Uncertainty [MGD Sections]

UNFCCC decisions and requirements
IPCC good practice guidance
Relationship to UNFCCC
GHGI coverage, approaches, methods and tiers
Design decisions relevant to national forest monitoring systems
Land cover, land use and stratification
Forest reference emission levels and forest reference levels
Quality assurance and quality control
Guiding principles – Requirements and design decisions
Estimation methods for REDD+ activities
Integration frameworks for estimating emission and removals
Selecting an integration framework
Activity data x emission/removal factor tools
Fully integrated tools
Practical considerations in choosing an integration tool
Guiding principles – Methods and approaches
Remote sensing observations
Coarse resolution optical data
Medium resolution optical data
High resolution optical data
L-band Synthetic aperture radar
C-band and X-band SAR
Global forest cover change datasets
Ground-based observations
National forest inventories
Auxiliary data
Guiding principles – Remote sensing and ground-based observations
Activity data
Methods for estimating activity data
Maps of forest/non-forest, land use, or forest stratification
Detecting areas of change
Additional map products from remote sensing
Estimating uncertainty of area and change in area
Estimating total emissions/removals and its uncertainty
REDD+ requirements and procedures
Reporting forest reference emission levels and forest reference levels
Technical assessment of forest reference emission levels and forest reference levels
Reporting results of REDD+ activities
Technical analysis of the REDD+ annex to the BUR
Additional advice on REDD+ reporting and verification
Guiding Principles – Reporting and verification of emissions and removals
Financial considerations
Country examples – Tier 3 integration
Use of global forest change map data
Relative efficiencies
Developing and using allometric models to estimate biomass

Record Keeping [MGD Sections]

Integration + Estimation [MGD Sections]

Ground Based Observations [MGD Sections]

Appendix D   Use of global forest change map data Previous topic Parent topic Child topic Next topic

Case Study - Guyana
Many tropical countries like Guyana are interested in establishing whether the UMD Global Forest Change products provide a useful and reliable source of information that can be used or adapted to satisfy reporting requirements, particularly for REDD+ or other programs that involve results-based payment for the avoidance of deforestation. Of particular interest to Guyana are the freely available Global Forest Maps provided by the UMD for estimating annual forest gains and losses. The Guyana Forestry Commission is interested in the value of these data in the context of REDD+ and specifically how statistics from the global products relate to gross deforestation.
Guyana undertook a careful assessment of the accuracy of the UMD Global maps by comparing these data with maps produced by its own MRV system, and with independent reference data. Global Forest Change Maps 2010-2104 were provided by the UMD (Hansen et al., 2013) and compared with the Guyana Forestry Commission’s forest change maps (MRV) which are based on careful interpretation of nationwide satellite image coverage. The Guyana MRV system had used 30m-Landsat TM and ETM+ imagery for the five epochs: 1990, 2000, 2005, 2009 and 2010. For 2011, 2012, 2013 and 2014 the Guyana MRV system used 6.5m-RapidEye imagery. In accordance with the Marrakesh Accords (UNFCCC, 2001), Guyana has elected to classify land as forest if it meets the criteria of minimum tree cover of 30%, a minimum tree canopy height of 5 m and a minimum area of 1 ha. The forest area was mapped by the Guyana Forestry Commission by excluding non-forest land cover types, such as water bodies, infrastructure, mining and non-forest vegetation. The non-forest land cover types are classified into six broad land use categories in accordance with IPCC reporting guidelines.
The UMD Global Forest Change data provides an estimate of tree cover percentage for each 30m-Landsat pixel (Hansen et al., 2013). The first step is to identify the percentage cover value in the UMD Global product that best corresponds to actual forest cover in Guyana.
The second step is to analyse the forest change (loss/gain) in the global data product and compare that with Guyana’s annual forest loss maps from the MRV system for the whole country.
The third assessment uses independent high quality reference data to assess the accuracy of both the UMD Global Forest Change and the Guyana MRV system data based on a two-stage stratified random sampling design with 143 (5 km by 15 km) first stage samples. The reference data consisted of 0.25 m aerial imagery and some independent reinterpretation of 6.5 m RapidEye scenes.
Figure 15 (a) shows Guyana’s non-forest area as a percentage plotted against the different tree cover percentage thresholds in the UMD Global Forest Change product. The result indicates that a threshold tree cover percentage of 90% provides the closest correspondence Guyana non-forest area as at 2000. This analysis shows that a user should not assume that forest definitions on the ground correspond well with the percentage canopy cover reported by the algorithm used in the global forest cover product. Using Guyana’s forest definition (30% canopy cover) results in an underestimation of non-forest and an overestimation of forest land cover (see Figure 15a).
Figure 15(b) shows estimates of annual forest loss from the UMD Global product and Guyana’s MRV system for years 2010 to 2014 inclusive. For years 2010 and 2011 Landsat data were used by both Guyana MRV system and UMD Global product to assess change. Comparatively the estimates of forest loss look broadly similar. For years 2012, 2013 and 2014 the UMD global product appears to underestimate forest loss when compared with the Guyana MRV system data. Over this period Guyana switched to the interpretation of 6.5 m RapidEye imagery.
Figure 15: Comparison of UMD global map and Guyana maps
Accuracy assessment
The results of an analysis of mapping accuracy are shown in Table 21. The analysis is based on a probability-based sample that uses an independent reference data to assess accuracy of both the UMD global forest change data and the Guyana MRV system data.
The UMD data gives a User’s accuracy of 94% and a Producer’s accuracy of 73%. This compares less favourably than the user’s and producer’s accuracies of 80% or better reported by Hansen et al., (2013) per climate domain and for the globe as a whole. The equivalent User’s and Producer’s accuracies for the Guyana MRV system data are 99% and 99.9% showing, as expected, that a deforestation estimate derived from the interpretation of high spatial resolution data is more accurate (Table 21) than a global map product.

Table 21: Error matrix between UMD global product and Guyana maps and reference data

Error matrix
Reference data
Non forest
User accuracy
Maryland map
Non forest
Producer accuracy
MRV map
Producer accuracy
Overall the UMD Global forest map products are easily accessed and provide a rapid spatial overview of forest area and forest change patterns across Guyana. When taken over the 14 years (2001-2014) of available data, the national-scale statistics for forest area and forest change match well with Guyana’s MRV system.
The ability of the UMD Global products to provide indicative maps of loss and gain is useful as it indicates where change occurs. Even if change is due to natural processes it is valuable to be able to visualise change on a year on year basis as shown in Figure 16.
There are however, some limitations. This is observed in Figure 16 which compares the area of annual change from the UMD Global product against Guyana MRV system mapped from RapidEye. It is clear that for any one year the Maryland global data does not capture all of the change. In essence this means the UMD Global product could not be relied upon to provide an accurate annual loss / gain or rate of change statistic.
This is to be expected and is one of the reasons that the MGD advocates the use of reference data to adjust for mapping errors when quantitative estimates of forest area or rates of change are required.
For Guyana, the UMD forest change maps underrepresent the patterns of gross deforestation in comparison with reference data. In recent years the greatest amount of forest loss has been associated with alluvial gold mining and mining roads and other infrastructure. In this regard the pixel size of Landsat is too large to capture small area change such as ribbon-mining infrastructure (Figure 16).
The inability to distinguish between anthropogenic loss (deforestation) and loss from natural processes is a limitation in the context of REDD+ where the causes of change is important in assigning emission/removal factors to activity data.
Figure 16: Illustration of forest area in Guyana
Forest loss has been mapped in the UMD Global product (left) and MRV system (right) maps between pre-2001 and 2014
In the context of REDD+, activity data may be associated with particular definitions. The UMD Global data does not take into account forest definitions such as areas more or less than a hectare. For example shifting cultivation and other types of forest degradation that appear as forest loss are mapped as deforestation in the UMD Global map. This results in an overestimation in forest change and carbon emissions from deforestation. Further country-specific definitions that operate below the minimum mapping unit of 1 ha may further limit the utility of UMD Global map product.
For Guyana there are areas that suffer from persistent cloud cover and that appears to be reflected in the UMD Global product for some parts of the country, especially along the Caribbean coastline. It is difficult to know how many Landsat scenes have been used for the change (loss/gain) detection and precisely from which time period these scenes are taken. Again, this limits the use of the UMD global products in annual REDD+ activity data monitoring.
Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A., Chini, L., Justice, C.O., & Townshend, J.R.G. (2013). High-Resolution Global Maps of 21st-Century Forest Cover Change. Science. 342:850–853. Data is available online Opens in new window.