The Earth is undergoing massive changes that are difficult to measure and even harder to forecast. Modeling these changes is required for predicting, planning for, and mitigating the effects of natural and man-made disasters. Processes that affect and feed off of these changes are governed by powerful and complex dynamics that occur at different spatio‐temporal scales. Examples of short‐time scale events include floods, hurricanes, earthquakes, tsunamis, wild fires, and volcanic eruptions. Long-time scale processes include drought, spread of invasive species, population growth, changes in land use, soil degradation, sea level rise, ocean acidification, permafrost melting, glacier retreat, sea ice loss, degradation of fresh water resources, land subsidence, climate change, deforestation, desertification, and critical habitat loss. Obtaining a better understanding of these processes involves the instrumental task of first harnessing the data associated with all relevant sensor modalities and then identifying the proper mathematical models and the computational machinery to process and extract relevant information from the data.
Terabytes of geospatial data are collected daily from a variety of sources. The amount of data is massive since these data are often high dimensional. Processing them and extracting useful information from them are major challenges that need to be overcome. Novel mathematical imaging techniques have already begun to address some of these problems. Mathematical and engineering advancements have led to methods for sub‐Nyquist sampling rates. These methods are significantly impacting the manner in which sensor acquisition systems collect, model, and analyze the data. To render the exploitation more tractable, researchers have developed innovative techniques for projecting the data into lower dimensional spaces. One such example has been the groundbreaking numerical developments for representing data with appropriately constructed dictionaries, sometimes learned from the data itself. These dictionaries are advantageous for many reasons, including computational efficiency, ease in which they are implemented, accuracy, and reliability. These and other new paradigms will significantly impact the manner in which we collect, store, transmit, and process data that has been derived from multiple sensors.
In conjunction with the special year on the Mathematics of Planet Earth, this workshop will address novel mathematics, new sensor technology, and computational imaging techniques that can lead to innovative ways of exploiting geospatial image information. Since cross‐discipline communication between those who observe Earth and those who strive to model the Earth’s processes must be effective and robust, participants will include scientists who are directly involved in the applications as well as mathematicians, computer scientists, and engineers who are developing mathematical approaches for representing, processing, and analyzing image data. The workshop aims to foster communication among these disciplines in order to validate and better understand models of the Earth’s complex processes by effectively using geospatial image data.