Parc de la Cimaise – Immeuble I – 27 rue du Carrousel – 59650 VILLENEUVE D’ASCQ 03.20.72.53.64 - 03.20.98.05.78 - E-Mail : [email protected]- Site Internet : www.sirs-fr.com S.A.S. au capital de 312.025 € - RCS LILLE B 444654271 - APE 6311 Z - N° d’identification FR 07444654271 - SIRET 444654271 00022 Systèmes d’Information à Référence Spatiale Dr Christophe SANNIER Head of Research & Innovation [email protected]Suitability of Global Forest Change data to report forest cover estimates at national level in Gabon Expert workshop on using global datasets for national REDD+ measuring and monitoring Christophe Sannier, Ron McRoberts & Louis-Vincent Fichet 9-10 November 2015, Wageningen, The Netherlands
22
Embed
Suitability of Global Forest Change data to report forest cover … · 2015. 11. 18. · [email protected] Suitability of Global Forest Change data to report forest cover
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Parc de la Cimaise – Immeuble I – 27 rue du Carrousel – 59650 VILLENEUVE D’ASCQ 03.20.72.53.64 - 03.20.98.05.78 - E-Mail : [email protected] - Site Internet : www.sirs-fr.com
S.A.S. au capital de 312.025 € - RCS LILLE B 444654271 - APE 6311 Z - N° d’identification FR 07444654271 - SIRET 444654271 00022
Systèmes d’Information à Référence Spatiale
Dr Christophe SANNIER Head of Research & Innovation [email protected]
Suitability of Global Forest Change data to report forest cover estimates at national level in Gabon
Expert workshop on using global datasets for national REDD+ measuring and monitoring
Christophe Sannier, Ron McRoberts & Louis-Vincent Fichet
• To develop the precursor of an operational national Forest monitoring system for Gabon
• Project funded by and coordinated
by
• is coordinating Gabon
activities in collaboration with AGEOS • Complete Coverage of Gabon (267,000km²)
in 1990, 2000, 2010 and now 2015 with production in Gabon by AGEOS
• Setting up wall to wall mapping requires substantial efforts
• Can global data sets be used at national level?
National Forest Cover and Forest Cover Change Mapping for Gabon
Forest definition: Minimum 1ha area, 30% Crown cover and 5m height at maturity
Objectives
• to assess whether the UMD GFC dataset can be processed to match the selected national forest definition for Gabon
• to determine the degree to which estimates of forest cover and forest cover change and their associated uncertainties are enhanced using the UMD GFC dataset relative to using reference data alone
• to quantify the loss, if any, of accuracy and precision resulting from using UMD GFC maps rather than a nationally produced map
• to determine the level of additional effort in terms of increased sample size and/or post-processing of the UMD GFC data that would be required to obtain comparable estimates with respect to accuracy and precision to estimates based on a nationally produced map
• to develop guidelines on how to use global products for national reporting
Page 4
Page 5
Accuracy assessment
Unaligned systematic area frame sample
Independant Interpretation of
sample sites
Comparaison SPOT 5 Landsat 7
Forest Cover Map 2010 accuracy
Fichet LV, Sannier C, Massard K. Makaga E., Seyler F (2014) Assessing the Accuracy of Forest Cover Map for 1990, 2000 and 2010 at National Scale in Gabon. IEEE Journal of Selected Topics in Applied Earth Observation and Remote Sensing. 7, 1346 - 1356. http://dx.doi.org/10.1109/JSTARS.2013.2271845
• Forest cover and forest cover change estimates can be produced based on samples alone (Direct estimate)
• Observations from reference samples and the map can be combined to improve the precision of estimates (Model Assisted Regression):
Sannier C, McRoberts R A, Fichet LV and Massard K. Makaga R. (2014) Using the regression estimator with Landsat data to estimate proportion forest cover and net proportion deforestation in Gabon. ForestSat 2012 Special Issue. Remote Sensing of Environment. http://dx.doi.org/10.1016/j.rse.2013.09.015
Direct Expansion Estimate Map area statistics Regression Estimate
Gabon Total Forest Area ≈ 23.6 million ha Gabon Total Area = 26.7 million ha
Gabon Forest Cover Change Estimates (1)
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
1990-2000 2000-2010
Gross Deforestation between 0.31 and 0.52%
Direct Expansion Estimate Map area statistics Regression Estimate
Gabon Forest Cover Change Estimates (2)
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
1990-2000 2000-2010
Regeneration between 0.16 and 0.22%
Direct Expansion Estimate Map area statistics Regression Estimate
Gabon Forest Cover Change Estimates (3)
-20,000
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
1990-2000 2000-2010
Net Change between 0.10 and 0.34%
Direct Expansion Estimate Map area statistics Regression Estimate
Processing of UMD GFC percent tree cover and loss/gain data for Gabon
• Global coverage • 30m Spatial resolution Percent Tree
Cover product for 2000 • Yearly loss/gain of forest cover at 30m
resolution • GFC data tiles downloaded, mosaicked
re-projected and clipped to Gabon map projection and boundaries
• GFC data processed to match selected Forest definition for Gabon
UMD GFC Data TCD 30% Threshold 1ha MMU
Sannier C, McRoberts R A and Fichet LV (2015) Suitability of Global Forest Change data to report forest cover estimates at national level in Gabon. ForestSat 2014 Special Issue. Remote Sensing of Environment. http://dx.doi.org/10.1016/j.rse.2015.10.032
• UMD GFC data set can be processed for producing national forest cover estimates
• Accuracy of estimates likely to be lower than that from nationally produced data sets
• 95% Confidence Intervals larger by a factor of 2-3 compared with that of national data for forest cover area
• Appropriate Level of precision to be related to ratio between additional costs and benefits generated by the increased precision from a performance based payment scheme
• Cost of processing UMD GFC data likely to be substantially lower than national map
• Comparison of change estimates do not provide any substantial improvement compared with direct estimates due to small magnitude of change: estimates are not significantly different from 0
• Bias and 95%CI for UMD GFC data can be further reduced if adjusted for local conditions