Somalia, with almost 2/3 of its land devoted to agriculture and livestock rearing, is facing the negative impacts of uncontrolled deforestation activities. A key driver of such a trend is the extensive and often illicit charcoal production, which leads to forest degradation dynamics and the depletion of the country's woody resources. To monitor and quantify these tendencies, remote sensing offers many advantages in terms of temporal and spatial coverage. Our study aimed to develop a workflow capable of integrating optical (Sentinel-2) and radar (Sentinel-1) imagery to detect charcoal production sites (i.e., kilns) in Somalia. Most of the processing was implemented in Google Earth Engine, enabling it to be fully replicable and easily scalable to other regions. Southern Somalia (Jubbaland State, approx. 110200 km2) was chosen as the test area since charcoal exploitation represents a critical issue in the region. Furthermore, a very detailed dataset, produced by the Food and Agriculture Organization (FAO-SWALIM) through photo-interpretation of kilns’ presence, was available for the area. Our methodology started by producing a single image containing both optical (NDVI) and radar (VV and VH polarizations) information over the first three months of 2016 and 2017. Subsequently, we calculated the difference between the two images and extracted the pixel values in correspondence with the known charcoal sites. Based on the extracted values, different thresholds (e.g., the mean +/- a set number of standard deviations) were tested for classifying the difference image. The results consisted of binary maps at 10 m resolution, showing locations with kilns’ presence or absence. A confusion matrix was used to evaluate the classifications. Overall accuracy reached almost 70% in some cases, while sensitivity and specificity were more variable (0.4 to 0.9), depending on the utilized threshold. Particularly, some of the classifications proved to be very balanced, with values around 0.7 for all three parameters of accuracy, sensitivity, and specificity. Our results demonstrate that a multi-sensor remote sensing approach is a valuable and reliable tool to monitor and quantify forest degradation dynamics, particularly considering the socio-political context of some countries, where in situ data collection is often difficult, when not dangerous.