, The Netherlands
Sustainable and optimal management of soil and water resources is required to preserve our environment and sustain increasing societal demands
Human activities, but also natural processes, lead to soil degradation worldwide:erosion, compaction, pollution, desertification, sealing, salinization, nutrient depletion
Surface water content is a key hydrological variable, controlling infiltration and runoff, evaporation, and energy exchanges between the soil and the atmoshere.
Management strategies and decision support
Observation: characterizing and monitoring the environment
Process understanding and modeling
GPR is promising to characterize surface water content at the field scale
InnaccessibilityHeterogeneity
Electromagnetic contrasts
GPR basic principles
GPR operates by transmitting microwave electromagnetic energy down into the ground through an antenna. The transmitted energy is reflected from various electromagnetic interfaces. An antenna then receives the reflected signal.
• Dielectric permittivity ε
• Electric conductivity σ
• Magnetic permeability µ
wave velocity
Tx Rx
ε1
, σ1
, µ1
Radar system
electronics
ε2
, σ2
, µ2
ε3
, σ3
, µ3
wave attenuation
reflection and refraction
Common GPR surface ground wave
txv =
2
⎟⎠⎞
⎜⎝⎛=
vc
rε
• GPR contact with the soil• Identification of the ground wave• Large antenna separation• Strong EM simplifying assumptions
Limitations
Common GPR surface reflection coefficient method
with
⇔
ε1
ε2
ε1
PECσ→∞
• Reference PEC measurements, at same h!• Strong EM symplifying assumptions
Limitations
Ei Ei EsEs
Advanced GPR forward and inverse modeling
• Parameter retrieval capabilities are limited
• Error in the estimates are inherently introduced
• Expert user is needed and techniques may be cumbersome
Common techniques
• SFCW radar with a VNA
→ UWB, international standard
• Directive horn antenna
→ to be used off the ground
• Monostatic
mode
→ accurate and efficient forward modeling
Low-cost and handheld radar
GPR system
• 3-D
multilayered
medium
• Dirac electric source/receiver
Exact solution of Maxwell’s equations
3D Green’s function
Inverse problem for the particular case of surface dielectric permittivity retrieval
tmin
h
εσ ≅ 0
tmax
zmax
with
Surface roughness: hmax
<λ/8
(Lambot et al., WRR, 2006b)
(Lambot et al., WRR, 2006a)
min
Advantages compared to the ground-wave or common reflection methods
Issues to take care of
• Effect of surface roughness (hmax
<λ/8)
• Effect of vegetation
• VNA measurements are an international standard
• 3-D
solution of Maxwell’s equations
• Antenna effects are filtered
• No time-zero and reflection time issues
• Reference PEC measurements are not needed
• Antenna height should not be known
• No need to identify the ground wave using multiple measurements
• Possibility to include effects of soil layering and high σ
• Fully automated
Water content
dry
Saturated
Aquiferex project: Tunisia
Measurements in different irrigated plots (Gabes):
36 measurements over 17 plots
GPRSoil
specific
model
TDR
Results taking into account electric conductivity during inversion of GPR data
r=0.94Soil
specific
modelr=0.93
→ Microvariability→ Measurement scale→ Surface roughness
(gravimetric) (gravimetric)
1-2 GHz 0.1-0.9 GHz
Extending the frequency range towards lower frequencies is required to estimate σ
z θ
x
Surface GPR TDR
Sample Remote sensingMeasurement scales
• Gravimetric
• Theta-probe
• Off-ground GPR
Transect
Olive trees
Measurements over a transect in Ben Gardane:
65
measurements
Dielectric permittivity and water content over the transect
0
0,01
0,02
0,03
0,04
0,05
0,06
0,07
0,08
0 10 20 30 40 50 60 70
Measurement
Wat
er c
onte
nt
0,00
1,00
2,00
3,00
4,00
5,00
6,00
0 10 20 30 40 50 60 70
Measurement
Die
lect
ric c
onst
ant 687119mE 3672243mN
688533mE 3670591mN
• Dry
• Small variation range
• Some spatial correlation
Observations
Integrated hydrogeophysical inversion for vertical profile reconstruction
EM model
Modeled signals
Measured signalsError
function
Optimization algorithm
Dielectric and electric profiles
G↑xx
(f,t)
ε(z,t)σ(z,t)
z
θ
t (min)
BC
∞
unknowns
EM model
Modeled signals
Measured signalsError
function
Optimization algorithm
Hydraulic properties
Hydrodynamic model
Modeled water content
Petrophysical relationships
Dielectric and electric profiles
0 20 40 60 80 100 120 140 160 180 200
0.15
0.2
0.25
0.3
Wat
er c
onte
nt (m
3 m-3
)
Time (min)
G↑xx
(f,t)
ε(z,t)σ(z,t)
θ(z,t)
?
P Time lapse
Constraint
(Lambot et al., GRL, 2006)
Integrated hydrogeophysical inversion
Simulated infiltration in a sandy soil
Hydrodynamic model configuration
• Hydraulic properties: Mualem-van Genuchten’s
model
0 20 40 60 80 100 120 140 160 180 200
0.15
0.2
0.25
0.3
Wat
er c
onte
nt (m
3 m-3
)
Time (min)
θr
= 0.00θs
= 0.357Ks = 8.24 cm/minα
= 0.0278 cm-1
n = 7.02λ
= 2.89
• 1-D Richard’s equation: WAVE
Precipitation
Water table
30 cmSimulation domain
• Boundary conditions unknowns
Initial condition:Hydrostatic equilibrium
(Lambot et al., WRR, 2002)
Water content profiles as f(t) Pedotransfer functions
Model of Ledieu
Model of Rhoades
2
⎥⎦⎤
⎢⎣⎡ −
=a
br
θε
( ) swba σσθθσ +′+′= 2
Electromagnetic model configuration and radar data
ε(z,t)σ(z,t)
S0.15 m
Δz=0.005 m
0.30 m
⎟⎠⎞
⎜⎝⎛
10minλ
(Thanks to Oliver Bucker, ZAM)
t
(min) t
(min)
JUMP: 1312 CPUs, 8.9 TFLOPSJUBL: 16384 CPUs, 45.8 TFLOPS
Objective function
( )2* ),(),(∑∑ ↑↑ −=Φ
f txxxx tftf GGb
Least squares
Optimization algorithm
Global multilevel
coordinate
search
(GMCS) algorithm
+
Nelder-Mead simplex (NMS) algorithm
91 frequencies
25 observation times
t
(min)
θr
= 0.00θs
= 0.357Ks = 8.24 cm/minα
= 0.0278 cm-1
n = 7.02λ
= 2.89
unknown
known
Successful
(~450 iterations)
θr
= 0.00θs
= 0.357Ks = 8.24 cm/minα
= 0.0278 cm-1
n = 7.02λ
= 2.89
unknown
known
Successful
(~800 iterations, ~8 hours, 16 CPUs)
3p 4p
Enough information is contained in the time-lapse GPR measurements
Optimization results
Parameter space constraints
A priori knowledge,
other sources of information
• UWB, dynamic range• EMI, seismic/acoustic
• BC• IC• Structure Computation
time
Petrophysical parameters, σw
A priori knowledge
Spatial correlation
Very
fine
Very
fineFine
Fine
Medium fine
Medium fineMedium
Medium
Coarse
Coarse
Estimated
Estimated
Water retention curve Hydraulic conductivity function
(HYPRES)
•
Full-wave
GPR
data inversion presents considerable theoretical and practical advantages for surface water content mapping
•
The proposed method is operational and fully automated
•
We are exploring methods to extend its applicability in high conductive and strongly layered soils, and better define characterization scale
Soil
moisture
mapping
and monitoringSoil
hydraulic
properties
mapping
Conclusions and perspectives
References
-Lambot, S., E.C. Slob, I. van den Bosch, B. Stockbroeckx, and M. Vanclooster, Modeling
of ground-penetrating
radar for accurate
characterization
of subsurface
electric
properties, IEEE Transactions on Geoscience
and Remote
Sensing, 42, 2555-2568, 2004.
-
Lambot, S., L. Weihermüller, J.A. Huisman, H. Vereecken, M. Vanclooster, and E.C. Slob, Analysis
of air-launched
ground-penetrating
radar techniques to measure
the soil
surface water content, Water Resources
Research, 42, W11403, doi10.1029/2006WR005097, 2006.
-
S. Lambot, E. C. Slob, M. Vanclooster, and H. Vereecken, "Closed loop GPR data inversion for soil hydraulic and electric property determination," Geophysical Research Letters, vol. 33, pp. L21405, doi:10.1029/2006GL027906, 2006.