Aucun titre de diapositive

1
Processing methodology for full exploitation of
daily VEGETATION data
C. Vancutsem, P. Defourny and P.
Bogaert Environmetry and Geomatics
(ENGE) Department of Environmental Sciences
and Land Use Planning UCL Université
Catholique de Louvain BELGIUM
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
2
Objective
Develop an operational compositing strategy to
produce spatially and temporally consistent
images over large areas
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
3
State of issue

• Selection of specific angle configurations and
  atmospheric conditions

• Use only 10 of the information (ten-days
  compositing)

• Requires stable land cover time series with low
  cloud frequency

• Requires the on-line use of a large archive of
  daily data

Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
4
Mean compositing strategy

• Robust and simple compositing

• The most stable parameter of a distribution

• Use of all the available information

The mean NDVI already suggested with AVHRR
simulations (Meyer et al., 1995)
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
5
Mean Compositing Strategy
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
6
Prerequisites of the methodology
1) Good superposition of the daily images
Multitemporal location error lt 500m
(http//vegetation.cnes.fr/userguide/userguide.ht
m)
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
7
Prerequisites of the methodology
2) Efficient cloud screening
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
8
Prerequisites of the methodology
3) VZA 5-days cycles
Compensation between backward and forward angles
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
9
Spatial consistency
Nigeria, first decade of november 99, (Red, Nir,
Mir)
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
10
Spatial consistency
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
11
Spatial consistency
West Africa , first decade of november 99 (Mir,
Nir, Red)
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
12
Spatial consistency
Mean composite
West Africa , first decade of november 99 (Mir,
Nir, Red)
Max NDVI composite
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
13
Spatial consistency
Mean composite
West Africa , first decade of november 99 (Mir,
Nir, Red)
Max NDVI composite
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
14
Spatial consistency
Mali , first decade of november 99
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
15
Temporal consistency
NDVI channel
MVC Mean
Tanzania, year 2000
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
16
Temporal consistency
MVC Mean Observations number
Tanzania, year 2000
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
17
Reduction of daily variations
Tanzania , year 2000
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
18
Conclusions

• Robust and operational approach
• Large spatial and temporal consistency of the
  results
• Use all the available information
• Low sensibility to BRDF effects
• Accessibility for all users
• Flexible methodology

Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
19
Two applications
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
20
For more information VANCUTSEM, C., BOGAERT,
P., DEFOURNY, P., 2002, Mean compositing strategy
as an operational temporal synthesis for high
temporal resolution, IJRS in press. Contact
Vancutsem_at_enge.ucl.ac.be
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
21
Rush
Department of Environmental Sciences and Land Use
Planning - UCL
GLC 2000, 18 22 march
2002
22
Temporal consistency
vancutsc For each pixel, selection of the signal
providing the maximum NDVI value amongst the last
10-day acquisitions At the end of 2001 the
processing chain will be improved as the cloud
screening The compositing strategy is not
validated for spectral bands !!!
Mir channel
MVC Mean
Tanzania, year 2000
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
23
vancutsc For each pixel, selection of the signal
providing the maximum NDVI value amongst the last
10-day acquisitions At the end of 2001 the
processing chain will be improved as the cloud
screening The compositing strategy is not
validated for spectral bands !!!
Speckle effect
MVC NDVI composite
Mean composite
Nigeria , first decade of november 99
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
24
Frequency of compositing
vancutsc The current compositing technique for
VEGETATION data (VGT-S10 product) shows
radiometric artefacts in the reflective bands
that may cause a significant noise for subsequent
retrievals of surface parameters. The
performances of various compositing strategies
are assessed as well for the reflective bands as
for the NDVI composites. Dedicated indicators
and statistical analysis are computed to provide
quantitative results by zone and by band. The
artefacts we can see on S1 products are more
visible on S10 products
Sliding window
Department of Environmental Sciences and Land Use
Planning - UCL GLC 2000, 18 22
march 2002
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)