In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Voi. XXXVIII, Part 7B
where i = y/—l.
2.2 Phase delay computing
It was shown in (Oguchi, 1983) that for the case of coherent prop
agation in rain the propagation phase delay, 5 in mm/km, can be
calculated as:
Ö = 10
-a 2-7T
k 2
/
Re(f(D))N(D)dD,
(6)
where k is the wave number (corresponding to a wavelength of 56
mm for the case of ERS1/2) and f(D) is the forward scattering
amplitude in mm and N(D) is in mm^'m" 3 .
In this case f(D) can be evaluated using Rayleigh approximation
(Bringi and Chandrasekar, 2001). The scattered amplitude for h-
polarized incidence wave can be then calculated as:
Figure 1 : Fitting Rayleigh approximation to a powerless one.
}(D) = tl) 3( e - 1)A , (7)
where e is the relative permittivity of water and A is given by:
1
A =
1 + Ai (e — 1)
(8)
where
A = and A 2 = 1 - jfan -1 (/)) , (9)
where
/ = e -2 — 1. (10)
As we can conclude the expression of Re(f(D)) is very com
plicated and cannot be used directly for deriving an analytical
expression for the phase delay related to the rain. Thus more
suitable expressions are needed.
In order to overcome this problem, we have simulated the for
ward scattering amplitude for D ranges between 1 and 8 mm. we
have fixed the temperature at 10 oc and the wavelength at 5.6 cm.
In these conditions, the value of e can be approximated by 70 +
30j. Then, for a given value of D, the corresponding axis ratio e is
computed using equation 2. Then, f{D) is computed using equa
tions 3 and 5-11. From the resulting curve of the real part we can
derive a powerless relation between Re(/(Z))) and D that have
the forme aD 13 . Coefficients a and /3 are estimated by the non
linear regression algorithm. For our simulation a = 3.7 x 10~ 4
and /3 = 3.02. Figure 1 shows the good agreement between the
Rayleigh approximation and the powerless one.
After replacing Re(/(D)) by its powerless expression and per
forming some simple mathematical manipulations equation 6 be
comes:
6 = J D 13 exp(—-D)cfD
, (11)
=
where T(-) is the gamma function.
Finally a simple expression relating the phase delay to the rain
rate can be derived using equation 4 as:
6 =
8\ 2 a4.1~ ß R°- 21ß r
(12)
mm, since particles with a larger diameter are unstable. In Figure
2 the result of these calculations is given. It can be seen that
for high rain intensities the path delay can be as high as several
centimeters.
Figure 2: Path delay as a function of the rain rate.
3 SIMULATED INTERFEROGRAMS
Our simulation for this work consists in generating two interfero-
grams for a perfectly flat area. The first one was generated with
out taking into consideration any rain effects and the second one
takes into a consideration a layer of rain that covers the imaged
area. Figure 3 and 4 show the considered computation geometry.
3.1 Computing geometry
In figure 3, designates the baseline vector, 0 the incidence
angle, T the radar swath and n and r 2 are ranges between a
fixed point in the imaged area and the position of the radar for
the first and the second acquisition respectively. In the absence of
rain ATp P rop depends only in n and r 2 (Abdelfattah and Nicolas,
2002):
4tt
Alpprop = ^“( r 2 - »*l). (13)
Thus, rain free interferogram can be directly generated using n
and r 2 . If we consider the configuration of figure 3 and after
performing some geometrical manipulations we can deduce that:
The path delay for different rain intensities were calculated. It
should be noted that raindrop diameter is usually truncated at 8
ri = yjx 2 + y 2 + H 2
r 2 = y/(x - B x ) 2 + (y - B y ) 2 + (H + B z ) 2