The NASA Remote Sensing component to CARPE is built largely upon the foundation laid by the NASA EOS LANDSAT PATHFINDER Humid Tropical Forest Project. 

The Landsat PATHFINDER Project is a cooperative project between NASA/USGS-EDC and EPA designed to map the rates of deforestation in the tropics using satellite data from three epochs (early 1970's, mid-1980's and mid-1990's). The University of Maryland and University of New Hampshire are mapping the tropical forest belt using Landsat data. The Central African component of the PATHFINDER Project is being implemented at the Geography Department of the University of Maryland in cooperation with scientists from NASA Goddard Space Flight Center. Emphasis is currently being placed on acquiring and producing derived data sets on deforestation for Central Africa, from the historical time series of Landsat data, from the early 1970's to the present. The objective of the PATHFINDER project is to generate a direct and spatially explicit measure of the rates of change in the forest cover over the last twenty years. The output products from PATHFINDER will provide a high resolution database for assessing and monitoring the forest resources of the region. 

Fig.2 

In comparison to other remote sensing initiatives in Central Africa, the Landsat PATHFINDER project has chosen an extensive (wall to wall) and high resolution (Landsat) approach to study deforestation. The European TREES project has adopted a two step approach : combining  low resolution (AVHRR data) analysis at a regional level with a high resolution (Landsat / SPOT) study on specific local sites (i.e. "Hot Spots"). The FAO FRA project has adopted a sampling approach: examining 12 sites in Central Africa using high resolution (Landsat) data. The FAO objective is to develop a global statistic on forest extent and rates of change. 

• The low resolution regional view of the TREES project is a single epoch view (ie.1990's). The PATHFINDER provides a multi-date analysis of the forest, focusing on the extent and timing of vegetation changes.
• The objective of the Landsat PATHFINDER project is to give a synoptic view of the forest changes since the 70's, initially to provide information for carbon modelling at global scale. To achieve this goal, the images are interpreted in only three main classes: forest, degraded forest and savanna.
• The 'Pathfinder' approach requires the acquisition of a large amount of Landsat data. The best available images have been selected for each of the three epochs : 70's period (Landsat MSS), 80's period (Landsat MSS or/and TM), 90's period (Landsat TM). Due to persistent cloud coverage, multiple images have been acquired for specific locations.
• The Landsat PATHFINDER project covers the tropical forest belt. For Central Africa, at least 120 images must be processed for each time period. An effective processing methodology has been chosen with an emphasis on simplistic and automated procedures.

A critical aspect of  this process is to assess the quality of the Landsat PATHFINDER products. The validation methodology is divided into different steps. In these steps, the products are checked, according to internal consistency, and compared to the available field information. Field information is crucial for the initialization and validation of the classification. Different approaches are used to obtain field information. 

Send Comments to: