Satellite Monitoring of Forest Degradation (Activity Data) and Estimation of Related Carbon Emissions
Franke, Jonas1; Jubanski, Juilson1; Ballhorn, Uwe1; Englhart, Sandra1; Siegert, Florian2
1RSS - Remote Sensing Solutions GmbH, GERMANY; 2RSS - Remote Sensing Solutions GmbH; GeoBio Center, Ludwig-Maximilians-University Munich, GERMANY
A successful implementation of climate change mitigation efforts and biodiversity conservation in tropical rainforests necessitates reliable remote sensing based forest monitoring capabilities with minimized uncertainties. Such monitoring has to be capable of detecting rapid changes in forest extent, i.e. deforestation, and subtle changes due to forest degradation by selective logging and fire. To improve current emission estimates from forest cover change for climate change mitigation efforts such as REDD (Reducing Emissions from Deforestation and forest Degradation), the accurate monitoring of changes in carbon stock is essential. This requires first a timely high-resolution monitoring of forest cover changes, i.e. activity data, and second, a reliable estimation of related carbon emissions.
In the present study, results from different REDD demonstration activities in Indonesia were combined and a comprehensive analysis of the varying approaches for estimating carbon emissions due to forest disturbances was conducted. Forest inventory data, high-resolution optical satellite data as well as airborne LiDAR (Light Detection And Ranging) data from various REDD projects were used to compare different methods of aboveground biomass (AGB) estimation and to analyze the variability of the estimates.
By the use of the forest monitoring system developed in the EU FP7 project REDD-FLAME (Fast Logging Assessment and Monitoring Environment), activity data on deforestation and forest degradation in Central Kalimantan, Indonesia was generated. This system was demonstrated by the use of RapidEye data, but is primarily designed to use future Sentinel-2 data. The monitoring system is capable of detecting even small scale forest disturbances due to illegal logging and was successfully implemented in Indonesia, Brazil and Mozambique. In addition, forest inventory data from REDD demonstration projects were used as a basis for AGB modelling of larger areas using airborne LiDAR data. A large area LiDAR biomass model was generated and combined with activity data from the REDD-FLAME monitoring system, in order to assess the AGB variability across primary forest, secondary forest and illegally logged forest. This allowed an in-depth analysis of the carbon emissions due to forest degradation and also demonstrated the wide intra-variability of biomass values within a forest type. For example, peat swamp forests showed mean AGB values between 220.1 and 246.1 t/ha, with a higher standard deviation for secondary forests (±54.7 t/ha) in comparison to primary forests (±32.9 t/ha). Illegally logged peat swamp forest showed a mean AGB value of only 169.4 t/ha (±59.2 t/ha). Small-scale illegal logging of peat swamp forests thus caused a reduction of AGB of about 50 to 76 t/ha.
The results of the forest inventory/LiDAR-based approach were compared to results of the stratify & multiply approach. It could be demonstrated that the use of LiDAR data drastically reduces uncertainties of large area biomass estimates and thus makes more reliable carbon emission estimates possible. By the use of LiDAR-based biomass estimates, accurate forest carbon stock maps can be derived that consider forest-type dependent variabilities. By the additional use of activity data, the carbon emissions due to small-scale illegal logging and fire could be estimated. This combined technique using forest inventory data, optical satellite data (e.g. Sentinel-2) and LiDAR data is currently being implemented in other REDD projects in Indonesia.
[For consideration for the special REDD Session]