Sentinel-1 satellite data unlocks possibilities to map deforestation at a high spatiotemporal resolution
In the next 24 hours, deforestation will release as much CO2 into the atmosphere as 8 million people flying from London to New York; a report from the Oxford-based Global Canopy Programme, a league of 29 scientific institutions across 19 nations, explains. As of now, around half of the tropical rainforests have been lost to deforestation, and at the present rate, it is estimated that there would not be any rainforests left in 100 years’ time! Knowing when and where deforestation occurs can help reduce these catastrophic emissions by preventing the loss of the earth’s essential forest systems and enable a quick response to illegal deforestation.
For a long time, scientists and researchers have used MODIS and Landsat satellite data to track deforestation on large scales. These satellite systems provide consistent global data with coarse to moderate resolution. For instance, consider a single MODIS 250 m pixel, which corresponds to an area of 6.25 hectares. While MODIS sensors provide a rapid assessment of forest change on a daily timescale, high-resolution measurements are needed to estimate the total deforested area and map illegal logging at a small scale. Landsat-8 with a 30m resolution can compensate for the spatial resolution requirements to some extent. But, the low temporal resolution (30 days) and consistent cloud cover in the tropics make it inevitable to rely on image composites that may not provide an accurate estimate of when deforestation happened. The Copernicus program’s Sentinel-1 satellite has revolutionized deforestation monitoring as it provides regular updates on the condition of ever-changing forests at a range of temporal and spatial resolutions.
What are we offering?
To combat deforestation and forest degradation in tropical rainforests, Space4Good, in collaboration with Indonesia-based Arsari Enviro Industri, is developing a Re-Forest-ER platform for agroforestry. The development of palm oil plantations, illegal logging for export, increased demand for coal, and local shifting cultivations in this area has been major drivers of tropical deforestation and therefore Space4Good aims to deliver a reliable and near-real-time monitoring platform that can detect deforestation events at a high spatial resolution (Figure 1b). In the context of mixed tropical agroforestry ecosystems, the approach we are developing is both novel and innovative.
Using the time-series data from the ESA Sentinel-1 radar satellite in combination with the local expertise (knowledge and feedback loop from local experts) and Artificial Intelligence, we are able to accurately map the deforested areas and alert the local authorities. But how does it work? and what is new about the Re-Forest-ER platform? The improved timeliness, scale, and resolution of the information derived from Sentinel-1 data have made it possible for the local stakeholders to monitor thousands of hectares remotely. There’s even more to it. The conventional cloud and haze cover problems associated with the tropical areas are completely eliminated with the inclusion of Sentinel-1 radar data that can see through the clouds and surpass these challenges and record and characterize the magnitude of change (Figure 2b). Through the approach, it is possible to assess both the extent as well as the time of the deforestation event.
Various challenges emerge throughout developments. In this case, the slope correction in the algorithm is of utmost relevance since the topography in this area is extremely diverse. In addition, the changes due to seasonality (wet/dry) also need to be accounted for in the algorithm for accurate results. With the methodology laid out, the next steps are to further develop and validate the algorithm, interpret the seasonal changes in the time series data, apply slope correction, and better automate deforestation alerts for the future Sentinel-1 images. The results will then enable alerts to be sent to the end-user (via email or WhatsApp) automatically when detected. The transferability of the algorithm to other areas needs further investigation.
Would you like more information or are you interested in collaborating with Space4Good? Visit our website or contact us via email@example.com.
Originally published at https://www.space4good.com on January 27, 2021.