Period: 2000 – 2017

Sensor: MODIS Terra

Frequency: Yearly

Resolutions: 250m

Extent: Global

Data can be accessed here.

Algorithm description: The algorithm for MODIS VCF product is based on Hansen et al (2002, 2003). It is similar to that used for USGS Landsat 7 ETM+ global tree cover product. MODIS reflectance in all 7 bands, NDVI and surface temperature data are used to generate annual phenological metrics which consists of spectral-statistical quantities. These metrices are then used against high resolution LANDSAT based forest cover training datasets to build regression tree algorithm. Training dataset was based on visual interpretation of forest cover types from Landsat images (DeFries et al, 1998)

Accuracy: Gao et al (2014) found that the MODIS VCF could successfully differentiate between forest and non-forest areas in two areas in Mexico. Primary and Secondary forest were also delineated well. However secondary forests with regrowth showed higher percentage tree cover than primary forests. Tropical dry forest categories were getting mixed up with non-forest categories.

Gerard et al (2016) have raised concerns about possible bias in the VCF products that could result from a bias in the training datasets used to generate regression trees. This issue could be serious in the areas with asymmetrical forest distribution, areas with tree cover gradient, or combination of both.

White et al (2005) report strong positive bias in the tree cover product against forest inventory data for two sites in USA


  • Product has been found to be less accurate in dry deciduous and sparsely forested areas.
  • Accuracy is also questionable in areas with asymmetrical cover distributions and/or areas marked with transitions in forest cover types.


Hansen, M. . et al. Towards an operational MODIS continuous field of percent tree cover algorithm: examples using AVHRR and MODIS data. Remote Sens. Environ. 83, 303–319 (2002).

Hansen, M. C. et al. Global Percent Tree Cover at a Spatial Resolution of 500 Meters: First Results of the MODIS Vegetation Continuous Fields Algorithm. Earth Interact. 7, 1–15 (2003).

De Fries, R. S., Hansen, M., Townshend, J. R. G. & Sohlberg, R. Global land cover classifications at 8 km spatial resolution: The use of training data derived from Landsat imagery in decision tree classifiers. Int. J. Remote Sens. 19, 3141–3168 (1998).

Gao, Y. et al. Assessing forest cover change in Mexico from annual MODIS VCF data (2000–2010). Int. J. Remote Sens. 1–18 (2018). doi:10.1080/01431161.2018.1479789

Gerard, F. et al. MODIS VCF should not be used to detect discontinuities in tree cover due to binning bias. A comment on Hanan et al. (2014) and Staver and Hansen (2015): GERARD et al. Glob. Ecol. Biogeogr. 26, 854–859 (2017).

White, M. A., Shaw, J. D. & Ramsey, R. D. Accuracy assessment of the vegetation continuous field tree cover product using 3954 ground plots in the south‐western USA. Int. J. Remote Sens.26, 2699–2704 (2005).