Time series of coarse resolution satellite imagery: some experiences and caveats

1
Time series of coarse resolution satellite
imagery some experiences and caveats
A contribution to the GLOBAL LAND COVER 2000

• Agustín Lobo
• alobo_at_ija.csic.es

2

• Classification of time series of Vegetation
  Index have produced vegetation charts at regional
  to global scales that are in general agreement
  with charts produced by compilation.

3
(No Transcript)
4
Summer-peaking vegetation
Spring-peaking vegetation
5
Summer-peaking vegetation
Spring-peaking vegetation
6
(No Transcript)
7
Summer-peaking vegetation
Spring-peaking vegetation
8
Summer-peaking vegetation
Spring-peaking vegetation
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12

• Single-date imagery would never produce such a
  result, no matter how many spectral bands would
  be considered...

13

• ...because time series of Vegetation Index are
  an estimate of the phenolgy of fPAR, which is a
  fundamental property of vegetation.

14

• Phenolgy of fPAR is shaped by climatic
  constraints (temperature and water availability)
  acting on the trade-offs of leaf maintenance,
  which implies that time series of Vegetation
  Index respond to climate.

15

• Nevertheless, there are two important
  shortcomings to be considered
• 1. Phenology is also fine-tuned by
  meteorological conditions, which implies that
  there is significant inter-annual variation and,
  therefore, mean annual series should be preferred
  for land-cover classification.

16
(No Transcript)
17
(No Transcript)
18

• 2. Although an important property, Phenology of
  fPAR is not enough to discriminate among some
  important land-cover types. Other properties
  should be measured from RS, such are

19

• Leaf/wood biomass ratio (woodiness)
• Leaf type and size
• Total biomass
• Height of dominant canopy
• Layering (vertical profile of leaf biomass)

20

• Leaf type and size
• Important to estimate the relative abundance of
  coastal conifers (i.e. P. halepensis) and
  evergreen oaks, because of their different
  behavior against fire. Note that vegetation
  changes due to increased aridity in the
  Mediterranean will be mediated through wildfires.
• RS Methods angular effects?
• Total biomass
• Height of dominant canopy
• Layering (vertical profile of leaf biomass)
• RS Methods Perhaps with SAR?

21

• Other important properties for vegetation
• functioning (but not for land-cover
• discrimination)
• Phenology of photosynthetic activity (PRI)
• Phenology of evapotranspiration
• Canopy roughness

22

• Leaf/wood biomass ratio
• Important for the C budget, to estimate fuel
  load, and for land-cover
• identification.
• Methods SWIR ?

23

• In High resolution imagery, SWIR has been found
  to be important to discriminate forest types

CAN-1 () CAN-2 () green 0.0435
(16.4) - 0.0448 (14.3) red 0.0915 (34.5)
0.0190 ( 6.1) nir -0.0447 (16.8) -
0.2027 (64.6) swir 0.0855 (32.2) 0.0472
(15.0)
24

• Using VEGETATION imagery, SWIR has been found
  also to be important to discriminate land cover
  types

burned forest
forest
arid vegetation
irrigated fields
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
Pastures
30
Cistus and broom dry garrigues
31
Quercus ilex forest and shrubland
32
Quercus rotundifolia shrubland
33
P. halepensis with Q. rotundifolia understorey
34
P. uncinata mountain forests
35

• Therefore, although there is some sensitivity to
  the wood/leaf ratio, the eventual presence of
  water complicates the problem.

36
P-OVNI, S-NDVI, S-OVNI?
21 images 1999-09-16 to 1999-12-13 Burkina-Faso (
Lobo Bartholome)
37

• S-NDVI

P-OVNI
38

• S-NDVI

P-OVNI
39
(No Transcript)
40
(No Transcript)