Ongoing Projects

1
Russian Academy of Sciences Space Research
Institute
Monitoring and understanding of changing
environment using Earth observation satellite
data and informational technologies RD at RAS
Space Research Institute Status Overview
Sergey Bartalev, Evgeny Loupian and Igor Savin
2
Primary Interacting Components within the Earth
System
3
Blue or green planet ?
Earth surface covered by vegetation 22,5
Global average Leaf Area Index x 4,5
Global Leaf Area 101,25 100
- IGBP DIS Land Cover Data Base, 2000 -
Gregory P. Asner et al. 2002, Global Synthesis of
Leaf Area Index Observations Implications for
Ecological and Remote Sensing Studies
Water is about 70,8 of the Earth surface
4
Importance to Study Terrestrial Ecosystems
Dynamics

• Climate change science
• Sustainable natural resources management
• Forest management
• Agriculture production
• International Environmental Conventions
• UN FCCC (1992) and Kyoto protocol (1997)
• UN Convention on Biological Diversity (1992)
• UN International Convention to Combat
  Desertification (1994)
• Convention on Wetlands of International
  Importance (1971)
• FLEGT/FLEG - Forest Law Enforcement, Governance
  and Trade (2003)

5
Global Climate Change
http//www.scar.org/news/newsletters/issues2008/ma
r08.html
The year 2007 tied for second warmest in the
period of instrumental data, behind the record
warmth of 2005. Figure on left shows 2007
temperature anomalies relative to the 1951-1980
base period mean. The global mean temperature
anomaly, 0.57C warmer than the 1951-1980 mean,
continues the strong warming trend of the past
thirty years that has been confidently attributed
to the effect of increasing human-made greenhouse
gases. The eight warmest years in the record have
all occurred since 1998, and the 14 warmest years
in the record have all occurred since 1990. The
most significant temperature increase is related
to the Northern Eurasia region.
6
Permafrost

• In Northern Eurasia the permafrost formation
  occupies an area greater than 10 million sq. km
• In the European part, permafrost occurs only in
  the tundra and the forest-tundra zone. In Siberia
  and Far East to the east of the Enisei the
  permafrost is spread almost everywhere, except
  for south Kamchatka, Sakhalin Island, and
  Primorjie

7
Forest ecosystems

• Forest is the largest biome and a single country
  in Northern Eurasia Russia is holding 22 of
  global and 38 of the Eurasian forest resources
  (FAO 2001)
• Russian boreal forest has an extremely unique
  type of forest cover. Extensive regions of Larch
  Forests underlain by permafrost.

8
Siberian pine regeneration under a larch canopy
9
Biosphere GHGs Inventory
EO data contribution
Production (CO2 uptake)
Consumption
Litter fall
Disturbances
Consumption Disturbances
Production
Respiration
Litter fall
CO2 respiration
Surface runoff
Deep leaching
Underground runoff
H
Deep leakage
CH4 release
Riverain discharge
A-atmosphere V-vegetation S-soil H-hydrosphere L-l
ithosphere
Stolbovoy et al., 2003
10
Main thematic focuses related to vegetation
monitoring at IKI RAS

• Land cover and change
• GLC 2000 for Northern Eurasia with SPOT-VGT
• Land cover annual mapping chain with MODIS
• Vegetation burning
• Operative burnt area mapping chains as a part
  national fire monitoring system (using both
  SPOT-VGT and MODIS)
• Forest burn severity assessment (using both
  SPOT-VGT and MODIS)
• Circumpolar post-season burnt area mapping using
  SPOT-VGT

11
Main thematic focuses related to vegetation
monitoring at IKI RAS

• Forest
• Forest characterisation (tree density, main
  species groups)
• Forest disturbances assessment (fire, insects,
  logging, industrial pollutions)
• Croplands
• Arable lands mapping
• Crop types classification
• Yield forecast and assessment
• Peatlands
• Peatlands types mapping with MODIS (initial stage
  of developments)

12
Main customers and users

• Sustainable natural resources management
• Russian Federal Forest Agency, including
• Forest Fire Protection Service (operational
  service since 2002)
• Forest Health Service (pilot project has started
  in 2007)
• Forest Inventory Service (pilot project has
  started in 2007)
• Ministry of Agriculture of Russia (operational
  service since 2004)
• Climate change science
• e.g. CLIVT - Climate Induced Vegetation Change
  Analysis Tool (project supported by Microsoft)
• International programmes and conventions
• UN FCCC and Kyoto protocol (resent request from
  national reporting group)
• FLEG - Forest Law Enforcement and Governance
  (potential cooperation on forest monitoring with
  EC)
• FRA 2010 (potential cooperation with FAO)

13
Some features of RD at IKI

• Focus is on national level (entire Russia)
  monitoring with expansion, if suitable, to
  sub-continental, pan-boreal or, potentially,
  global coverage
• Primary sources of EO data are moderate
  resolution satellite instruments (mainly MODIS
  and SPOT-VGT), while resent developments in
  Russia are rapidly increase the potential role of
  high-res. (e.g. SPOT-HRV/HRVIR) data for national
  monitoring
• Focus on long-term time-series data analysis for
  land cover mapping and monitoring
• Development of automatic satellite data receiving
  and processing chains to perform monitoring in
  the routine and repeatable manner

14
Earth Observation Instruments for Terrestrial
Ecosystems Monitoring
Thematic focuses of the consideration
15
VEGETATION and MODIS data archive for N. Eurasia
SPOT-VEGETATION Maximum NDVI ten-day composites (S10 product) Geographical coverage 42º-75ºN 5º-180ºE Time frame 1998 ongoing Temporal resolution 10 days Spectral bands 430 470 nm 610 680 nm 780 890 nm 1580 1750 nm Spatial resolution 1.15km (nadir view) Data delivery Internet access with 3 months delay (http//free.vgt.vito.be) Terra-MODIS Surface Reflectance Product (MOD09GHK, MOD09GQK, MODMGGAD, MOD09GST) Geographical coverage entire N.Eurasia (in process, 70 completeness) Time frame 2002 ongoing Temporal resolution daily Main spectral bands used 440 480 nm 620 670 nm 841 976 nm 1630 1650 nm Spatial resolution 250500m (nadir view) Data delivery Internet access with 2-7 days delay (http//modis.gsfc.nasa.gov)
16
MODIS data pre-processing steps
17
Cloud-free summer MODIS composite over Northern
Eurasia
Spatial resolution 250 ? June-August 2005
18
(No Transcript)
19

20
Forest Change Analysis
21
Land cover automatic mapping method
Thematic source data
Satellitedata
GLC 2000
Forest map
Peatlands map
Training data preparation
Histogram filtration
GIS analysis
Manual selection
Expert evaluation and correction
Training samples
Spectral mixture modeling
Classes signatures for cell-grid nodes
Training samples spatial regularization (gridding)
Contextual Maximum Likelihood classification
Auxiliary thematic products
New land cover map
GIS analysis
Burnt area
Croplands
Water mask
Urban mask
22
Filtering of local mapping errors and land cover
changes
Input sampling
Output sampling
Histogram filtering
Filtered histograms
Initial histograms
23
Training samples spatial regularization (gridding)

• Spectral signatures of each class are estimated
  for regular grid nodes with 25 km step using
  growing window approach. The window size depends
  on the spatial samples distribution and fixed
  threshold on number of samples.
• The estimated signatures include
• covariation matrix
• vector of average values
• number of samples.

24
MODIS data derived land cover map (250 m res)
25
30
SPOT-Vegetation data derived Circumpolar burnt
area for 2000-2004
50
2000
2001
2002
2003
2004
150
Forest cover
Water bodies
40
26
Forest logging mapping in Komi region with
Landsat-TM/ETM
Logging 1995-2001
Logging 2001-2002
Logging 2002-2004
Forest
Non-forest
Logging 1989-1995
27
Dynamics of Agricultural Lands in Russia
Landsat-TM, 06.10.1988
Sown area decreased at about 37 during last 25
years
Landsat-ETM, 05.10.1999
28
PVI time-series analysis

• Inter-annual PVI dynamic similarity assessment
• Phenological features retrieval start, max, half
  senescence, length of season

29
An arable lands map for Russia generated from
MODIS data
Draft product, status on July 2006. Mapping for
some regions is in progress.
30
Contextual Maximum Likelihood Classification
Local spectral-temporal signatures of classes
Spectral-temporal MODIS data composites
Covariation of metrics
Average of metrics
Number of samples
Metrics for the pixel
Maximum likelihood classifier
Probabilities for classes
31
Agriculture lands abandonment in Volgograd Region
Landsat-ETM, 05.10.1999
Landsat-TM, 06.10.1988
32
Abandoned lands detection

• Comparison between multi-temporal arable lands
  map shows abandoned and newly ploughed lands.

Green corresponds to previously arable lands
abandoned during year 2002
Red corresponds to previously abandoned lands
ploughed up during year 2008
Arable lands map for years 2002-2007
Arable lands map for years 2003-2008
33
Russian Academy of Sciences Space Research
Institute Moscow, June, 2009
Earth Observation and Informational Technologies
for Analysing of Climate Change and Vegetation
Dynamics Interactions
The research is performed within Joint IKI RAS
and MS Research CLIVT Project (Climate Induced
Vegetation Change Analysis Tool)
34
Main Characteristics of Vegetation Dynamics
The beginning of the understanding

• Dates of start and end of vegetation season
• The vegetation season duration
• Value of the seasonal VI maximum
• Dates of the seasonal VI maximum
• S VIFAR (photosynthetic potential) during
  vegetation season

35
Earth Observation data sets for CLIVT
Source data sets
Instrument / Satellite Resolution Resolution Time period covered
Instrument / Satellite Temporal, days Spatial, km Time period covered
NOAA-AVHRR 15 8 1981-2006
SPOT- Vegetation 10 1 1998-2008
MODIS-Terra 1 0,25 2001-2008
Spectral indexes to estimate fraction of
photosynthetically active radiation absorbed by
vegetation Normalised Difference Vegetation
Index NDVI (NIR-RED)/(NIRRED) RED and NIR
surface reflectance in red and near-infrared
spectral channels Optimised Vegetation
Normalised Index OVNI (for SPOT-VGT). The OVNI
index allows to take into account the variations
in the signal due to changes in the geometrical
conditions
36
Vegetation season duration as estimated using
SPOT-VGT data
The beginning of the understanding
SPOT-Vegetation multispectral composite Red
610-680 nm, Green 780-890 nm, Blue 450-520 nm
Duration of vegetation season

Decades
26
0
37
Integration of Earth Observation and
meteorological data
Methods for joint vegetation and meteo. data
analysis
Multi-annual NDVI time-series by land cover
types
NDVI for Evergreen Needleleaf Forest
Regular cell-grid for data integration
NDVI averaging for 2,5x 2,5 cell-grid by land
cover types

1st decade of June 1999
2nd decade of June 1999
Land cover map GLC2000
3rd decade of June 2007
Integrated analysis
Multi-annual NDVI time-series
Air-Temperature
1st decade of June 1999


1st decade of June 1999
2nd decade of June 1999
2nd decade of June 1999
3rd decade of June 2007
3rd decade of June 2007
Multi-annual time-series of meteorological data
38
Vegetation and regional climate change detection
algorithms
Parameters trends in 1982 -2006 with confidence
by F-distribution
Annual STgt0
Annual SDays when Tgt0
Annual precipitation-evaporation ratio
Annual SP when Tgt0
Evergreen needle-leaf forest NDVI
Steppe NDVI
Sedge tundra NDVI
Shrub tundra NDVI
lt0 0 gt0
39
Spatial distribution of the climatic factors
limiting condition of vegetation
Limiting factor
Relationship direction of limiting factors and
vegetation
1
3
R²
2
1
2
3
Temperature, ?
Precipitation, kg/m²
Precipitation, kg/m²
Evergreen Needle-leaf Forest
Limiting factor
Relationship direction of limiting factors and
vegetation
R²
1
2
3
3
1
Precipitation, kg/m²
Temperature, ?
Precipitation, kg/m²
2
Deciduous Broadleaf Forest
Limiting factor
Relationship direction of limiting factors and
vegetation
R ²
- Pozitive relationship - negative
relationship
- temperature - precipitation
0
0,9
40
Climate parameters trend directions during
growing seasons for all vegetation types

• Mean temperature

• Precipitation sum

41
Climate induced vegetation trends for all
vegetation types

• Vegetation trend

• Climate induced vegetation trend

• Test region covered by vegetation

42
Climate induced vegetation trends (1982-2006)

• Deciduous Broadleaf Forest

• Evergreen Needle-leaf Forest

• Deciduous Needle-leaf Forest

• Mixed Forest

• Steppe

• Humid Grasslands

• Tundra

• Bogs and Riparian vegetation

• Vegetation trend

• Climate induced vegetation trend

• Class mask

• Croplands

43
Development of scientific toolkits
EO and meteorological integrated database access
EO and meteorological data time-series joint
analysis
Steppe NDVI NOAA-AVHRR data in 1999 - 2006
Steppe OVNI SPOT-Vegetation data in 1999 - 2006

• Meteo data/Vegetation data

Meteo - Temperature/Precipitation rate/Relative
humidity/.. Vegetation - NDVI SPOT/OVNI
SPOT/NDVI NOAA
Steppe precipitation rate in 1999 - 2006
Steppe precipitation rate in 2006
Steppe temperature in 1999 - 2006
Steppe NDVI NOAA data in 2006
44
TerraNorte Data Products on-line
http//terranorte.iki.rssi.ru/