The Federal Ministry of Economics and Climate Protection (BMWK) is funding a new research project entitled Dense Satellite Time Series for Forest Monitoring (DESTSAM). In cooperation with Prof. Michael Schmitt from the Universität der Bundeswehr München, innovative AI-based methods are to be developed to supplement optical satellite image time series with radar satellite images. Clouds in optical satellite imagery often prevent regular monitoring, leading to gaps in the time series. Radar satellites do not have this limitation.
By means of the AI-based amendment of time series of optical satellite data (Sentinel-2) with radar satellite data (Sentinel-1), a temporally dense and application-oriented monitoring of forest areas should be possible. The increased information gained through dense time series is necessary in order to be able to identify information about disturbances in forest ecosystems worldwide in a timely manner. Methods that are only based on optical satellite data are not very suitable, especially in tropical areas, since weeks, usually even months, often pass between the deforestation event and the next cloud-free satellite image, during which part of the vegetation grows back. The damages are then no longer visible. The rapid regeneration of the vegetation prevents the timely detection of small-scale disturbances such as illegal logging. During forest fires, smoke and haze prevent timely detection of burned areas with optical satellites. Using time series supplemented with radar images, we expect to quantitatively record various processes of forest degradation and destruction such as logging, deforestation for plantation management, slash-and-burn and damage caused by storms and fire more quickly and accurately.