Assessing Landscape Change and Population‐Environment Interactions via Panel Analysis

Abstract

Landscape change involving landuse/landcover (LULC) traditionally was considered the territory of remote sensing, a result of clustering of spectral responses and subsequent interpretation. However, a more integrative approach to landscape change has evolved into a spatially explicit and scale-dependent practice more appropriately couched in the context of Geographic Information Science (GISc). GISc promotes the importance of combining multiple data sources and techniques into a scientifically grounded information system with an integrative and derivative capacity for characterizing scale, pattern, and process variables across thematic and spatial/temporal domains. Combining GISc and landscape ecology provides a meaningful way to interrelate theory and practice in the analysis of scale, one of the most fundamental and inherently geographic factors in assessing landscape form, function, and change. This research models the relationship between 25-years of landscape change and variables derived in a Geographic Information System (GIS) representing population, environmental gradients, and accessibility. These relationships are statistically tested across spatial scales to assess human-environment interactions at work on the landscape. In addition, particular attention is given to the characterization of multitemporal LULCC (LULC change) over the landscape at the pixel level (900 m landscape cells) as well as the characterization of forces at work on the landscape via correlation analysis and tests of significance across spatial scales to discern population-environment interactions. Pattern metrics are used to examine the composition and spatial organization of the landscape across spatial scales by computing selected indicators of LULC structure from the panel change histories derived through an expert system approach, defining the emergent properties of LULC dynamics across space and time. Lastly, three measures of spatial autocorrelation are assessed relative to landscape trajectories. The basic intent of the research is to study humanenvironment relationships in Nang Rong district in northeast Thailand by examining the propensity of LULCC in support of agricultural extensification through deforestation as a consequence of 1) total population at the village level, 2) resource endowment reflected in soil moisture potential, and 3) access to critical infrastructure for agriculture. To this end, landscape classification trajectories are created to measure agricultural extensification, deforestation, and LULC stability.