Forest Change using Sentinel-2 dataSubmitted by RenneTergujeff on Tue, 2018-04-24 23:00
Forest Change using Sentinel-2 data
Updated by Renne Tergujeff, 27 August 2018 (Forestry TEP v2.4.1)
When forest is converted to agricultural land or logged for other purposes, the measured reflectance in various parts of the visible and infrared spectrum changes a lot. This is evident when investigating the various spectral bands of the MSI instrument aboard Sentinel-2 satellites. Especially if bare ground is exposed in connection with the logging operation, red reflectance (and the infrared reflectance in the 1.6 µm band) rises. The ratio of red reflectance after vs. before can be used as a measure of detecting changes in forest cover. The Forestry TEP Sentinel-2 Forest Change service computes a red-reflectance ratio layer, which can then be further analysed e.g. in the QGIS service. A forest or land cover map produced using the Sentinel-2 Land Cover service is most useful in this analysis.
The service requires as its input data two Sentinel-2 images in the new format, i.e. one tile per image. The new format, valid from the end of 2016, can be recognized by file names that begin with “S2A_MSIL1C” or “S2B_MSIL1C”, not with “S2A_OPER”. The images must cover the same area (the same Sentinel-2 tile). The service is most useful with completely cloud-free images. Cloud masking is performed with the included cloud masks of Sentinel-2 level 1C products. How to search and select satellite images in the Explorer view is described here.
Running the service:
The service requires defining parameters in the Service dialogue panel (workspace):
- Start product: The first input Sentinel-2 image, which has been acquired before the assumed forest change. Drag the input Sentinel-2 image from a search results listing or from databasket contents at the bottom of the screen, by grabbing from the horizontal lines in front of its name.
- End product: Select the second image in the same way and drag it here.
- Target CRS identifier: This field is OPTIONAL. It defines the target coordinate reference system (CRS). If left empty, the CRS of the input data is used as default. This must be a string beginning with 'EPSG:', and the following numeric code must be a valid system as defined by the European Petroleum Survey Group (EPSG). For UTM zones in the northern hemisphere and with the WGS84 datum, the numeric code is 32600 + the UTM zone number. An example of a valid CRS specification is EPSG:32635, where "6" or "7" defines Northern/Southern hemisphere, and 35 represents the UTM zone.
- Area of interest: This field is OPTIONAL. It defines the area of interest (AOI) that the output image is cropped to. Note that full Sentinel tiles are processed regardless of this setting. If left empty, an image covering the entire tile area is produced. The AOI can be copied from the area drawn on the map with the Copy from Map button. Alternatively, it can be specified in the Well-known text (WKT) POLYGON format. An example of a valid specification is: POLYGON((-92.906633 16.190411,-92.066559 16.188383,-92.070266 15.376645,-92.907004 15.378567,-92.906633 16.190411)).
- Target image resolution: This field is OPTIONAL. It specifies the desired image resolution in metres, controlling resampling of the output image (pixel spacing). If left empty, resolution of the input data (10) is used as default. For Sentinel-2 optical data, feasible values are between 10 and 100 (meters).
- Label: OPTIONAL. Allows free-form tagging for later identification of this processing job.
After all input fields have been filled in, the processing service is launched by clicking the round play button at the bottom right corner of the input dialogue area. The processing can typically take 25 minutes. The resulting output is GeoTIFF file that can be downloaded, re-used or displayed in one of the GUI applications, such as Sentinel Toolbox or QGIS. Tutorials on opening data within the Sentinel Toolbox and QGIS are provided. The image below shows a generated change map (red reflectance ratio) within SNAP. Most clouds are masked to white (with "no data" value of 32767) while cloud shadows show as artefacts.