A novel methodology using an object-oriented approach, based on initial image segmentation followed by a rule-based classification, was developed and applied to Quickbird, SPOT and Landsat imagery. The methodology created was considered to be satisfactory for both single-year thematic mapping as well as for multi-year change mapping.
As for the land cover change maps and associated statistics it was found that the re-plantation project has resulted in a 220 ha increase in new plantation area inside the project area in the period from the project start in 2007 to the beginning of 2008. This increase is less than the project target of re-planting 750 ha new plantations per year until the termination of the project in 2011. It is, however, merely a status assessment and no indicator for whether the project is off target since new plantations will undoubtedly be established for the remainder of 2008. It was also found that new plantations are being replanted outside the EIB re-plantation area albeit less significantly. Moreover, there is evidence that Kolombangara as a whole is being subject to a gradual increase in human activities which may have implications for sustainable forestry management.

Re-plantation Monitoring in Kolombangara Forests


The EIB is supporting a re-plantation project on Kolombangara, a small island located in the Western Province of the Solomon Islands.  Re-planting started in 2007 and will continue until 2011. Objective of this service case was to provide the EIB with information on the re-plantation progress and on fragmentation of the forests.  Most recent (2008) and historic Earth Observation data before start of re-planting (2005/04) needed to be acquired to set the baseline and to enable land cover change detection.  Expected results of this trial are a recent orthophoto map of the entire island, land cover & change maps & area statistics and fragmentation metrics (edge density) covering both, historic and recent dates.   

Executive Summary (extracted from the Service Operation Report prepared by the service provider GRAS)

An approach combining satellite sensors at various scales and multi-temporal analysis was adopted to provide up-to-date information about the land cover distribution as well as to observe the land cover evolution following the start of the replantation project in 2007.


Assessment of impact and benefit by the user


Plus:

  • Previous data on this tiny island were very limited: only a paper map was at EIB disposal; the promoter had some GIS data but these were inadequate due to a recent earthquake and were in the process of being updated. The maps which were developed during the course of this service represent important progress.
  • EO data and derived products provide
  1. sufficient resolution of the maps to detect main land cover changes and to verify progress of planting within the EIB project area and the entire island - providing insights into the forestry practices being implemented within the project area and how these may influence what is taking place elsewhere;
  2. objective monitoring of changes over time, both within and around project areas;
  3. useful insight allowing EIB to see patterns of change regarding species composition: distinction between ‘old’ plantations and ‘new’ plantations was made as a proxy for tree species (‘old’: higher percentage of white teak; ‘new’: higher percentage of eucalyptus);
  4. information about areas which may be very large, very remote and difficult to reach (eg Kolombangara), or which are very difficult to check in their entirety (linear infrastructure)
  • Satellite imagery is an excellent communication tool which can allow the EIB to visualize and to demonstrate the benefits of its investments, providing important added value to our work as a public institution
  • The service would be very useful for providing an overview of changes / impacts which allows us to then better target our monitoring resources, or to decide whether more monitoring is needed. Such a monitoring overview would be potentially useful to many projects that the Bank finances.
  • Procedures applied by LuxSpace  in cooperation with the EIB during the service were satisfactory within resource constraints. Administration and coordination by LuxSpace between the various partners was excellent.

 
Minus:

  • The analogue integration of the information (using the report, not the geo data) is assessed as “medium” (out of 5 marks).
  • The conclusions provided in the report are very general, and the technical details too much for a non-expert in Remote Sensing. The user expected a more precise time window as it was possible but this was unrealistic in this case, due to cloud coverage in the tropics. It is felt that the work would have benefited from the involvement of a sectoral expert. The definitions of ‘grassland’ and ‘forest’ seem to allow the possibility of very young plantations being mistakenly identified as grassland.  Overall, the report had too much emphasis on technical aspects what is less useful or helpful for us – we were interested in the results, and what this means for project implementation and impacts.
  • This project focused on project implementation over quite a short period of time. We feel that satellite information would be better suited to the monitoring of medium to long-term project impacts. For monitoring implementation over short periods of time, sufficient precision seems to be difficult to achieve.
  • LuxSpace  made the process as straightforward as possible by translating EIB’s user requirements into the technical specifications necessary to launch the study. However, the major challenge is co-ordination. Fortunately, a fruitful co-operation was achieved. Despite the need for considerable resources for this new activity, the added value and enhanced quality which such services can bring to the EIB’s work are likely to make this worthwhile.
  • Issues:
  1. Delayed delivery, although the reason is clear (waiting for suitable satellite image);
  2. Cloud cover or distortions (eg Landsat linear distortion) cause problems. It is important to manage the users’ expectations in this regard! Expectations may have been too high - cloud cover was a big and rather unexpected obstacle for this study.