Course:Cons452/Hansen

From UBC Wiki

Hansen Global Forest Change dataset

Description

The Hansen Global Forest Change dataset is a time-series analysis of Landsat images which shows global "forest extent" and change from the year 2000 to the year 2017. It is very important that you understand what is meant by "forest extent" in this dataset. What the authors did was map "global tree cover extent, loss, and gain... at a spatial resolution of 30 m, with loss allocated annually"[1]. In this dataset trees are defined as vegetation taller than 5m in height. With a spatial resolution of 30 m, it is a pretty amazing dataset in that it can provide a detailed picture of deforestation and reforestation[1]. It is the most comprehensive dataset of forest cover (recall: tree cover > 5 m) and forest change available publicly; however, there are some disadvantages. For example, it does not distinguish tropical forests from plantations and herbaceous crops (such as bamboo and banana), which thus can underestimate forest loss in tropical areas[2]. The newest version of the Hansen dataset is version 1.5, but version 1 is still available for download (version 1 only covers years 2000-2012). The new version has notable updates which have helped improve the detection of some forest change nuances, like better detection of selective logging. They are planning for a Version 2 in the future (no release date).

The Hansen dataset contains 6 layers, each its own .tif file:

  1. Tree canopy cover % for the year 2000 (treecover2000)
    • Tree cover in the year 2000, encoded as a number for each output grid cell. The number represents the % of canopy closure for vegetation above 5m tall in that cell. See discussion below in "Common Problems" on how to turn this continuous raster data into a binary map of forest (forest/non-forest).
  2. Global forest cover gain from 2000-2012 (gain)
    • Defined as a change from non-forest to forest state within the period 2000-2012. Encoded as 1 = gain, or 0 = no gain.
    • "Forest loss was defined as a stand-replacement disturbance or the complete removal of tree cover canopy at the Landsat pixel scale. Forest gain was defined as the inverse of loss, or the establishment of tree canopy from a nonforest state"[1].
    • A "forest state" is considered a tree canopy cover density of > 50%.
    • "Gain was defined as the inverse of loss, or a non-forest to forest change; longer-lived regrowing stands of tree cover that did not begin as non-forest within the study period were not mapped as forest gain"[3].
  3. Year of gross forest cover loss event (lossyear)
    • Defined as a change from forest to non-forest state at a point between 2000-2017.
    • "Forest loss was defined as a stand-replacement disturbance or the complete removal of tree cover canopy at the Landsat pixel scale"[1]. In other words, it's a pixel where the tree canopy cover went from > 50 % to ~ 0 % at some point in the time period.
    • Encoded as 0 = no loss, or a value between 1 – 17. The number corresponds to the primary year in which the forest loss occurred.
  4. Data mask layer (datamask)
    • 0 = no data, 1 = land surface, 2 = permanent water bodies
  5. Cloud-free image composite (c. 2000, Landsat 7) (first)
    • Reference multispectral imagery for the first available year, usually 2000
    • Composite imagery is composed of median observations in 4 spectral bands: Landsat bands 3 (red), 4 (NIR), 5 (SWIR), 7 (SWIR) are available (observations are taken during growing season).
  6. Cloud-free image composite (c. 2017, Landsat 7) (last)
    • Reference multispectral imagery for the last available year
    • Composite imagery is composed of median observations in 4 spectral bands: Landsat bands 3 (red), 4 (NIR), 5 (SWIR), 7 (SWIR) are available (observations are taken during the growing season).

Metadata

Metadata Component Description
Theme Forest (meaning tree cover) change   
Source Matthew Hansen et al. (University of Maryland, 2013) created the dataset, and they derived it from Landsat satellite imagery
Purpose To visualize and analyze forest cover change in detail across the world
Time Frame 2000 to 2017
File Type Spatial
File Format .tif
Structure Raster
Projection and coordinate system EPSG:4326 - WGS 84 - Geographic   
Extent Global (data are sub-divided into 10-degree “granules” so that you do not need to download the entire global extent)
Resolution or scale Spatial resolution is 30 metres (1 arc second per pixel)    

Common Problems

  • If your project examines forest cover in a tropical area, be careful not to assume that all "forest" (or tree cover) is the same. The Hansen dataset does not differentiate tropical forests from plantations and herbaceous crops[2]. Keep this in mind when carrying out your analysis.
  • For layer #1: tree canopy cover % in the year 2000, be aware that this is continuous data. In other words, it shows the % tree cover in each pixel. If you would like to turn this layer into a forest cover layer, you will need to decide what % tree cover "counts" as forest for your specific question. Commonly used thresholds are 10 % (used by the FAO in their 2015 Forest Resources Assessment[4]), 30%[5], but the threshold can also be higher depending on what it is you are trying to examine (and where it is geographically). Using a higher threshold may underestimate forest cover in dryland or savanna regions; whereas using a lower threshold may increase the apparent forest cover. Experiment to assess the effects of using different thresholds in your analysis.

Downloading Instructions

  1. Go to the url: https://earthenginepartners.appspot.com/science-2013-global-forest/download_v1.2.html .
  2. Scroll down to the subheading “Download Instructions”.
  3. Click on your desired 10 degree “granules” on the page. These cubes exist so that you don’t have to download an unwieldy amount of data, but rather just what you need for your specific project. You’ll notice that once you click on your granule, the links below the map will change to match up with the latitude of it.
  4. Each link corresponds to a layer file (.tif) for that granule (one of the 6 layer files mentioned above), and you can download select layers or all of them for your desired tile.

Restriction on Use

Creative Commons Attribution 4.0 International License

Please see https://creativecommons.org/licenses/by/4.0/legalcode for full license terms but this essentially means that you are free to use, share, and adapt this data as long as you attribute it and indicate how you have manipulated or built upon the data (if you do so).

  1. 1.0 1.1 1.2 1.3 Hansen, M. C.; Potapov, P. V.; Moore, R.; Hancher, M.; Turubanova, S. A.; Tyukavina, A.; Thau, D.; Stehman, S. V... (2013). "High-Resolution Global Maps of 21st Century Forest Cover Change". Science. 342: 850–853.
  2. 2.0 2.1 Tropek, R.; Sedlacek, O.; Beck, J.; Keil, P.; Musilova, Z.; Simova, I.; Storch, D. (2014). "Comment on "High resolution global maps of 21st century global forest cover change"". Science. 344: 981.
  3. Hansen, M. C.; Potapov, P. V.; Moore, R.; Hancher, M.; Turubanova, S. A.; Tyukavina, A.; Thau, D.; Stehman, V.; Goetz, S. J... (2013). "Supplementary Materials for High-Resolution Global Maps of 21st-Century Forest Cover Change". Science. 342: 850.
  4. Food and Agriculture Organization of the United Nations. (2012). "Forest Resources Assessment Working Paper 180: FRA 2015 Terms and Definitions". 1-28.
  5. Laura Vang Rasmussen, et al., Global Food Security, https://doi.org/10.1016/j.gfs.2019.100331