Post on 25-Jan-2021
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
On the retrieval of soil moisture content over agricultural sites using
L band SAR data
Francesco Mattia(1), Giuseppe Satalino(1), Anna Balenzano(1), Malcolm Davidson(2)
(1) CNR-ISSIA, Bari, Italy
(2) ESA ESTEC, Noordwijk, The Netherlands
This work was partially funded by ESA ESTEC under contracts 19558/06/NL/HE and by MI-PAF under contract 209/7393/05.
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
develop and assess a retrieval algorithm estimating soil moisture over agricultural sites from L band SAR data
Objective
Application context: agriculture and/or hydrology
Outline
Experimental data: Terra SAR SIM 03 campaign
The retrieval algorithm: direct and inverse approach
Results
Conclusions and future work
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
TerraSARSIM campaign: 2-25 April, 2003(ESA project n. 16736/02/NL/FF)
• Barrax, La Mancha, Spain• Mediterranean climate• Average yearly rainfall: 400 mm• Flat morphology• Agricultural area• 65% dry land, 35% irrigated land
SAR data (5 Flights)
Sensor: E-SAR onboard DLR Dornier Frequency: C-, L- and X-bandIncidence angle: 26° - 56°Doy: 92, 92, 94, 94, 115
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
In situ dataN° of monitored fields: 28Land use: 7 irr. wheat, 4 irr. barley, 5 irr. alfalfa, 12 not-irr. barleyData:
gravimetric and volumetric soil moisture at 0-5 cm (some problems!)qualitative information on surface roughnessfresh and dry biomass & crop height
Selected 12 cereal fields: 8 non irr. barley; 2 irr. barely; 2 irr. wheat
0,0
0,5
1,0
1,5
2,0
2,5
90 95 100 105 110 115 120
Dates of SAR acquisitions (Doy)
Fres
h bi
omas
s (K
g/m
2)
0,0
5,0
10,0
15,0
20,0
25,0
30,0
90 95 100 105 110 115 120
Dates of SAR acquisitions (Doy)
Vol.
moi
stur
e co
nten
t (%
)
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
a priori estimates of surface and vegetation parametersˆ jp
direct model
( ) ( )0 1,...ˆ , , j M ii pσ θ λ ε== +iF total model & measurement error
backscattering
Maximum likelihood solution for parameters
( ) ( )( ) ( )
2 20 1..
2 20
, , ˆ1 1i j M j jii j
i j
p p pC
N M p
σ θ λ
σ=− −
= +∆ ∆
∑ ∑F
error on a priori estimates
total rms error
( ) 20 iiσ ε∆ =
jp
Retrieval algorithm
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
( ) ( ) 2 sec( )0 0ˆ ˆ i hSi i eτ θσ σ −=
h
, ,..i HH VV=
( )0 , , , .,...w v dg m f strε ρ′′ = ( )vv vb W≈ vW [Kg/m3]2 2
VV VV s VV sh b W b Fπ πτλ λ
≈ ∝ sW [Kg/m2]
sF [Kg/m2]
(Veget. Water cont.)/m2
(Fresh biomass )
( )0ˆ S iσ Integral Equation Model (IEM)2
VV oπτ ελ
′′∝
Direct model
rε
An anisotropic uniaxialdielectric slab with an effective permittivity
θz
x
0 00 00 0
e er
e e
o o
jj
j
ε εε ε ε
ε ε
′ ′′+⎛ ⎞⎜ ⎟′ ′′= +⎜ ⎟⎜ ⎟′ ′′+⎝ ⎠
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
Free parameterssoil:
surface height rms (s), surface corr. length (l) and soil moisture (mv)
vegetation:
fresh biomass (F), structural parameters bhh & bvv
Multiple regression: ∆σ0s & ∆F Updating guess values
Temporal updating: 2 SAR acquisitions
σ0t =σ0s e(-g ·F)∆σ0t/ σ0t=∆σ0s/ σ0s – g ∆F
1st SAR date: s, l, mv, F, bhh, bvv
measured estimated on the 1st date
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
SAR data used in the study
L band HH & VV SAR data acquired on:
- April 2, 2003 (flight n.1, in the morning)
- April 4, 2003 (flight n.3, in the morning)
- April 25, 2003 (flight n.5, in the morning)
The same guess values obtained as averages of in situ measurements
guess values obtained by temporal updating
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
0
5
10
15
20
25
30
0,0 0,5 1,0 1,5 2,0
Fresh biomass (Kg/m2)
Vol.
moi
stur
e co
nt.
(%)
measured
retrieved: no attenuation
retrieved: with attenuation
Retrieved soil moisture: April 2 & 4
: mean =17 %
: mean=9.4 %, rms=9%
: mean =18.7 %, rms=4.2 %
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
-0,5
0,0
0,5
1,0
1,5
2,0
2,5
10 12 14 16 18 20 22 24
vol. moisture content (%)
Fres
h bi
omas
s (K
g/m
2)
measured
retrieved
Retrieved biomass: April 2 & 4
: mean = 0.8 Kg/m2
: mean = 1 Kg/m2, rms=0.4 Kg/m2
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
Biomass updating: April 4 & 25
0,0
0,5
1,0
1,5
2,0
2,5
1 2 3 4 5 6 7 8 9 10 11 12Selected Fields
Incr
emen
ts o
f Fre
sh b
iom
. (K
g/m
2)
true increments of Fresh biomass
estimated " "
: mean = 0.9 Kg/m2
: mean = 1.3 Kg/m2, rms= 0.64 Kg/m2
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
10
15
20
25
30
1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4
Fresh biomass (kg/m2)
Vol.
soil
moi
stur
e (%
)
measured
retrieved (updated guess values)
Retrieved soil moisture: April 25
: mean =16.4 %
: mean=19.6 %, rms=4.6%
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
1,0
1,5
2,0
2,5
3,0
10 12 14 16 18 20 22 24
Vol. soil moisture (%)
Fres
h bi
omas
s (k
g/m
2)
measured
retrieved (updated guess values)
Retrieved biomass: April 25
: mean =1.7 Kg/m2
: mean=1.8 Kg/m2, rms=0.4 Kg/m2
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
Vol. Soil moist. Content (%)3 3015
Soil moisture map: April 25
POLinSAR 2007, 22-26 January 2007, ESRIN Frascati, Italy
Conclusions
Future worksystematic assessment of the algorithm accuracy through
simulation studies;
application of the algorithm to other data sets and extend the approach to other crops
soil moisture and vegetation biomass maps of cereal fieldshave been retrieved, with accuracies of approximately 5 % and 0.5 Kg/m2, respectively, by inverting L band SAR data and using a priori information;
a technique to update a priori information on surface parameters using multi-temporal SAR data has been proposed and tested on the Barrax test site;