Monitoring the Pulse of our Planet: Towards the 100 Trillion Pixel Challenge

Over the last half century, satellite remote sensing has played an important role in describing and assessing landscape dynamics of Planet Earth. With increasing spatial and temporal resolutions of remotely sensed data, mapping physiographic features along with the built environments has progressed from broad classes of land use land cover (LULC) to individual objects such as buildings, roads, cars etc. Moreover, our ability to detect, identify, and characterize changes in these LULC classes or objects on, under, and above the Earth's surface has vastly improved. For example, recent advancements in GPU based high performance computing architectures combined with machine learning and deep learning, exploiting very high resolution (sub-meter resolution) imagery has proven to be convenient for creating accurate human settlement maps at national scales. This has also enabled potential (social and vulnerability) characterization of population from variability in settlement structures. Emergence and expansion of small satellites across the globe are bringing a paradigm shift where daily cadence of ~5m resolution data will lead to rapidly growing data cubes. Consequently, the community must prepare for "the 100 trillion pixel challenge"; develop a feasible technology pathway through which such pixel volume could be analyzed daily to measure and track the pulse of our planet.