An approach to spatial data gathering technology: Realizing large-scale fire modeling in the wildland urban interface

Environmental planning continues to evolve, modify and introduce models that test, simulate and attempt to answer questions that arise where humans and nature interact. Due to historical constraints in the capability of equipment and computing resources, many environmental models ignored heterogeneity in the environment and modeled the subjects at a highly aggregated level. Although this might help in policy making at a national, state and even regional level, it does not serve solutions at the large-scale or neighborhood level where most landscape design problems and practical solutions must be advanced. Recent advances in technology such as: computing power, geographic information science, remote sensing, global positioning systems, the Internet, and wireless communication, have made it possible to now reexamine these models and their implementation.; This dissertation recognizes the call for a shift in modeling scale, and proceeds to solve the basic problem encountered by many environmental models during such a scale change---collecting large-scale spatial data. Although some modelers are beginning to recognize the need for this shift, few have proposed and fewer have attempted to design systems for automating large-scale data collection, resulting in a delay for this movement. Using Wild land Urban Interface (WUI) modeling of wildfire as an example, this dissertation proposes two new approaches that employ and guide both humans and robots to advance to necessary large-scale modeling: (1) developing the tools for massively paralleled human-based Internet mapping, and (2) developing tools to guide massive robotic remote sensing.; The first approach, human-based Internet mapping, targets applications when automated methods such as remote sensing alone could not provide satisfactory results and human interpretations are intricate to success. Here a general Internet mapping system, introduced as iMap, is developed to support simple or good enough data input and two way data transfer over the Internet. Unlike other complex geographic information systems (GIS) or limited web based GIS (WebGIS), iMap is designed to be simple enough for the average citizen with no GIS experience to use, yet powerful enough so that general mapping and interpretations are all supported. In order to meet highly varying mapping requirements, a hierarchy of an object model is carefully designed and exposed to other applications through Microsoft COM and ActiveX technology. Consequently iMap is highly customizable and can be easily adopted for many types of planning needs. A preliminary test of the iMap mapping system is undertaken for an area of the University of California, Berkeley, Botanical Garden and the acceptable results are evaluated and reported using a conservative metric.; The second approach, guiding massive robotic remote sensors, is based on wireless sensor networks which will be necessary if we are to shift to real-time large-scale data gathering to fuel our models in dense heterogeneous environments. This dissertation provides solutions to the critical problem of maintaining dynamic wireless sensor network coverage and connectivity in harsh environments. An alpha-shape based algorithm is developed and dynamically applied to detect and monitor sensor coverage, and a beta-skeleton based algorithm is designed and deployed to analyze dynamic sensor network structure and evaluate potential vulnerable links to assist in redeployment decision making. alpha-shape and beta-skeleton workbenches are implemented with graphic user interfaces and scripting support to accomplish this second objective.; This dissertation simulates a WUI wildfire in the Claremont Canyon of Oakland, California employing: (1) fuels classified using coarse images, and (2) using iMap to delineate large-scale fuels. The simulation results vary significantly and divulges that large-scale data, where roads are detected as fire breaks, restricts the fire spread to an area 1/3 t...