• Posted on: 3 March 2022
  • By: bc_editor


While forest cover is increasing in many industrialized countries, deforestation is still widespread in the tropics and subtropics. According to new data from the University of Maryland, 12.2 million hectares of tree cover were lost in the tropics between 2019 and 2020 - the second year in a row that primary forest loss worsened in these areas. Of this, 4.2 million hectares are in wet tropical primary forests, which are particularly important for carbon storage and biodiversity. The carbon emissions resulting from this loss of primary forests (2.64 Gt CO2) are equivalent to the annual emissions of 570 million cars. Deforestation is largely driven by forestry and agricultural expansion (Figure 1).

Figure 1: Emission sources for deforestation-related carbon dioxide emissions are diverse and vary by region. Emissions embodied in production are quantified for each commodity group in each country. The width of a region on the x-axis corresponds to the embodied emissions produced in that region, while the y-axis shows the proportion of emissions attributed to each commodity group within each region, meaning that the rectangles within the graph are scaled according to the embodied emissions in each region-commodity combination. The percentages within the rectangles indicate the share of total (2.6 GtCO2 yr-1) embodied emissions. For forest products, the results show emissions associated with the expansion of tree plantations, but not emissions due to timber extraction alone without subsequent land use expansion.

Between 2001 and 2014, Cambodia was one of the countries with the fastest forest loss globally. Densely forested landscapes have been cleared in large parts of the country over the past decade. We can place forest loss due to large-scale agro-industrial plantations (rubber) in a series with other agrarian conversion and illegal logging factors. The global rubber prices and a surge of land-concession deals have played key roles in accelerating Cambodia's rate of deforestation. The Cambodian government leases concession lands to domestic and foreign investors for agriculture, timber production, and other uses. Although other countries, such as Brazil, have lost more forest cover in recent years, Cambodia stands out for the rapid rate of deforestation of its forests (Figures 2,3,4). As per government data (satellite data from INPE) released beginning of February, Brazil recorded the largest ever deforestation of the Amazon rainforest this January. At 430 sqkm last month, this makes deforestation in the Brazilian Amazon five times higher than in January 2021.

To reduce emissions from deforestation and forest degradation, and for the sustainable management of forests and the conservation and enhancement of carbon stocks in developing countries, the UNFCCC Conference of the Parties (COP) for forest sector activities established a framework (REDD+). REDD+ targets the implementation of activities by national governments to reduce human pressures on forests that result in greenhouse gas emissions at the national level, but also recognizes subnational implementation as an interim measure.

The Sentinel-2 Global Mosaic (S2GM) service contributes to REDD+ monitoring by providing analysis-ready composites from time series of Sentinel-2 surface reflectance observations to monitor large forest areas (e.g., at country scale). Following 2019 and 2020, the S2GM service provided pre-processed products for 2021 to the REDD+ community for REDD+ regions in the tropics according to IFORCE specifications. A detailed description of the regions and covered periods can be found here. For the processing of the 2021 data, a total of 1427 Sentinel-2 MGRS tiles (10’000 km² each) have been processed, each tile containing multiple observations (roughly every 3 to 5 days) in several spectral channels (six spectral bands between the blue and short-wave infrared region). The size of the output products equals 1.1 TB for the year 2021. The datasets contain the spectral bands as well as metadata including quality indicators and details about the product lineage.

The figures below show a quarterly (January to March) mosaic of Cambodia (Figure 2). This example nicely demonstrates how the mosaic provided through the Sentinel-2 Global Mosaic service can help track changes in a timely manner. In the detailed view (Figure 2, right), human activities close to or in the forested areas are clearly visible. Comparing the S2GM mosaic product with Google Earth imagery (Figure 3) shows the timeliness of the provided information. For instance, in the central part, new clear cuts are visible in the Sentinel-2 based products (Figure 3b), which are not fully captured in the Google Earth imagery (Figure 3a). Thus, the S2GM products allow faster detection of human-induced land cover changes and clearly demonstrate the mosaicking service's value. Figures 4 show a mosaic time series between 2019 and 2021 (from left to right). The red rectangular marks the area of land cover change. In 2019, the beginning of deforestation can already be seen, 2020 clearly shows a significant progress, which has multiplied again in 2021.

Figure 2: Quarterly (January to March 2021) mosaicking product for Cambodia.

Figure 3: Comparison of (a) most recent Google Earth imagery, presumably a composite of the period 2018-2020, and (b) the Sentinel-2 Global Mosaic (S2GM) service product - January to March 2021 for the region indicated with the red rectangular in Figure 2. The S2GM products clearly provide more recent information, which can be seen, for instance, through the newly evolving clear cut in the central part in (b).  

Figure 4: Mosaic time series between 2019 and 2021 (from left to right). The red rectangular marks the area of land cover change. In 2019 (a), the beginning of deforestation can already be seen, 2020 (b) clearly shows a significant progress, which has multiplied again in 2021 (c).