Project aims for Forest Dragon 3:

1) Scaling effects in forest ecosystems mapping

2) Long-term analysis of GSV & forest structure over northeast China, for the years 2005-2010

3) Linking Forest Dragon products with existing products: CLC/ NLCD/ MODIS Time Series

⇒ Production of validated GSV-Change-maps

Thus aims were achieved by solving following tasks:

Task 1) Pure Forest Pixel Investigation

Task 2) Mixed Pixel Investigation

Task 3) Plausibility Check with existing Products

Approach Task 1:

Pure Forest Pixel Investigation.

Pure Pixel Analysis, based on the combination of Globeland 2010, Corine LC 2005 & 2010 and MODIS LC 2005-2010. Distinction between forest and non-forest pixel to extract stable pure forest pixel. Inter-yearly comparison of BIOMASAR GSV values, for stable pure forest pixel, based on the BIOMASAR GSV Maps 2005 & 2010. The Outcome was the detection & extraction of high difference GSV pairs for pure forest pixel. This analysis aims to serve as a cross-validation of the BIOMASAR maps using global land cover products. By combining information of four products, classification errors introduced by land cover products get minimized.

Approach Task 2:

Mixed Pixel Investigation.

Mixed Pixel Analysis, based on Corine LC 2005 & 2010 to compute the percentual forest mixture combination within each 1km BIOMASAR GSV pixel. Following mixed-pixel classes were chosen: forest-crop, forest-grass and forest-shrub. As well as an agreementcheck with the GSV-Values from BIOMASAR GSV – Maps 2005 & 2010. In this analysis we compared GSV distributions of BIOMASAR pixels containing a mixture of forest and other classes related to vegetation (shrub, grass, crop). The goal was to investigate influences of vegetation related classes in combination with forest regarding GSV derivation in C-band, which is used in the BIOMASAR algorithm.

Approach Task 3:

Plausibility check.

The objective of the plausibility check was to find an Agreement or Disagreement for the change of GSV in the BIOMASAR GSV Map 2005-2010, with other landcover-change products. The investigation was concentrated on areas with a percentual change over 35% in the GSV change map of 2005–2010. These areas were compared with vegetation products: MODIS NDVI 2005-2010, MODIS VCF 2005-2010, as well as with fire products: MODIS burned Area 2005-2010, CAF Burned Area 2005-2010, to explain the GSV-change from 2005 – 2010.