International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
&€spor (d) accounts for geometric errors of the SPOT data, that
is a geometric bias for the exterior orientation and the
residuals of the inner ortientation :
ESPOT (d)= BIAS+ £iNNER(d)
BIAS is constant within a SPOTS image
The inner orientation error EINNER(d) only depends on the
detector number d.
For each image matching measurement we have an estimation
Of EINNER(d) :
EINNER(d)\(,p,fref) + BIAS = dm(l,p,Iref) — Ecor(l.p) -
Erer(Iref) — EDEM(I.p) — EpsF(l,p)
The same detector views a landscape 5 km wide (the width of
our super site), that is 500 to 1000 rows depending on the
mode to be processed, each pixel of the landscape being
viewed in turn by 4 to 5 images of the reference data set. The
image matching measurements was done with a density of one
measurement every 5 rows.
Following is an example of the scattering of measures for a
given HRGI panchromatic detector ; in this example, 418
measures made over 20 different images of the reference
dataset showed correlations better than 0.97. After filtering at
2 sigmas, the 1 sigma ellipse of confidence is a circle with
radius 0.1 pixel.
MEASURES FOR SPOTS HRG1 Pan, detector 4414
24 T T T T T T T T T
measure with couælation higher than 0.97 =
313 confidence ellipse
€
&
)*
pixels along track
o
i
1.4 1 i L i A i | L i
22% 20 3 X] 132 33 a4 35 a6 37
pixels across track
4 -THE SPOTS DATA
4.1 SPOTS instruments
In this paragraph we briefly review the main geometric
characteristics of SPOTS payload.
SPOTS high resolution payload is composed of 2 HRGs
instruments, with 60 km wide swath, 5 meters resolution in
panchromatic standard mode called HMA and 10 metres
multispectral mode called XI. Since registration between
spectral bands is measured with relative methods, it is not
treated in this paper and for multispectral mode, only B2 mode
(closest to panchromatic in radiometry) is concerned here.
The payload is complemented with the stereoscopic HRS
instrument built with 2 similar telescopes : forward and
bakward HRS, with 120 km wide swath, 5 metres resolution
along track and 10 m cross track looking 20° forward and
backward |Baudoin 2004].
4.2 Data acquisition :
During the summer of 2002, the supersite has been imaged at
least once with each instrument and in each of the modes
described previously. The 120 km swath of HRS is covered in
2 steps using 2 tracks at 5 days interval. As far as it was
possible, all the data were acquired with almost nadir viewing
conditions.
The following table shows the data used for inner orientation
of SPOTS instruments. An assembly map completes the
description of the data.
9 i
28/05/02 X x ox
18/06/02 | X x X A
23/06/02 X Le Monnet
19/07/02 x SR T
14/08/02 S
MANOSQUE SUPERSITE - INITIAL REFERENCE DATASET 1997
INDEX AND SPOTS SEGMENTS SWATH
44.00
f
/
/
asso f À
j
4
/
#
#
43.00 — d
Reference dat
HRG 18/06/02
HRS 18/06/02
HRS 14/08/07
42.50 1 i 1
e o > 5 =
o v a 2 S
Us v xS = =
À control acquistion at one year interval was planed in july 03
5 -RESULTS
5.1 HRS Inner orientation
Along track, a major third degree polynomial distorsion with
an amplitude of 5 pixels was observed with the reference
dataset within the first month of in-flight commissioning on
both HRS instruments and included into the ancillary data (so
called PSTS), thus improving the quality of the geometric
model.
The availability of the new dataset, allowed to observe the
residual inner orientation. Following is a figure representing,
in pixels, the residual inner orientation for both instruments
along and across track.
Internat
oooc
555b
pixel
4d. trm Ota ed
pixel
o555 oooo
Uu t2—0—t125 5
Residu:
3.5%10
degree |
into the
as follo
Along t
Across
most of
set as d
the hig
angle, |
5.2 HR
Main 1
HRGs
pixel. /
PSTS, |
and 0.0
The fol
polynoi
pixel
o
ixe
eo
