The area of interest is divided into three sub-areas:

1. Africa,

2. Southeast Asia (including insular tropical forest areas),

3. Amazonia (Brazilian and non-Brazilian).

The objectives, approaches and field studies mentioned here can also be adopted in studies of boreal and temperate forests.

Objectives

1. Determine changes in land cover types using remote sensing;

2. Determine land use intensities in anthropogenically affected cover types;

3. Obtain necessary parameters to feed climate models and models of the behaviors of biogeochemical cycles: carbon, trace gases, nutrients, hydrological cycles.

Note: The parameters which drive models which predict future rates of Humid Tropical Forest (HTF) loss or alteration (and the consequent fluxes of carbon, nutrients, water, trace gases, and climate change) include road building, population growth, economic alternatives outside the HTF, improvements in pasture management techniques, introduction of disease-resistant forage grasses, domestic and foreign market pressures for timber and meat, and very probably some other economic, political and social parameters. This working group does not include experts in these fields, so that predictive models of future land use will not be discussed here. Modelers with this expertise might be invited to a future meeting. The human dimension of Land Use Cover Changes in the HTF is also the subject of the Human Dimension Program of the IGBP.

Approach

1. Use whatever imagery is available, ideally MSS and TM, but it may be necessary to rely on AVHRR 1-km and orbital radar sensors (RADARSAT, JERS-1, ERS-1, dual polarizaton ENVISAT, dual polarized, one-time SIR-C covering ± 60° latitude) in some areas.

2. Identify the natural boundary between HTF and the neighboring savanna/woodland biome. Save this decision in a widely used GIS file format.

3. Classify HTF into at least four categories and include in a widely used GIS format:

   a) Primary forest plus secondary forest over a threshold age [threshold spectral criterion];

   b) Deforested: includes bare soil, short cycle crops, perennial crops;

   c) Secondary forest up to the threshold age [threshold spectral criterion?]; and

   d) Non-forest.

4. For deforested and secondary forest areas, determine the history of land use, pixel by pixel, in order to derive a classification based on the history of the pixel. Develop index of land use intensity based on this classification; for example, number of years pixel was classified as NPV/bare soil.

5. Modify the land use intensity index with such parameters as soil texture, soil fertility, land use type (pasture vs. swidden) to develop a site degradation index.

Field Studies

1. Ground truth spectral classes; i. e. supposed land-use/land-cover types, including the different stages input into the land use history classes, as seen in temporal series of images.

2. Determine behavior of carbon, trace gases, nutrients, and hydrology under different land use regimes and histories. For example after a single slash-and-burn of primary forest, after repeated crop/fallow cycles in swidden agriculture, and after long-term pasture.

3. Determine LAI, APAR, NPP, and GEP of different cover types.

4. Create a temporal transformation matrix where boxes represent different land cover or land use types, including different land use history/intensity classes. See recent work of P. Fearnside.

Problems

1. Availability of data: TM & MSS not available everywhere.

2. Archiving problems: encourage places like INPE to maintain readability and archival security of old MSS data. This is irreplaceable data spanning such important natural change periods as 1982/83 El Niño. Some data has already been lost. Express need of the scientific community in writing for this archival effort. Write to:
   Dr. Luiz Alberto Vieira Dias
   Chefe, Observacoes da Terra
   INPE-Instituto Nacional de Pesquisas Espaciais
   Caixa Postal 515
   12.225 Sao Jose dos Campos, SP
   BRAZIL

   Note:Dr. Vieira Dias has stated that such letters from the scientific community, showing a demand for these products, is required in order to obtain the necessary funds for archiving and data recovery.

3. It is impossible to discriminate many types of forest degradation. Only the most intensively logged areas are visible on TM images and the spectral distinction disappears within three years.

4. There is ambiguity in separation of anthropogenic secondary forest vs. some natural forests. This problem is greatest in SE Asia, a small problem in the non-Brazilian Amazon, but not a problem in the Brazilian Amazon. The error should be estimated for each region by field checking.