IEEE Geoscience and Remote Sensing Magazine - June 2015 - 35

on the data provider. Interoperability requires completeness of metadata;
◗ The processing level is not harmonized, in the same
dataset, for some product types additional pre-processing stages on the user side are needed, for applying
terrain corrected-processing and atmospheric corrections for instance;
◗ The specific processing of each data provider makes it
difficult to compare the same quality parameter for different missions, as for instance in case of cloud cover;
◗ The lack of traceability regarding the reference data used,
results in a dramatic planimetric accuracy loss.
The exhaustive validation of all products included in a
dataset is not in the scope of the CQC. It is operationally
not feasible. For instance in case of URBAN ATLAS, the
check of geometric accuracy would require very accurate
reference data (0.5 m RMS) for each covered urban zone.
However, the geometric accuracy of URBAN ATLAS data
produced by using IMAGE2006 as reference data, itself
based on IMAGE2000 and USGS Land Cover reference
datasets, [14], might be overestimated from an absolute accuracy point of view.
More generally, regarding the geometric processing, the
production of a terrain corrected product is basically done

as a combination of three subsequent processing stages. In
the first stage, supporting data are collected; Ground Control Points are selected from
a raster reference dataset for
planimetric information toThe band To band
gether with terrain elevation
regisTraTion accuracy
information (Digital Elevahas a signiFicanT impacT
tion Model, DEM), mission
specific parameters (orbit,
on The reliabiliTy oF
attitude, camera model) are
geophysical measures,
repatriated. The second stage
and also on The
uses supporting data for
classiFicaTion resulTs.
image geo-referencing and
The undersTanding oF
optionally viewing model
accuracy measures and
calibration. The last stage apvalidaTion meThods by
plies the geometric model to
all parTies is crucial For
drive the image geo-coding
process. Errors may come
The mosT appropriaTe
at each processing stage, in
use oF daTa.
particular, because of inaccurate supporting data and
also an inappropriate georeferencing model. However, in most cases, in the product
metadata, the supporting data (version, accuracy) are not

MODIS Band 3 (Libya 4), RE Band 1, IK02 Band 1

1
0.8
TOA Reflectance

TOA Reflectance

0.8

MODIS Band 4 (Libya 4), RE Band 2, IK02 Band 2

0.6
0.4
0.2
0
2000

3000

4000

5000

0.6
0.4
0.2
0
2000

6000

Epoch (2008): Days Since 1/1/2008

MODIS Band 1 (Libya 4), RE Band 3, IK02 Band 3

4000

5000

6000

MODIS Band 2 (Libya 4), RE Band 5, IK02 Band 4

0.8
TOA Reflectance

3000

Epoch (2008): Days Since 1/1/2008

0.6
0.4
0.2
0
2000

3000

4000

5000

6000

Epoch (2008): Days Since 1/1/2008

0.6
0.4
0.2
0
2000

3000

4000

5000

6000

Epoch (2008): Days Since 1/1/2008

Figure 5. Libya-4 site data - The radiometric co-registration of working data with reference data is assessed by using visual inspection of
calibration trend and using a statistical approach. The stability of radiometric calibration is demonstrated.
june 2015

ieee Geoscience and remote sensing magazine

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2015