Results from time-series analysis of Landsat images characterizing forest extent and change. Trees are defined as vegetation taller than 5m in height and are expressed as a percentage per output grid cell as ‘2000 Percent Tree Cover’. ‘Forest Cover Loss’ is defined as a stand-replacement disturbance, or a change from a forest to non-forest state, during the period 2000–2014. ‘Forest Cover Gain’ is defined as the inverse of loss, or a non-forest to forest change entirely within the period 2000–2012. ‘Forest Loss Year’ is a disaggregation of total ‘Forest Loss’ to annual time scales. Reference 2000 and 2014 imagery are median observations from a set of quality assessment-passed growing season observations. detailed information at: http://science.sciencemag.org/content/342/6160/850 This global dataset is divided into 10x10 degree tiles, consisting of seven files per tile. All files contain unsigned 8-bit values and have a spatial resolution of 1 arc-second per pixel, or approximately 30 meters per pixel at the equator.

Global 25m resolutions PALSAR-2/PALSAR mosaic and forest/non-forest map are free and open dataset generated by applying JAXA's powerful processing and sophisticated analysis method/techniques to a lot of images obtained with Japanese L-band Synthetic Aperture Radars (PALSAR and PALSAR-2) on Advanced Land Observing Satellite (ALOS) and Advanced Land Observing Satellite-2 (ALOS-2). The global 25m resolutions PALSAR/PALSAR-2 mosaic is a seamless global SAR image created by mosaicking the SAR images in backscattering coefficients measured by PALSAR/PALSAR-2, where all the strip data within 10x10 degrees in latitude and longitude are path-processed and mosaicked for the sake of processing efficiency. Correction of geometric distortion specific to SAR (ortho-rectification) and topographic effects on image intensity (slope correction) are applied to make forest classification easy. The size of one pixel is approximately 25 meter by 25 meter. The temporal interval of the mosaic is generally 1 year. The global forest/non-forest map (FNF) is generated by classifying the backscattering intensity values in the global 25m resolution PALSAR-2/PALSAR mosaic so that strong and low backscatter in HV-polarization are classified "forest" (colored in green) and "non-forest" (colored in yellow), respectively. Here, "forest" is defined as the natural forest with the area larger than 0.5ha and forest cover over 90%, as same to the FAO definition. Since the radar backscatter from the forest depends on the region (climate zone), the classification of Forest/Non-forest is conducted by using the region dependent threshold of backscatter. The classification accuracy is more than 84% according to the comparison with reference data such as in-situ photos and high-resolution optical satellite images.