GFW-Commodities RSPO Land Use Change Analysis Methodology DRAFT v06/02/2014 Context: The Roundtable on Sustainable Palm Oil (RSPO) Principles &Criteria (P&C) (Criteria 7.3) stipulates: New planting since November 2005 have not replaced primary forest or any area required to maintain or enhance one or more High Conservation Values (HCV). RSPO palm oil producer members are required to have completed HCV assessments on their land holdings for new plantings since November 2005. According to the Specific Guidance for 7.3.1: Where land has been cleared since November 2005, and without prior and adequate HCV assessment, it will be excluded from the RSPO certification program until an adequate HCV compensation plan has been developed and accepted by the RSPO. The RSPO Remediation and Compensation Procedures were developed in order to establish a formal mechanism to identify and, where appropriate, remediate or compensate for potential losses of HCV associated with clearance of areas without prior HCV assessment and to provide a pathway for certification applicants in control of areas non-compliant with 7.3 to resolve their non-compliance in order to proceed with certification (or maintain their certified status. Effective May 9, 2014, the staged implementation of the Remediation and Compensation procedures requires all RSPO members who own and/or manage land for oil palm production to disclose any non-compliant land clearance and to submit a historic land use change analysis for these non-compliant areas. Additionally, the revised 2013 P&C (7.3.2) further states that HCV assessments shall include Land use change analysis to determine changes to the vegetation since 2005. This analysis shall be used, with proxies, to indicate changes to HCV status.
Approach: Global Forest Watch Commodities (GFW-Commodities) is an online mapping and analysis tool created by the World Resources Institute (WRI) and over 40 partners. WRI has developed an “RSPO Land Use Change Analysis” tool for GFW-Commodities that produces an estimated result for the historic land use change assessments required by the RSPO Remediation and Compensation Procedures and New Plantings Procedures. This analysis is based on the guidelines described in Section 7 and Annex 2 of the Remediation and Compensation Procedures and summarized below. The output for a user-defined area of interest describes hectares of annual tree cover loss beginning in 2006 within four land cover categories, which
are approximations of the vegetation coefficient categories described in Section 7, with a baseline year of 2005. These categories are: “primary forest”, “secondary forest”, “agroforest”, and “non-forest.”
Vegetation Coefficient Category in RSPO Remediation and Compensation Procedures Coefficient 1.0: Structurally complex forest (including primary forest), regenerating, selectively logged forests with elements of high canopy.
Approximated Category for GFW-C RSPO Land Use Change Analysis Tool “Primary forest”
Coefficient 0.7: Structurally degraded but ecologically functional natural forest.*
“Secondary forest”
*Includes other degraded but still functional low-canopy secondary forest and pioneer-dominated, heavily and/or repeatedly logged or previously burned forest and regenerating forest. Coefficient 0.4: Multi-species agroforestry.
“Agroforest”
Coefficient 0: Monoculture tree and non-tree plantations; other permanently cultivated, developed or open degraded land
“Non-forest”
The GFW-Commodities RSPO analysis uses ArcGIS Image Server to intersect Annual Tree Cover Loss data (Hansen et al., 2013) with a reclassified version of the Southeast Asia 2005 Land Cover dataset (Gunarso et al., 2013). This analysis identifies pixels where loss occurred in the years of interest and the land cover type upon which that loss occurred. All land cover types were reclassified into “primary forest”, “secondary forest”, “agroforest”, or “non-forest”. See data descriptions and table below for full reclassification scheme. For detailed descriptions of land cover categories defined in the Southeast Asia 2005 Land Cover dataset, including description on bare soil, materials and methodology used in the analysis, please refer to this document: http://www.rspo.org/file/GHGWG2/4_oil_palm_and_land_use_change_Gunarso_et_al.pdf.
Data: Land cover classifications: The Southeast Asia Land Cover dataset (Gunarso, et al., 2013) identified 22 different land cover types using visual interpretation of Landsat 4, 5, and 7 images over three time periods (1990 to 2000, 2001 to 2005, and 2006 to 2009/2010). Three types of natural forest were identified: upland forest, swamp forest, and mangrove. These forest classes were further characterized by their level of human activity: undisturbed or disturbed. Undisturbed forest classes are characterized by an absence of logging roads and settlements. Disturbed forest classes showed evidence of logging including roads and small-scale
clearing. Undisturbed upland forest, swamp forest, and mangroves were reclassified as “primary forest” for the purposes of this RSPO analysis whereas disturbed upland forest, swamp forest and mangrove were reclassified as “secondary forest.” All other land cover classes including managed tree plantations (timber, rubber, and oil palm) were reclassified as “non-forest” with the exception of “mixed tree crops / agroforest” which fit into RSPO’s “agroforest” land cover class. The 2005 time step of the land cover data, representing 2001-2005, serves as the baseline land cover for the RSPO Land Use Change Analysis. Land cover change: A complete land cover change dataset is not available for each of the required years for analysis under the RSPO procedures, so the GFW-C RSPO Land Use Change Analysis tool utilizes a global layer describing annual tree cover loss (Hansen et al., 2013) as a proxy for land cover change. This data set measures areas of tree cover loss at 30 x 30 meter resolution across all global land except Antarctica and some Arctic islands. The data were generated using multispectral satellite imagery from the Landsat 7 thematic mapper plus (ETM+) sensor. Landsat 7 images were compiled and analyzed using Google Earth Engine. Clear land surface observations in the images were assembled chronologically and pixel tree cover loss was identified using a supervised learning algorithm. “Tree cover” is defined as all vegetation taller than 5 meters in height and identifies the biophysical presence of trees in the form of natural forests as well as plantations over a range of canopy densities. “Tree cover loss” indicates the complete removal or mortality of tree canopy cover at the Landsat pixel scale due to a variety of factors including but not limited to mechanical harvesting, fire, disease, or storm damage.
Cautions: This analysis is intended to be used as a preliminary estimate. It should not be used to fully meet the land use change analysis reporting requirements for the RSPO Remediation and Compensation Procedures. The results of this analysis have not been systematically verified. WRI highly recommends the use of multiple sources of satellite imagery and other contextualizing, site-specific information.
Land cover classifications from Southeast Asia 2005 Land Cover (Gunarso et. al, 2013) which was adapted from Badan Planologi, Ministry of Forestry of Indonesia (2001). Value
Code
Class
Description
Corresponding GFW-C RSPO Land Use Change Analysis Classification
Natural forest cover with dense canopy, highly diverse species and high basal areas. It has no logging roads, indicating that it has never been logged, at least under large-scale operation, and in some areas in Indonesia located in areas with rough topography. Canopy cover of undisturbed forest is usually >80%. In satellite image, it is indicated by high value of vegetation index and infrared spectrum channels, and lower value in visible spectrum channels.
Primary Forest
1
UDF
Undisturbed Forest
2
DIF
Disturbed Forest
Natural forest area with logging roads and degraded forest cover or logged spots.
Secondary Forest
3
USF
Undisturbed Swamp Forest
A swamp forest is a natural forest in wetland featuring temporary or permanent inundation of large areas of land by shallow bodies of water.
Primary Forest
4
UDM
Undisturbed Mangrove
Undisturbed mangrove is area along the coastline with high density of mangrove tree species, usually consists of diverse mangrove species composition, and has never been logged.
Primary Forest
5
DSF
Disturbed Swamp Forest
Logged-over swamp forest is swamp with natural forest cover that has sign of been logged or degraded.
Secondary Forest
6
DIM
Disturbed Mangrove
Logged-over mangrove is area along the coastline with various species of mangrove trees, has been logged in the past and partly degraded.
Secondary Forest
7
RPL
Rubber Plantation
Rubber Plantation.
Non-Forest
8
OPL
Oil Palm Plantation
Oil Palm Plantation.
Non-Forest
9
TPL
Timber Plantation
Monoculture timber plantation (e.g. Gmelina sp., Paraserianthes falcataria, Acacia mangium) where the area is less than 1 ha. Tree canopy cover is around 30-50%.
Non-Forest
10
MTC
Mixed Tree Crops
Agroforest is a mixed tree based system with more than 30% of the area consists of various species of trees. Mixed garden usually located in 0.5-1km distances to settlement or road. Tree canopy cover can reach 5-60%. Several example of agroforest are rubber agroforestry system, coffee agroforestry system, and home garden.
Agroforest
11
SCH
Shrubs
Non-tree-based system consists of non tree vegetation usually less than 5-6 m (15-20 ft) tall, usually resulted from swidden agriculture activities or logging area that has been left for 2-3 years as part of the fallow/rotational systems.
Non-Forest
12
SSH
Swamp Shrubs
Non-tree-based system consists of non tree vegetation usually less than 5-6 m (15-20 ft) tall, usually resulted from swidden agriculture activities or logging area that has been left for 2-3 years as part of the fallow/rotational systems on the area that temporary or permanent inundation of large areas of land by shallow bodies of water.
Non-Forest
13
DCL
Dry Cultivation Land
Open area characterized by herbaceous vegetation that has big probability planted or intensively managed such as row crops. Sometimes it mixed with wide spaced brushes or trees. The features usually associate with settlements.
Non-Forest
14
SET
Settlements
Characterized by settlement-included homestead, urban, rural, harbor, airports, industrial area, open mining. Associates with road network or constructed materials.
Non-Forest
15
GRS
Grass
Extensive cover of grasses with scattered shrubs or trees.
Non-Forest
16
SGR
Swamp Grass
Extensive cover of grasses with scattered shrubs or trees in swamp area.
Non-Forest
17
RCF
Rice Field
Open area characterized by herbaceous vegetation that has big probability human managed such as paddy field. The features usually associate with settlement or irrigation structure.
Non-Forest
18
CFP
Coastal Fish Pond
Open area in the coastal with block pattern and always inundated.
Non-Forest
19
BRL
Bareland
Areas characterized by bare rock, gravel, sand, silt, clay, or other earthen material, with little or no woody vegetation present regardless of its inherent ability to support life; include forest clear-cut, forest conversion, and changes due to natural causes (e.g. fire, flood, etc.).
Non-Forest
20
MIN
Mining
Open area with mining activities.
Non-Forest
21
WAB
Water Bodies
Water bodies - seen from its reflection of water on the image.
Non-Forest
22
NCL
Not classified including cloud cover
Not classified including Cloud Cover.
Non-Forest
Citations: Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A. Turubanova, A. Tyukavina, D. Thau, S. V. Stehman, S. J. Goetz, T. R. Loveland, A. Kommareddy, A. Egorov, L. Chini, C. O. Justice, and J. R. G. Townshend. 2013. “UMD Tree Cover Loss and Gain Area.” University of Maryland and Google. Roundtable on Sustainable Palm Oil Principles and Criteria for the Production of Sustainable Palm Oil. 2013. Endorsed by the RSPO Executive Board and accepted at the Extraordinary General Assembly by RSPO Members on April 25th 2013. Roundtable on Sustainable Palm Oil Remediation and Compensation Procedures Related to Land Clearance without Prior HCV Assessment. Endorsed by RSPO Board of Governers for staged implementation on March 6th 2014 (Effective date May 9th, 2014). Gunarso, P., Hartoyo, M., Agus, F., and T. Killeen. 2013. Oil Palm and Land Use Change in Indonesia, Malaysia and Papua New Guinea. Reports from the Technical Panels of the 2nd Greenhouse Gas Working Group of the Roundtable on Sustainable Palm Oil (RSPO). Published November 2013 at www.rspo.org.