Training on national land cover classification systems Toward the integration of forest and other land use mapping activities. Guiana Shield 9 to 13 March 2015, Paramaribo, Suriname Background Sustainable management of human and environmental interactions is crucial for the functioning of economic, social and environmental processes. Land policies and measures are based mostly on national geospatial data and data on land quality such as e.g. land productivity. Many initiatives exist for land mapping, whether at local, national and regional level. In most cases, the estimates obtained are not consistent in space and in time, lacking comparability and robustness. In a context where land assessment is crucial to guide mitigation and adaptation to climate change, land use planning and food production, it is essential to support integrated efforts. Semantic interoperability is one of the unsolved problems in the use of land cover/use data. Uncertainty is an inevitable element in all types of geographical information which is always relative to context. Since 1996, FAO has been developing a way to approach the problem. A new version of the FAO land cover classification system (LCCS) will be released in 2015. LCCS places importance on standardizing attribute terminology rather than final categories. It works by creating a set of standard diagnostic attributes (called classifiers) to create/describe different land cover classes. The classifiers act as standardized building blocks and can be combined to describe the more complex semantics of each land cover class in any separate application ontology (classification system) (Ahlqvist,2008). The LCCS approach, in this way, is different from most of the other examples of standardized land cover systems, (e.g. Anderson CORINE etc.) which follow a single ontology approach where all semantic descriptions available have been created with a very similar view on a domain and have to be shared by all users (Lutz and Klien,2006). When reporting to international conventions and participating to voluntary performance-based mechanisms, the consistency of the information provided is a key principle. At current status, almost all countries possess land cover/ use maps that are inconsistent in time or in scale. Countries in the Guiana shields experience the same problems and issues. While the Guiana Shield is one of the largest blocks of primary tropical forest worldwide, pressures on natural ecosystems are increasing. Governments are keen to drive their development in a sustainable manner and in that perspective, they have shown strong interest in REDD+ as a mechanism that would enable to financially value their efforts, thanks to carbon revenues. The project REDD+ for the Guiana Shield was initiated, at first by Guyana, Suriname and France, shortly after joined by the state of Amapá in Brazil, at the occasion of the UNFCCC s fourteenth Conference of Parties, held in 2008 in Poznan. The project aims at providing information and tools to be used by countries to establish sound science-based policies and measures to tackle deforestation and degradation, in the framework of REDD+ mechanism. The 3rd working group meeting of the project was dedicated to the topic Design of a Multipurpose Page 1/6
National Forest Inventory. At this occasion, it was proposed to solve problems of stratification, definition and classifications, using the third version of the land cover classification system. The objective of this concept note is to introduce a regional training on LCCS3 to support the integration of forest and land use mapping activities in the Guiana Shield. Objectives The proposal aims at providing support to the REDD+ countries through the development of new approaches toward the implementation of land cover mapping activities. In particular, the following will be carried out: - Present and compare existing land cover classification systems: assessment of the possible inconsistencies and gaps performed; - Make recommendations based on existing experiences to improve comparability of different legends; - Explain the LCCSv3 methodology for creating a standardized legend. In particular the object based approach to land cover classification through the use of UML and LCML schema will be presented; - Introduce several examples of integration of different legends - Present automated approach for maps analysis based on LCCS3 methodology: ADG software; - Share training materials to support land use mapping activities in an integrated approach; - Discuss how best we can deliver capacity building and good advice to countries as regards the land cover mapping activities for REDD+. Activities to be performed Given its regional dimension, a general training will be organized taking into consideration the current Forest and Land Classification system in each participating territories (Guyana, Suriname, State of Amapá in Brazil and French Guiana). It will aim at identifying the main gap areas to harmonize the classification scheme followed at national level and also at comparing these national systems and discussing the setting up of a classification scheme that could be followed for regional monitoring studies. Some specific case studies will be used for describing the general concept. The main activities are listed below: A review: Introduction to land classification systems and forest definitions in the countries. Presentations carried out by the institutions invited (i.e. GFC - Guyana Forest Commission, SBB - Foundation of Forest Management and Production Control in Suriname, SEMA - Amapá State Secretariat of Environment, ONF-G - National Forests Office of French Guiana); Identification of constraints related to the harmonization of available data in the context of the establishment of a national system for monitoring forest cover and REDD+ activities; Use of LCCS3 and ADG in each country; Practical Exercises on LCCS3; General recommendations for a national and regional classification system that would allow the consideration of existing data in line with UNFCCC decisions and IPCC guidelines; Open discussions among participants and experts Page 2/6
Expected Outputs Updated knowledge on the current status of the land cover mapping activities and relative legends in the Guiana shield; Identification of the main gaps and inconsistencies between the different classification systems at national and regional scale; Increase the coherence and consistency of the current legends towards an integrated approach; Technical recommendations for the integration of existing national legends and mapping efforts; Capacity development on legend harmonization and classification improved. Requirements for the implementation of a training on LCCS The training workshop will need the following data from the institutions/participants involved. historical data on land cover classification systems existing land cover maps at national and sub-national level legends and the existing documentation that describe in detail the classes; develop a table with: Country / Scale / Name of legend/ Classes/ description/ Year/ adopted field calibration data and documentation available documentation on the methodologies used for the related mapping activities Expert Required: Dr. Antonio Di Gregorio (FAO) Dr. Rémi d Annunzio (FAO) Page 3/6
A detail workshop schedule has been developed after consultation with all UN-REDD Taskforce. Agenda REDD+ for the Guiana Shield Training on national land cover classification systems Toward the integration of forest and other land use mapping activities Tentative Agenda Venue: Paramaribo, Suriname 9 to 13 March 2015, Paramaribo, Suriname Participants: Staff members from the forest services of Surinam, Guyana, Amapá State and French Guiana. Up to 4 staff members of each country are welcome to attend. 1st day 8:30 Registration 9:00 9:10 Welcome and introduction of the training 9:10 9:30 Presentation of the data used for international reporting in the Guiana Shield (national communication to the UNFCCC, FAO-FRA, CBD, etc.). 9:30-10:50 Presentation of the different land cover mapping activities in the ONFI GFC, SBB, SEMA, ONF-G partner countries (20min each) 10:50 11:20 Introduction to the FAO Land Cover Classification System Antonio Di Gregorio 11:20 11:50 The Structure of the Land Cover Meta-Language Antonio Di Gregorio 11:50 12:30 Demonstration case study in Bangladesh Remi d Annunzio Lunch break 13:30 Land representation in the context of the IPCC Remi d Annunzio 14:00 Installation of LCCS3 By Group 15:00 Exercise 1: Creation of land cover classes using LCCS3 By Group 16:00 Exercise 2: Translation of national legends By Group 17:00 Group restitution By Group Page 4/6
2nd day 8:30 Summary of the first day ONFI 9:00 Principles of the Land Cover Classification System Antonio Di Gregorio 10:00 Image analysis processing chain and Advanced Database Gateway Antonio Di Gregorio 10:30 Coffee Break 11:00 Exercise 3: Development of legends using high resolution satellite By Group images (SPOT 5 imagery can be used for that purpose) 14:00 Exercise 4: Satellite image visual interpretation By Group 16:00 Restitution By Group 3rd day 8:30 Summary of the second day ONFI 9:00 Field Work collection of field inventory data following a transect 16:00 Data compilation and preparation 4 th day 8:30 Summary of the third day ONFI 9:00 Exercice 5: preparation of the classification system based on the high resolution images and field inventory measurement By Group 11:00 Exercice 6: Image visual interpretation for the field area surveyed By Group 12:00 Demonstration of the potential use of the results as training data for automatic image classification process comparison with a non LCCS based legend Rémi d Annunzio 14:00 Exercice 7: development of a land cover map for forestry purposes By Group 16:00 Exercice 8: Development of a land cover map for agriculture purposes By Group 17:00 Restitution By Group Page 5/6
5 th day 8:30 Summary of the fourth day ONFI 9:00 10:00 Preparation of a roadmap Remi d Annunzio 10:00 11:00 Integration of LCCS in the context of REDD+ Remi d Annunzio 11:00 11:30 Coffee break 14:00 16:00 Feedback from the participants Move forward: objectives and steps towards a harmonize approach for the land cover classification system 16:00 Closing ceremony Budget required: The estimated budget for this training is 25 591, including expertize and logistics. As the training will be provided in English some additional translation costs might be needed to facilitate the comprehension by all participants. Page 6/6