Period: 2000, 2005, 2010, 2015

Sensor: Landsat 5 TM/ETM+

Frequency: 5 yearly

Resolutions: 30m

Extent: Global

Data can be accessed here.

Algorithm description: The Landsat Vegetation Continuous fields (VCF) tree cover maps provide estimates of percentage of horizontal ground covered by woody vegetation greater than 5m in height.

The algorithm is based on Sexton et al (2013). The tree cover is estimated as a piecewise linear function of surface reflectance and temperature. MODIS 250m tree cover estimates (2000 to 2005) and Landsat surface reflectance products are utilised to generate training datasets. MODIS crop cover product is used as an additional information to improve representation of agriculture areas. The  regression model was developed using Cubist regression tree algorithm.

Accuracy: Sexton et al  (2013) compared the product with LIDAR based tree cover estimates over four sites (one in Costa Rica and three from US). Considering all sites together, the Landsat products had R2 of 0.81 and RMSE of 17%. Song et al (2015) report overall accuracy of 82%, producer’s and user’s accuracy of 80% and 91% respectively, over forested parts of two counties in the State of Maryland, US. Global assessment of the product has been done by Feng et al (2016), who compared the product against 27,988 independent reference points visually interpreted by experts. Overall accuracy of static forest layers was 91% and that of change was >81%.

Limitations: 

  • Product was found to be less accurate in sparse forests and savannahs.
  • Areas that had ~30% forest cover mainly affected as this value was set as a threshold for forest -non forest discrimination.

Reference:

Sexton, J. O. et al. Global, 30-m resolution continuous fields of tree cover: Landsat-based rescaling of MODIS vegetation continuous fields with lidar-based estimates of error. Int. J. Digit. Earth 6, 427–448 (2013).

Song, X. P. & Tang, H. Accuracy assessment of Landsat-derived continuous fields of tree cover products using airborne LIDAR data in the Eastern United States. ISPRS – Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. XL-7/W4, 241–246 (2015).

Feng, M. et al. Earth science data records of global forest cover and change: Assessment of accuracy in 1990, 2000, and 2005 epochs. Remote Sens. Environ. 184, 73–85 (2016).