Large-scale ecological vegetation mapping in Indonesia
Vegetation maps are a key tool to depict the mosaic of vegetation communities in the landscape are mirroring the state of land health and environment. They are compulsory in all integrated landscape management and development programs. Despite several UNESCO and FAO recommendations, these ecological vegetation maps are practically non-existent in South East Asia. In an effort to promote more bottom-up landscape management, CIRAD and CIFOR have developed a large-scale vegetation mapping initiative in line with the new regulation for mapping at the regency level.
Ecological mapping principles
The ecological vegetation map methodology used by CIRAD and CIFOR was developed by the Vegetation School of the Toulouse University in France. The method has been adopted by UNESCO and FAO in many countries. The principle is that it focuses on understanding the decisive ecological factor(s) that will induce the different vegetation types and their successional stages, and therefore it underlines the fact that vegetation is the finest indicator of the environment. The convention governing the use of colours and symbols is also specific to this kind of mapping has been recommended by UNESCO.
For Indonesia, it makes use of baseline internationally used ecological classification for the country enhanced by introducing the eco-floristic classification concept (Laumonier 1997).
The procedure used at CIFOR follows the general principles for this kind of mapping although there have been successive adaptations in the course of its application in Sumatra, Kalimantan, Sulawesi, Sumba and the Moluccas.
Pre-stratification and hierarchical ecological classification for systematic sampling of the vegetation units.
- Small scale bioclimatic data and maps are first analysed, and small-scale physiographic regions are refined using geological maps and land unit maps at the scale of 1:250,000. Particular attention is paid to the following geological formations and substrats: recent and old alluvium, recent and old volcanic, sedimentary, metamorphic and intrusives rocks.
- General characteristics of slopes (regularity, shape, gradient) are analysed next, in conjunction with interpretation satellite imageries. Topographical large-scale maps (1:50,000) or Digital Elevation Models using radar data are used in conjunction with landform classification to help identify slope patterns and landforms units.
- Altitudinal zonation. Refinement of Van Steenis classes (1972), following results of Symington 1943 and Laumonier 1997, later on confirmed by Cannon (2008) (<300; 300-800; 800-1300; 1300-1800; 1800-2500; > 2500m)
- Remote sensing data interpretation
Some principles for remote sensing data interpretation have to be respected, such as:
- The accuracy of the interpretation for all categories should be approximately equal, with a minimum level of accuracy in identifying vegetation cover categories of at least 85 percent.
- Replicable results should be attainable from one interpreter to another.
- The classification system used should be applicable over extensive areas, and usable for remote sensing data obtained at different times of the year.
- The classification system used should allow addition of subcategories that can be derived from ground surveys or from larger scale remote sensing data, and aggregation of categories should also be possible.
The vegetation is mostly interpreted at 1:50 000 scale from data acquired by the LANDSAT satellite instruments in 2017 with a spatial resolution of 30 m. The method used follows the procedure recommended by King (2002) for combining computerized (supervised classification) and manual interpretation. For the later 4,5,3 band combination of red, near infrared and green are used for color composites. Variations in moisture content are better identified with this set of bands. We also use the 7,4,2 band combination (shortwave infrared, red and blue) to interpret areas with swamp vegetation more accurately. Besides the use of these band combinations, it is adviced to differentiate between “Localized, spectrally mixed, usually large, with few occurrences (LSM: localized spectrally mixed), not really interpretable using supervised classification and “Widespread, spectrally consistent, usually small and with many occurrences” (RSC: repetitive spectrally consistent) suitable fot supervised classification. At 1:50 000 scale, on screen interpretation is often preferred to supervised classification in complex mosaic of swidden agriculture landscape or swamp vegetation mosaic.
Ground check and plot survey to validate vegetation map units
Using a geographic information system (GIS), (1) (2) and (3) above are overlaid with the remote sensing interpretation (4) to identify homogeneous units in term of environmental variables relevant to vegetation. Within each strata, units are equally systematically sampled using a network of small 0.2 ha survey plots laid down approximately every 100 m along transects perpendicular to the major topographical gradient. Positioned in this way, sample plots covered a wide range of land facets, which are defined here in terms of their topographic position (valley bottom, slope, ridge). Such a design maximizes the habitat range of data collected and is considered more suitable for extrapolation to landscape level and modeling purpose. This design also helps separate the effects of soil variations according to topographic position on different sites. In each plot (100 x 20 m), diameter at breast height (DBH) ≥ 10 cm, total height and height of the first branch are measured. The contribution of the major life forms in the forest structure is also recorded, botanical collection systematically made and vouchers later identified at the Bogor Herbarium.
In addition, soil samples are often collected at soil horizons of soil pits (0–100 cm) for physico-chemical properties analysis and whenever possible other taxa sampled for biodiversity assessment (mainly birds, soil fauna, small mammals or bats).
Applications: Integrated Landscape Management and conservation
The large-scale ecological vegetation maps provide new data on present geographical distribution of a whole range of forest and vegetation types (secondary forests, mixed gardens, agroforestry, degraded lands, oil palm and other plantations etc.), geographical distribution of plants (mainly tree species diversity), providing at the same time information on the state of the environment, such as land degradation. This information is important to analyse patterns of diversity and carbon, and may provide a basis for assessment of ecosystem services. The applications are numerous, mainly:
- Integrated Watershed Management
- Land health evaluation / Vegetation and land units (soil)
- High Conservation Value forest identification / High Carbon Assessment
- Eco-floristic zoning to select potential conservation areas
- Fragmentation and connectivity studies related to the need for restoration and ecosystem-based adaptation initiatives
- Bioclimatic maps
- Forest function allocation
Fullscreen 
Share 
