Tropospheric delay in
InSAR data
Heresh Fattahi
(Jet Propulsion Laboratory, Caltech)
Aug 2021
InSAR (Interferometric Synthetic Aperture Radar )
Before Earthquake After
Earthquake
ϕ (t A ) ϕ (t B )
δϕ (t A , t B )
λ
δr = δϕ
4π
interferogram
δϕ = δϕ dis + δϕ atm + δϕ geometry + δϕ decor
Atmospheric delay
acquisition 2ndacquisition
Index of refraction
c ! n =1 vacuum
d = vt v= "
n
# n >1 atmosphere
[Modified from Z. Li]
ti tj
Atmospheric delay
Atmospheric delay for double
difference InSAR measurements r
between two pixels (p and q) and
P q
between acquisition times ti and tj:
ti ,t j # rp
= % ∫ N(r, t j )dr −
rp & # rq rq
N(r, ti )dr &(
δL pq $ 0 ∫ 0
N(r, ti )dr ( − % ∫ N(r, t j )dr −
' $ 0 ∫ 0 '
With r the range from the radar to the target and N the refractivity
Pd e e ne
N = K1 + K 2 + K 3 2 + K 4Wcl + K 5 2
T T T f
Troposphere Cloud Ionosphere
Pd: dry air partial pressure, e: water vapor partial pressure, T: atmospheric temperature,
Wcl :liquid water content [kg/m3] , ne: electron number density/m3, f: radar frequency
Methods to correct the Tropospheric delay in InSAR data
Estimated from InSAR Predicted delay using
data independent data
Empirical Satellite Atmospheric
Filtering GPS
approach Spectrometers models
Stacking Phase-elevation
(linear or non-linear) GPS Reanalysis
MERIS
Temporal-Spatial only models
Filtering Phase-elevation
(moving window)
MODIS Forecast
Common scene
models
interferogram Phase-elevation GPS +
combination (multi-resolution Atmospheric
Wavelet based) models
Filtering (given
Statistical properties of atmosphere)
Correction using MERIS
InSAR MERIS (wet delay) corrected InSAR
The interferogram is formed from ASAR data onboard Envisat which also was carrying a
spectrometer called MERIS
Correction using MERIS
InSAR MERIS (wet delay) corrected InSAR
The interferogram is formed from ASAR data onboard Envisat satellite which also was
carrying a spectrometer called MERIS
Troposphere correction with MERIS and ERA-I
Interferogram MERIS-corrected ERA-I - corrected
MERIS more
accurate than
atmospheric
models.
Limitations:
Requires daylight,
limited by cloud and
not available at SAR
acquisition time
except for Envisat
data.
Troposphere correction with Empirical approach
Caused by changes in tropospheric layering – correlates with topography (wet & dry)
Phase-elevation correlation can be estimated
DEM Interferogram
Differences in tropospheric layering cause
delays correlating with topography.
Ebmerier et al., 2013 Empirical estimation across a portion or the whole scene
Tropospheric delay correction with Atmospheric model (ERA-I)
Interferogram Ddtropo = Dddry + Ddwet
Dddry << Ddwet as expected
[Jolivet et al, 2014]
Troposphere correction with Atmospheric model (different models)
Interferogram NARR ERA-I (ECMWF) MERRA
3 hours 0.7° grid 0.7° long, 0.5 lat
temporal 6 hours, 6 hours
spacing 37 levels 42 levels
NARR performs best in this case [Jolivet et al, 2014]
Tropospheric delay has seasonal variations
Blue and Red circles: InSAR time-series
Gray dots : predicted tropospheric delay
The atmospheric models seem to successfully predict the seasonal delay while the
improvement in the stochastic component may be minimal.
[Fattahi & Amelung, 2015]
Amplitude of the Seasonal delay across US
Smaller seasonal GPS, MODIS and
variation in ERA-I are
western US consistent for the
seasonal delay
GPS delay
(Unavco)
MODIS
(NASA-JPL)
ERA-I
(ECMWF
and pyaps)
[Fattahi, 2015]
Tropospheric delay correction with atmospheric models accounts for
systematic seasonal delay and reduces the standard deviation of the
stochastic components by 10-20%.
[Fattahi & Amelung, 2015]
References:
Fattahi & Amelung, InSAR bias and uncertainty due to the systematic and
stochastic tropospheric delay, JGR-Soild Earth, 2015.
https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2015JB012419
Jolivet, R., P. S. Agram, N. Y. Lin, M. Simons, M. Doin, G. Peltzer, and Z. Li (2014),
Improving InSAR geodesy using Global Atmospheric Models, J. Geophys. Res.
Solid Earth, 119, 2324– 2341, doi:10.1002/2013JB010588.
