Click here  to view the map (1.28MB)

ESSC recently participated in two consecutive validation activities of the country's forest cover as invited national experts for the Philippines. 

Jose Andres Ignacio, ESSC's Geomatics Manager, was in Ispra, Italy from 12 to 19 June 2010 to contribute to the ongoing Global Forest Monitoring Project (TREES-3) for Southeast Asia and specifically for the Philippine datasets.

The TREES-3 Action provides quantitative measurements and mapping of changes in forest resources for EU policies related to global environmental and forestry issues, with a focus on Eurasian boreal forests and tropical forests, including the Caribbean and Pacific regions. It also addresses forest cover and cover change issues related to EU commitments to multilateral environmental agreements, especially to UN conventions such as the UNFCCC, UNCCD, UNCBD, and the UN Forest Forum, as well as Action Plans such as on Forest Law Enforcement Governance and Trade (FLEGT).

The TREES-3 generates regional forest maps, tracks areas of rapid forest change and produces statistically valid estimates of cover change for the current and previous decades (from the mid 1970s to 2005-10). It identifies the regional drivers of deforestation, with a focus on tropical and Eurasian boreal forests.

The TREES-3 products are used as inputs for future climate change impact scenarios and, through close cooperation with DG Environment, provide a basis for developing non-Annex-1 countries' inputs into the Kyoto Protocol process.

The regional forest maps and estimates of cover change are shared with Commission services, EU Delegations , international organizations (particularly the Food and Agricultural Organization or FAO ), and partner countries. Biomass maps and carbon emission/storage estimates are produced for selected forest ecosystems. Due to persistent cloud in the tropics and poor illumination in winter months in the boreal ecosystems, the TREES-3 Action develops forest monitoring techniques which include radar technologies, as these have good cloud penetration properties and operate without sunlight.

Andres was invited to validate tree cover classification that they did for the Philippines based on 1990 and 2000 data sets. The TREES-3 project is mandated to do a global assessment of forest cover (tropical and boreal) on a regular basis.

This activity Andres was invited to was an exercise to validate their data classification using 20km x 20km sampling grids. They developed a software tool to facilitate the coding of segmented polygons based on the 1990 and 2000 images.

Andres provide their programmer additional functionality requests to further aid in the validation process for future updates using the same tool.

On the whole, the decision-making procedure was very similar to the one ESSC eventually developed for the ESSC 2002 Philippine forest cover dataset, mainly using visual photointerpretation as the eventual basis for determining the type of forest.

It was good to know that even within the EC-JRC level, the validation process is not any different from what ESSC developed considering the various problems associated with automated statistical classification methodologies applied to large areas.

Global Forest Resources Assessment (FRA) 2010 Remote Sensing Survey (RSS) Technical Workshop, Asian Institute of Technology, Pathumthani, Thailand, 6 to 19 August 2010

Emmanuel "Maning" Sambale, ESSC's Geomatics Associate, went to Thailand in August 2010 as part of the Philippine team to focus on the validation of land cover and land use labels for all FRA RSS sample locations by national experts from each country in the region. The results obtained will be crucial in the completion of the Global Remote Sensing Survey and to support individual countries' work on forest extent and change detection.

The Remote Sensing Survey is part of the several research initiatives in preparing the Forest Resource Assessment 2010 Report of the Food and Agriculture Organisation (FAO). The main goal is to obtain systematic information on the distribution and changes in forest cover and forest land use from 1990 to 2000 and 2005 at regional, ecozone, and global levels.

LANDSAT images were used as the main image platforms because these provide spatially consistent coverage for most of the world. A systematic sampling design based on each longitude and latitude intersection was implemented for 13,689 sites) and each sample tile covers a 10 by 10 kilometer square.

The image processing includes segmentation of the images into polygons based on similar satellite image characteristics and labeling these following a simplified form of the FAO land cover and land use classifications.

Click here to enlarge picture

The regional and country experts will then provide validation and correction of the pre-labeled attributes.

After the validation, the FAO-RSS unit will compile the data to estimate trajectories of forest cover and land use changes at regional and global scales. The data will not be used for country-specific analysis because the national statistics provided by the national correspondents (in the Philippines' case, the DENR-FMB) will be the official results.

Apart from the validation results, the organizers are encouraging the participants to use, modify and adapt the methodology within their own forest monitoring systems.

For further details, please go to FAO Forestry

Philippines datasets

Andres Ignacio already completed the initial work on land cover in Ispra, Italy in June 2010. For the land use validation, Maning was with Dr. Nel Bantayan of the College of Forestry and Natural Resources, UPLB and they divided the sample sites/tiles and worked independently.

There were 32 sampling tiles, and 23 have three LANDSAT images (ca 1990, 2000, 2005 and beyond) while the rest have only two images (ca 1990 and 2000).

Since bulk of the work was on land cover already completed by Andres, the land use validation activity in Thailand was straightforward. Some minor corrections had to be done, especially for dense coconut plantations, which in some cases were pre-labeled as forest. But these misclassifications were fairly minimal. Overall, the validation was simple because of the intuitive software used and the general familiarity with the sampling areas and image platform. Moreover, the classified forest polygons are mostly similar to the ESSC 2002 forest cover classification.

The FAO-RSS Unit is expected to consolidate the global and regional statistics and the final report will be available by mid-year of 2011.

A web portal was also created to provide public access to the global sample tiles used in the study. Through this facility, the public can download the sample tiles and conduct their own analysis.