Skip to main content

Reducing Emissions from Deforestation and Degradation in Africa

Final Report Summary - REDDAF (Reducing Emissions from Deforestation and Degradation in Africa)

Executive Summary:
EO application is essential in providing spatial information related to carbon stock changes. The REDDAF Project achieved technical results in the application of optical and SAR EO data for the monitoring of forest cover, land cover, forest cover changes and biomass measurements. Key aspects of the research have already supported the methods used to successfully produce the forest cover and forest cover changes products in Cameroon and CAR. Mapping changes in land use can be achieved with more than 90% accuracy using optical and/or RADAR data, mapping changes of forest land remaining forest remains problematic due to the spatial nature of the mapped feature and the lack of a common definition. It is also worthwhile noting that the uncertainty in distinguishing some of the IPCC land cover categories still remains high. EO also contributes to estimating emission factor. Current techniques are still in research but massive strides towards operationality are expected as soon as the P-band BIOMASS satellite is launched.
REDD+ countries are expected to develop national MRV systems inter alia. A critical issue in preparing countries for the REDD+ process is to ensure that the MRV technologies and methods developed are transferred to the countries. Capacity building – institutional, infrastructure and human resources – is crucial. Partnerships between European research institutions and service providers with the partner institutions from the User community ensured that the REDDAF products and services were effectively taken up. Nevertheless, much still needs to be done in enhancing capacity for the establishment of national MRV systems in both Cameroon and CAR.

Project Context and Objectives:
The United Nations Framework Convention on Climate Change (UNFCCC), Conference of the Parties (COP) requests REDD+ countries to establish robust MRV systems to account for emissions from deforestation, forest degradation and the role of conservation, sustainable forest management and enhancement of carbon stocks. The role of EO and in-situ measurements has been underscored as primordial tools for the evaluation of activity data and emission factors that are fundamental inputs for the estimation of Green House Gases (GHG). The objective of the REDDAF project was to test and provide improved methodologies using both optical and radar EO data for deforestation/degradation assessment and above ground biomass estimation for monitoring carbon stock changes in two tropical countries actively involved in the REDD process: Cameroon and Central African Republic (CAR). The improved methodologies are intended to support the development of operational service chains which can be scaled up to national level and therefore contribute to the current national REDD+ Measuring, Reporting and Verification (MRV) activities in these countries. The project emphasized the involvement of the user community and the subsequent transfer of the developed methods and techniques. The UNFCCC Focal Points in both Cameroon and CAR hosted in the Ministries of Environment, Nature Protection and Sustainable Development (MINEPDED) and Ministry of Environment and Ecology (MEE) respectively endorsed and supported the REDDAF concept and project implementation. It should be noted that the REDDAF project has a strong legacy in the GMES Service Element on Forest Monitoring (GSE FM) REDD Project in Cameroon which was implemented 2008-2010 and formulated several research and development needs.

Project Results:
The REDDAF project was conceptualised around five main tasks: User Requirements which provided the national context for project implementation; Methods Development to offer cost-effective methods with known certainty levels for products and services development; Service Development and Integration to develop customised products and services that are compliant to international standards and norms and can be easily integrated into the working cycles of the Users; Validation and Proof of Concept to demonstrate and provide evidence that the developed services and products are fit for purpose and lastly Dissemination and Training to ensure that the methods and techniques developed are transferred to the User and to a wider external audience through seminars, workshops, publications etc. This Section will present the main achievements of these Tasks.
1.2.1 User Requirements
Cameroon and CAR are still at the initial phase of the REDD+ MRV developments, characterised by institutional arrangements, identification of roles and responsibilities, capacity needs assessment and capacity building. The technical needs and standards for implementation of a national REDD+ MRV were assessed in both countries to elaborate the technical specifications of the REDDAF services. The existing work practices, current geospatial capacity and the range of standards and protocols on information exchange/dissemination that the REDDAF service portfolio must comply in both countries were also analysed. A training needs assessment was conducted in both countries as basis for the establishment of training modules. Based on this information Service Level Agreements (SLA) with the intended services and products was signed between the REDDAF Service Providers and the Users in Cameroon and Central African Republic respectively.
There are currently no cartographic norms, national definitions of forest and other key categories. Most geo-spatial standards or quality assurance systems are done on a project basis. In the absence of national norms and standards it was agreed that the REDDAF products and services should be compliant to international standards for forest cover mapping in REDD+ as explained in the IPCC Good Practice Guidelines for AFOLU (2006) and the GOFC GOLD REDD Source book (2011).

1.2.2 Methods Development
Methods development in REDDAF was built on the findings from previous pre-operational EO-based forest mapping projects which identified the following challenges:
• the problem with frequent cloud cover in the region which hampers the use of the optical data
• the use of multi-scale and multi-senor data for forest monitoring
• no operational methodologies for degradation assessment and mapping
• no operational methods for the use of remote sensing for direct biomass assessment and mapping

Consequently, the REDDAF project sought to address these issues.
1. Improving EO-based Forest Cover Mapping
The continuous impairment of optical signals by clouds in the tropics has necessitated the use of multi-sensor, multi-temporal datasets for effective wall-to-wall coverage of tropical countries. This requires robust, cost-effective pre-processing techniques. Three main pre-processing issues were investigated and techniques developed and tested under pre-operational conditions. The first was related to cloud shadow compensation. The cloud shadow algorithm developed, radiometrically corrects cloud shadow areas and consequently increases the amount of interpretable pixels which otherwise will be discarded. A second pre-processing step which is the block-based radiometric adjustment improves the similarity between neighbouring images since multi-sensor, multi-temporal images are required for production. Spectral analysis shows that areas belonging to the same land cover class form more compact clusters in the scatter plots after the adjustment than before. Thus large image blocks are adjusted to similar grey values, permitting cheaper and more accurate classification of image mosaics. Another important aspect of pre-processing for the integration of optical and SAR data, required the development of a robust and fully automatic matching procedure to ensure geometrical congruence between optical and SAR data sets.
2. EO Methods for Forest Degradation
The complex task of assessing forest degradation is being addressed in the REDDAF Project. The following methods were developed and tested: (i) an approach based on the calculation of NDVI from optical data mosaics to map disturbances due to selective logging; (ii) an improved SMA to track to detect disturbance and track forest regeneration signals (Figure 1). A major challenge in mapping forest disturbance in the tropics is that the signals are lost quickly; (iii) identifying canopy gaps as an indicator of forest degradation from Cosmo Skymed stereo data.

Figure 1: Degradation product of a test site in Central African Republic.
3. Direct EO Biomass Assessment
The activities related to direct biomass assessment with SAR data had as objective to provide a transferrable methodology to deliver an EO based product of Above Ground Biomass (AGB). The approach focused on improving existing image processing and biomass inversion methods to fully exploit the potential of the currently available data (ALOS PALSAR-L-band) and developing generalized models that are easily transferable to other regions. Reference plots of 1 ha were identified and biomass field measurements for each plot were carried out during two successive missions in the Adamawa and Centre region of Cameroon. The plots for the first mission were used for the calibration of the model while the second mission served for model validation. The improvement of the methods beyond the state of the art in biomass mapping (in particular in Cameroon, as reported by Mitchard et al., 2011) include several aspects: improvements of the spatial resolution of the biomass map, based on an optimized use of multi-temporal and polarimetric data, reduction of the uncertainties in biomass for biomass values beyond 100-150 ton/ha by taking into account the perturbing effects which mask the sensitivity of the backscatter to biomass (e.g. topographic and temporal variation effects); and the use of a limited number of in-situ data for model calibration. A Bayesian inversion method is also used to improve the retrieval performance. Changes in biomass are observed in more details with ALOS PALSAR data from 2007, 2008, 2009 and 2010.
Figure 2 presents the preliminary biomass map of the study area. The map is derived from ALOS PALSAR 2007 data using an inversion model developed for L-band SAR data (Le Toan et al., 2004).

Figure 2: Biomass mapping results from inversion of dual-polarized ALOS –PALSAR data for the Adamawa region, Cameroon.
The map in Figure 2 highlights the dense humid forests (Mbam and Djerem National Park) with biomass less than 150 ton/ha, and the gallery forests in the savanna, with biomass lower than 100 ton/ha. It should be noted that large areas of gallery forests and transitional forests (with biomass reaching 60-80 ton/ha) contain a large amount of carbon stocks, which is often neglected in carbon estimates.
1.2.3 Service Development and Integration
The objective of this Task was to integrate the findings of the methods development to develop products and services on an operational scale. The service area for Cameroon constituted the Centre (82 000 Km2) administrative region while for CAR it comprised the regions Ombella, Mpoko, Lobaye, and Sangha Mbaéré (87 000 Km2). The following products were developed:
• Forest cover maps: 1990, 2000, 2010
• Forest cover change maps: 1990-2000, 2000-2010
• IPCC compliant land cover maps: 2010
The products are based on optical data complemented by ALOS PALSAR radar to fill out residual gaps using techniques developed in Task 2. Optical datasets comprised principally Landsat TM4, TM5 for the historical periods (1990 and 2000) and RapidEye, SPOT4&5 for the most recent period (2010). A multi-temporal composite approach based on segments is used to interpret the images and create forest cover and forest cover change maps for the corresponding periods. A post processing approach which comprises a visual inspection is performed to reclassify segments that were initially wrongly classified in order to improve the accuracy of the final product. Figure 3 shows the IPCC compliant land cover maps (2010) for Cameroon and CAR.

Figure 3: IPCC compliant land cover maps (2010) for the demonstration areas in CAR (right) and Cameroon (left) based on HR optical satellite data (Landsat TM4/ TM5/ ETM, RapidEye, Deimos)
The accuracy assessment is based on an area frame sampling. 20x20 Km grids are overlaid over the service area and 2x2 Km rectangles are randomly created within each grid. Using all available datasets these 2x2 Km rectangles are interpreted by an independent interpreter. The accuracy is determined by creating 50 point samples within these rectangles and then comparing the products and the reference interpretation. The overall accuracy of the forest/non-forest maps is approximately 95%; for the IPCC compliant land cover maps 2010 approximately 75% for CAR and 94% for Cameroon; and for the forest cover change maps (2000 to 2010) approximately 77 % for CAR and 95% for Cameroon.

1.2.4 Validation and Proof of Concept
REDD+ reporting requirements will be guided by the general principles of transparency, consistency, comparability, completeness and accuracy. To assure the quality of the REDDAF products and services, an effective validation scheme was required. The basic premise of validation is to ensure that the products and services are fit for purpose (fit for the intended use); quality and efficacy are designed into the products and services; and that each step of the production process is controlled to assure that the finished product meets product specifications. A Validation Plan which provides the organisational structure of the validation process as well as reference to specific guideline documents, templates and information on the validation reference information that serve as a basis for the validation in REDDAF was elaborated. A three-stage process was drafted for REDDAF, comprising: (i) a First Party assessment checking the conformity of the achieved results with the previously agreed specifications in a self-assessment step by the Service Providers; (ii) a Second Party assessment undertaken by the Users: Ministry of Environment, Nature Protection and Sustainable Development (MINEPDED) in Cameroon, and Ministry of Environment and Ecology (MEE) in Central African Republic; (iii) a Third Party assessment performed by a body that is independent of both supplier and customer organisations. In the scope of REDDAF, this was done by the Geospatial Technology Group (GTG). The products were found to be compliant with specifications in IPCC and GOFC GOLD and would support Components 3 and 4 (MRV and Reference Scenario respectively) of the R-PP in both countries.

1.2.5 Dissemination and Training
This Task had the objective of ensuring that REDDAF had a wide-reaching impact on various stakeholders. The Capacity Building/Training transferred the methods developed to the counterparts in Cameroon and CAR. Trainings were organised in the form of theoretical courses and practical exercises based on the technical concepts used in the project and adapted to the needs, knowledge and skills of Users. The training material developed included case studies and exercises to build experience in service portfolio utilisation; demonstrations and information sessions on service chain operations, validation processes and data access conditions. Three training workshops on Remote Sensing, Satellite Image Interpretation and GIS were conducted per country. The training courses enhanced the knowledge of the participants on the concepts of REDD+ MRV, remote Sensing, image processing, terrestrial biomass inventory and GIS applications. Open source software SPRING, Q-GIS, BEAM, and Dinamica were used to ensure sustainability. As part of the workshop, field surveys were organised to procure ground truth information to improve the understanding of the land cover/use classes. Furthermore, sessions were organised with policy makers from both countries (climate change negotiations) in the sideline of SBSTAs and COPs to discuss the utility of the REDDAF services and products.
To ensure awareness of the REDDAF project and the corresponding outcome, a variety of dissemination activities were undertaken within the User and the global REDD+ communities. These include participation and presentation at User organised REDD+ fora, meetings with the respective Users during COPs and SBSTAs, as well as presentations at various global REDD-related meetings/fora. Furthermore co-ordination actions with other REDD projects for synergies was also implemented. A REDDAF project website has been developed and is constantly being updated to reflect the progress made in the project implementation. The website URL address is www.reddaf.info.

Potential Impact:
The REDDAF project produced significant scientific, technical, and social impacts. The scientific impact is illustrated by the number of peer-reviewed papers that have been published or still in press. Also, different outputs of the REDDAF project were presented and discussed with the scientific community during various scientific forums related to the REDDAF theme organised in the last three years.
Specific deliverables were intended to address particular audiences. The brief for policy makers and the white paper, which were intended to identify technical problems and research needs are designed to address scientists, practitioners, and policy makers. Capacity Building and technology transfer ensured that the methodological developments were transferred to different REDD+ stakeholder groups in the respective User communities. Participants in the capacity building workshops included technical staffs of government, representatives of some civil society institutions and research.
The project has also had relevant impacts on the MRV developments in both Cameroon and CAR as it adequately addresses some key issues that are outlined in Component 3 (Reference Scenario Development) and Component 4 (National MRV System) of the R-PP in both countries. The participation of REDDAF in this process resulted in the enhancement of future involvement of EU technologies, methods, and institutions in the REDD business of the Congo basin also contributing to improve awareness of the importance of EO and information technology as cross-sector decision support tools. The work developed by the consortium enabled a strong positioning in the process in cooperation with other national and international organizations.
Moreover, the relevance of REDDAF products is well accepted in CAR and Cameroon, which through their national UNFCCC focal points have manifested the intention of using REDDAF products for supporting other land use management sectors. REDDAF products have been used to feed the periodic report “Sate of the Congo Basin statistics”, which is an inter-governmental document of the Congo Basin countries.

List of Websites:
GAF AG, Germany (Project Coordinator)
E-mail: forestry@gaf.de

Project Website: www.reddaf.info