Sparse-Data Forecasting of Megacity Growth

摘要

The Department of Defense stands ready to respond to security threats and human assistance and disaster relief (HA/DR) across the globe. Rapid response relies on forecasting probabilities of needs far in advance to ensure that soldiers and materiel are sufficient and adequately prepositioned, which is challenging in rapidly changing human and natural threats. One of those dimensions involves rapid global megacity growth, which can result in ungoverned urban areas attractive as safe havens and recruiting targets for terrorist groups. Forecasting when and where such areas are likely to develop would allow military planners to better anticipate future challenges. Unfortunately, to date computational capabilities to project future growth were developed for developed countries where historic and current land use/land cover data is readily available. As a result, there is a need for inexpensive methods to project urban growth to support the identification of dynamically changing social challenges, where megacities are the most challenging environments forecasting changing conditions. In this article, we present a novel solution using the regional urban growth (RUG) model, a spatially dynamic, extensible approach for assessing the relative attractiveness within cities for various densities of residential growth within a region. This model estimates the attractiveness of development for every location in a rasterized landscape based on proximity to development attractors, such as existing dense development, roads, highways, forest, and water. The level of attraction can vary with distance of the attractor and can even be negative. RUG can be rapidly installed, parameterized, calibrated, and run for any megacity across the globe. Here we illustrate RUG capabilities by applying it to Dhaka, Bangladesh and its surrounding areas. The principal raw data is digital elevation and 8-band, 2m resolution World-View-2 satellite imagery. Our findings suggest that the RUG model provides a cost-effective means of predicting where urban growth will occur in megacities based on development attractiveness factors derived from sparse, but ubiquitous, global data. This information will improve Department of Defense (DoD) long-term preparation and planning for security and emergency needs within megacities by providing critical neighborhood scale household density information for constructing sociocultural analysis maps.