Fundamental Concepts of Intensity Analysis to Understand Changes Among Categories

Comparison of maps of land categories from different time points for the same site is a common approach to reveal land transitions over time. Typically scientists attempt to explain these transitions in a manner that links the patterns of change with the processes of change. This requires a proper characterization for the observed transitions. However if scientists fail to identify signals of systematic and/or stationary transitions among categories, they might end up explaining processes that are spurious and/or ephemeral. This research develops methods to quantify the behavior of land categories across several time intervals to answer the following interrelated questions for any given case. (I) In which time intervals is the annual rate of overall change relatively slow versus fast? (2) Given the answer to question I, which land categories are relatively dormant versus active in a given time interval, and is this pattern stable across time intervals? (3) Given the answers to questions 1 and 2, which transitions are intensively avoided versus targeted by a given land category in a given time interval, and is this pattern stable across time intervals? (4) For each answer of the previous three questions, how much error in the data can explain the non-uniform results? (5) How can we aggregate the categories to simplify the analysis while still focusing on the important transitions? This dissertation illustrates the concepts with applications to land change in Massachusetts and Florida USA, while the concepts apply to changes in any type of categorical variable.;The broader impacts of this research lie in its potential to combine different levels of analysis into one unified framework, where the more detailed levels are conditional on the less detailed levels. This integration improves the ability of scientists to characterize patterns of change appropriately at different levels of details and over several time intervals in order to solicit complimentary information from other experts about the processes and causes of change. Furthermore, this research provides supporting methods to expose the size of data error that would affect the results and to understand the influence of aggregating categories on the measurement of transitions. All concepts in this research have been incorporated into two computer programs, which are available for free at http://sites.google.com/site/intensityanalysis.