Sensor: Landsat 7 ETM+
Data can be accessed here.
Algorithm description: The precise algorithm for the product has not been described in any one manuscript or ATBD. However, the algorithm is similar to those published in Hansen et al. 2002, 2008 and 2013. The algorithm uses training datasets of continuous forest cover and annual phenological metrics obtained from global datasets to build up regression tree algorithm that generates tree cover map. The ‘annual phenological metrics’ consists of information about phenological variations in terms of annual means, maximum, minima and amplitude of spectral data. The metrics are built up for different bands (red, NDVI and surface temperature). The regression tree is nonlinear and distribution free algorithm has been designed to handle complexity of global spectral land cover variations. The training dataset is built up based on visual image interpretation of images.
Accuracy: There has been limited verification of this product. Some research work has been done by applying slight variations of the algorithm to produce region specific forest map. Gross forest loss producer’s and user’s accuracy >88% and 89% respectively was found over Eastern Europe. The maps were able to capture changes in forest cover from 1985 to 2012 over this region. Song et al (2015) have evaluated the product over two counties in the State of Maryland, US. They report overall accuracy of 81% and producer’s and user’s accuracy ~85% for forested areas.
- Product is available only for 2010.
- Not much information is available for evaluation of this specific collection.
Hansen, R. . DeFries, J. R. . Townshend, R. Sohlberg, C. Dimiceli, and M. Carroll, “Towards an operational MODIS continuous field of percent tree cover algorithm: examples using AVHRR and MODIS data,” Remote Sens. Environ., vol. 83, no. 1–2, pp. 303–319, Nov. 2002.
M. C. Hansen, R. S. DeFries, J. R. G. Townshend, M. Carroll, C. Dimiceli, and R. A. Sohlberg, “Global Percent Tree Cover at a Spatial Resolution of 500 Meters: First Results of the MODIS Vegetation Continuous Fields Algorithm,” Earth Interact., vol. 7, no. 10, pp. 1–15, Oct. 2003.
M. C. Hansen, D. P. Roy, E. Lindquist, B. Adusei, C. O. Justice, and A. Altstatt, “A method for integrating MODIS and Landsat data for systematic monitoring of forest cover and change in the Congo Basin,” Remote Sens. Environ., vol. 112, no. 5, pp. 2495–2513, May 2008.