A Remote Sensing Data Fusion algorithm for Near Real time monitoring of Volcanic Radiative Power

Nowadays, near-real time volcano monitoring at global scale is made possible thanks to the thermal infrared sensors on board of several satellite platforms providing accurate estimates of the volcanic thermal emissions. In particular, they are able to provide reliable estimates of the Volcanic Radiative Power (VRP), i.e. the heat radiated during the volcanic activity. In addition, Remote Sensing Data Fusion (RSDF) techniques allow to combine data from multiple satellite sensors to improve the potential values of the single source and to produce a high-quality data representation. Fusion techniques are useful for a variety of applications, ranging from object detection, to object tracking, change detection. In particular, we aim to use them to integrate the different satellite data acquired with different spatial and spectral resolutions to produce fused data that contains more detailed information than each of the data sources. We introduce a novel RSDF algorithm deployed in a Cloud Computing environment to monitor VRP worldwide from multiple multispectral satellite sensors, namely the polar MODIS, SLSTR and VIIRS and the geostationary SEVIRI. The RSDF algorithm demonstrates heightened sensitivity in detecting high-temperature volcanic features and thus VRP monitoring compared to conventional already processed Level 2 products available online. Specifically, the overall accuracy has improved in terms of omitted rate and false detections, reducing from 78% to 5.3% and from 6.5% to 4.7%, respectively. The decision to combine the use of different satellite sensors stems from the need to offer complete continuous monitoring of each volcanological phenomenon, taking advantage of each sensor's own characteristics.