Frédéric Achard European Commission – Joint Research Centre

Frédéric AchardEuropean Commission – Joint Research Centre

2nd GOFC-GOLD Workshop on reducing emissions from deforestation

Santa CruzBolivia16 -19 April 2007

Introduction:

The Joint Research Centre and tropical forest monitoring

JRC TREES 3 / FAO FRA 2010 projects:

Joint activities relating to the technical and practical issues for Reducing emissions from deforestation in developing countries

Example of regional forest cover change estimation:

The Congo Basin for the period 1990-2000

Part of the JRC’s work program is:

o to provide Scientific & Technical support to EU policies for o the Protection of the Environment

o to contribute to the EU Strategy on o Sustainable Development

The JRC has therefore been active in monitoring tropical forestsfor over 20 years

These activities have been carried out under the TREES project

- baseline inventory of the distribution of pan tropical forests

- methods to highlight areas of rapid change

- Quantification of pan tropical forest change for the period 1992 to 1997 using a sample of high resolution satellite data

Mapping

Monitoring

Measuring

TREES I / TREES II 1992 - 2002

In co-operation with FAO FRA :

- providing the input to FAO FRA RS Survey

- in collaboration with the FAO FRA regional / national networks

TREES-3 (2007 – 2013)

Forest change assessment: 1990 – 2000 - 2005 (2010?)

Phase I: tropical humid forests (start 2007)

Phase II: tropical dry forests and Eurasian boreal forests

Changes estimates to be valid at global & regional scales

FAO FRA 2010 RSS SCOPE & OVERALL PURPOSE

• Spatial distribution of forests at global & regional levels

• Land use change information (deforestation, fragmentation and forest degradation)

• Consistent information over time and space

• Independent monitoring system at global and regional level

• Complementary to national reporting and with close links to national monitoring systems

• Increased in-country capacity

FAO FRA 2010 RSS OUTPUTS

• Regional, biome and global monitoring of forests 1975-1990-2000-2005:– area change statistics– information on land use dynamics (including

deforestation, expansion of forests, fragmentation and forest degradation)

• Forest maps• A framework to support further monitoring of forests, land

use and the environment• Baseline data for research & modeling• Capacity building

RSS SAMPLING DESIGN

PROPOSED TIME FRAME

19751990

2000

?

2005

IMPLEMENTATION

• Ca. 9 000 sample units, 4 dates• Centralized pre-processing & website• Standard methodology and classification• Regional training workshops• Interpretation by national / regional teams• Regional validation workshops• Centralized statistical analysis

A regular sample of sites based on the geographic grid across the tropics (an adapted scheme outside the tropics)

TREES-3 Forest change assessment

Systematic Sample with units at 1° x 1° lat/long

TREES-3 Number of sampling units

TROPICAL DOMAIN

Latin America Africa S& SE Asia

Central + Caribbean: 60 Humid: 451 South: 314

South America: 1168 Dry: 14731 Southeast: 421

Region: 1228 Region: 1924 Region: 735

Total: 3887

1 including full dry forest domain of Africa

• 20km x 20km sample units at each 1º lat-long grid intersection• based high spatial satellite imagery (Landsat TM, SPOT,

ASTER, similar..) for each date

a) forest cover change assessment within units of 20km x 20km

b) assessment of indicators for degradation processes (openings, roads..)

TREES-3 Forest change assessment

TREES-3 Land Cover LegendFAO LAND USE→:

↓TREES-3 LAND COVER:

Forest Other wooded

land

Other Land

With TC

Burnt Other……..

Water No Data

• Trees

• Mosaics Trees & other Land Cover

• Shrubs

• Burnt Areas

• Other Vegetated Cover

• Other Non-Vegetated Cover

• Water

• No-Data

TREES / FAO sampling scheme over South Asia

TREES / FAO sampling scheme over South America

Around 50 sites over Bolivia

1 degree grid0.5 degree grid0.25 degree grid

Sampling Coverage

Data download and site extraction

Site image control and pre-processing

Segmentation

Class labelling

Change layer generation

Internal consistency control

External control and added value

Automatic Box extractor (example PNG)

30 km X 30m boxes of year 2000 ETM data

30 km X 30m boxes of year 1990 TM data

Dry forest in NE Bolivia (Bosque Seco Chiquitano)

1990 2000Example of sample in Bolivia : 16 S / 61 W

Semi-automatic segmentation and change label for paired dates

Paired segmentation Automatic change label

1990

2000

TREES-3 Forest change assessment

Minimum mapping unit has implications for control

With MMU 0.5 ha: 1,500 polygons (average size 26 ha)

If MMU is too large : features are lost (e.g. burnt area)

With MMU 2 ha: 400 polygons (average size 94 ha)

Forest cover change assessment procedure

1. Image segmentation (automated)

2. Preliminary labeling of objects and changes (automated)

3. Verification and Adjustments of Labels- regional partner co-operation -

4. Validation - regional partner co-operation -

5. Adding forest and land use dimension (FAO FRA)- regional partner co-operation-

Example of regional forest cover change estimation:

The Congo Basin for the period 1990-2000

Dense forest

Degraded forest Mosaics Urban Savana

For. dense

For. DégradéeMosaïque Foret/Agriculture

Sol nu / urbain

Savane

Change matrice for the Congo Basin case study

Central AfricaForest area = 180.4 M ha

n average C.I.

101 2.20% 0.88%

104 1.48% 0.74%

98 1.94% 1.02%

97 2.73% 1.22%

400 2.08% 0.48%

Total deforestation over 10 years

1 degree X

1 degree

1/2 degree X

1/2 degree

Congo Brazzaville Dem. Rep CongoForest area= 22.2 M ha Forest area = 108.3 M ha

n average C.I. n average C.I.

14 0.86% 0.79% 68 2.45% 1.10%

16 0.67% 0.76% 68 1.95% 1.03%

11 0.03% 0.06% 65 2.72% 1.46%

15 1.02% 1.10% 67 2.72% 1.24%

56 0.69% 0.56% 268 2.45% 0.55%

Total deforestation over 10 years

1 degree X

1 degree

1/2 degree X

1/2 degree

Activity data :

Global and regional forest conversion rates

Methods for national forest conversion rates

Constant data :

National carbon stock estimate per forest type

Conclusions

Potential contribution of TREES-3 / FAO FRA 2010 to

the “reduction of emissions from deforestation” item

36

Mid-Decadal Global Land SurveyMid-Decadal Global Land Survey

Jeff MasekBiospheric Sciences Branch, NASA GSFC

Steve CovingtonAerospace Corporation / USGS

February 28, 2007

37

Mid-Decadal Global Land Survey (MDGLS)Mid-Decadal Global Land Survey (MDGLS)

Follow-on to the GeoCover orthorectified global data sets (1975, 1990, and 2000 epochs) centered on 2004-2006

Partnership between USGS and NASA, in support of CCSPSupport global assessments of land-cover, land-cover change, and ecosystem dynamics (disturbance, vegetation health, etc)Landsat-5 TM and Landsat-7 imagery, with ASTER and EO-1 ALI data as needed

38

Phase 1: identify all candidate scenes and ingest into the USGS archive (USGS lead)

Phase 2: Process selected data into an ortho-rectified dataset compatible with previous surveys (NASA lead)

Phase 3: Analyze data set to quantify trends in land cover and vegetation dynamics (NASA LCLUC)

MDGLS DevelopmentMDGLS Development

39

What data are available? Landsat-7What data are available? Landsat-7

Green = Base ≤ 1% CC, Fill ≤ 5% CCYellowYellow = Base ≤ 5% CC, Fill ≤ 10% CC

3 month acquisition windows, 95% fill coverage

40

Archived L7 Coverage Meeting SpecificationArchived L7 Coverage Meeting Specification

Global land coverage using L7 composite image pairswith >95% coverage

Prime ACCA> ,1Fill ACCA < 5

63%

No SuitableL7 Composite

Coverage18%

Prime>Fill < 20 ,5 8%

Prime>Fill < 10 ,5 5%

Prime>,1 Fill < 10

6%

41

Combined Archived Coverage in EROS ArchiveCombined Archived Coverage in EROS Archive

GreenGreen = ETM+ 5%/10% CC FillYellowYellow = TM <10% CC in EROS ArchiveRedRed = TM = ??% CC in IC Archives

>91% of the P/R Locations Covered

42

Phase 1 StatusPhase 1 StatusPhase 1: satellite tasking, ground station coordination, scene selection, data transfer, and ingest into the USGS archiveSince December 2005

Developed and implemented an MDGLS acquisition strategyDeveloped QA management tool and automated scene selection toolEstablished a network of 6 campaign stations to collect Landsat 5 data

4 have provided data (Kiruna, Moscow, Irkutsk, Maspalomas) 1 is under construction and will begin collections in early 2007(Chetumal)1 is in progress (Malindi)

Most International Cooperators have agreed to supply image data in support of the MDGLS Project

6 have provided metadata to USGS6 stations have confirmed Jpeg browse – for easier inspection

43

Mid-Decadal Global Land Survey (MDGLS)

Phase I: Identify and Acquire L5 and L7 Data

Phase II: Process MDGLS Data

Phase III: Analyze MDGLS Dataset for Land Cover/ Land Cover Change

44

• Establish MDGLS Product Specifications

•Select data source and scenes (where multiple options are available)

• Implement automated orthorectification at EROS

• Process selected data- Orthorectification- Gap-filling (for Landsat-7)- Product format

• Distribute MDGLS dataComplete dataset to be available Fall 2008

Phase II Tasks

45

Data Source Selection: Status

Recommendation: - For cloud-free scenes (<2% CC): lean toward Landsat-7 ETM+- Cloudier scenes (2-10% CC): lean toward Landsat-5 TM- Humid Tropics: multiple acquisitions for compositing

Sensor choice must be balanced against acquisition date, overall cloud cover, and acquisition date of 2000 Geocover

- Optimization algorithm being developed to assist selection

46

MDGLS Orthorectification

Need to reprocess previous GeoCover datasets in high-relief areas to maintain continuity with MDGLS

- model absolute error due to Geocover DEM choice- reprocess locations with errors >15m using SRTM

Use 2000 GeoCover chips as geodetic control, SRTM DEM for terrain correction

- L7 automated 1Gt processing available May 2007- L5 automated 1Gt processing available Spring 2008

Geodetic accuracy relative to 2000 Geocover of 30m RMSE (or better). Maximum absolute geodetic error of 100m.

47

MDGLS Product Specification (Draft)

• UTM / WGS-84 projection• 14.25 / 28.5/ 57 meter resolution• Cubic Convolution resampling (1 step)• GeoTiff format• Orthorectified, Gap-filled

Processing by USGS EROS

FTP distribution of individual MDGLS scenes at no cost, with limited provision for bulk distribution of entire dataset (e.g. via hard disk transfer).

48

MDGLS ScheduleMDGLS Schedule

CY 2007 CY 2008

Image Acquisition

Phase 1 Activities

Scene Selection

*IC = International Cooperator

IC Data Collection

Phase 2 Activities

Phase 2 Planning

Product Generation

IC* Metadata Collection

Product GenerationPrime Acquisition Period

CY 2006CY 2005CY 2004