Special Issue Foreword: Mapping & Modeling Land Use/Land Cover Dynamics in Frontier Settings
by Stephen J. Walsh, Joseph P. Messina, and Daniel G. Brown
We have showcased the work in this Special Issue from projects that have been supported by an array of research initiatives including, for instance, the NASA Land Cover/Land Use Change Program, the NASA Ecology-LBA (Large-Scale Biosphere-Atmosphere Experiment in Amazonia) Program, the NASA New Investigator Program, and the NSF Coupled Natural-Human Systems Program. Their intent is to describe the nature of land use/land cover (LULC) dynamics in frontier settings and investigate the causes and consequences of LULC by mapping and modeling landscape patterns and dynamics and evaluating these in the context of human-environment interactions. The growing community of scholars focused on sustainability science and global change is increasingly recognizing that LULC change and system dynamics, as driven by human behavior and agency and the interactions between population and the environment, are among the most critical set of determinants of both environmental and societal aspects of sustainability. As a result, various branches of science and policy are relying on maps and models of LULC change at multiple and interacting, spatial and temporal scales.
A de facto Land Change Science that emphasizes the causes and consequences of LULC dynamics has developed through the efforts of a diverse community, a number of interacting organizations, and a growing number of research programs, and challenges the scientific community, engages policy-makers, and involves decisions- makers and landscape agents who interact with each other and the land. Concerned chiefly with the spatial and temporal patterns of LULC dynamics, Land Change Science examines the linkages between endogenous and exogenous factors and pattern-process relations, thereby necessitating the development of new tools to enhance our ability to map and model LULC patterns and to extend our understanding of how people and the environment interact.
The collection of papers in this special issue emphasizes the mapping and modeling of LULC dynamics in frontier settings where the imprint of landscape change has been relatively recent, temporally and spatially profound, and/or where global reach of market economies has not been fully extended. Collectively, the papers emphasize the integration of remote sensing data types, analytical approaches, spectral and non-spectral data, and image analyses that are used to characterize LULC change and pattern-process relationships across space and time scales. The papers add to the remote sensing literature, but also to the GIScience and Land Change communities through their explicit links to geospatial data and methods for analyzing spatial patterns of human and natural systems interactions. The general goals of the special issue are to focus on remote sensing by featuring the use of optical and non-optical systems, multi-spectral and hyper-spectral data, field and laboratory analyses of spectral and spatial data, innovative processing and analysis approaches, and local to regional scales of study for historical to contemporary periods. In addition, the papers emphasize models for assessing the causes and consequences of LULC dynamics and linking mapped trajectories of change to the complex interplay of people, place, and environment in diverse and complex settings.
Arima and colleagues examine the spatial fragmentation of forest in the Brazilian Amazon that is associated with the building of roads by loggers. GIS, geo-statistical methods, and microeconomic theory are used to characterize the spatial organization of forest cover in an important tropical forest setting. Behavioral processes of key agents are linked to the simulated spatial patterns of forest fragmentation through models that explicitly associate human characteristics to landscape outcomes. With implications for GIScience and landscape ecology, methods and findings are linked to the human dimension through pattern-process relations.
Santos and Messina classify African Palm in the Ecuadorian Amazon using a combination of synthetic aperture radar (SAR) data, ground-based digital video data, and Landsat ETM+ imagery to map and model African Palm plantations through a data fusion approach. Classification results, generated through (a) the fusion of optical and non-optical remotely-sensed data and texture measures, and (b) the fusion of single Landsat ETM+ bands with texture measures, are described. Classifying LULC types in such an ecologically and spectrally complex environment has been a persistent remote sensing challenge.
Also working in the Ecuadorian Amazon, Walsh and colleagues assess reforestation patterns on household farms, primarily using hyper-spectral Hyperion data. Reforestation is described as an important element of LULC change in the Amazon Basin and as a critical landscape factor that influences ecosystem goods and services. Linear mixture modeling is described as an approach to derive sub-pixel fractions of LULC types through the processing of Hyperion data, and spectral endmembers defined using the hyper-spatial Ikonos data. A longitudinal household survey and a cross-sectional community survey of socio-economic and land use variables are used to relate the mapped patterns of reforestation patterns to socio-economic and demographic drivers.
The trajectories of LULC change in the Northern Ecuadorian Amazon are examined by Mena by assessing the composition change of forest landscapes through time, the spatial configuration of the trajectories of LULC change, and the probabilities of trajectories occurring in this dynamic forest frontier. A Landsat image time-series is used to characterize land use change and to construct pixel-based trajectories. Cluster analysis and landscape pattern metrics are used to derive measures of spatial structure, and spatially-explicit models are described that assess the stability/ dynamism of the trajectories. Findings are interpreted relative to social and ecological theory, ecosystem health, and the sustainability of protected areas.
The Lower Okavango Delta of Botswana is examined by Neuenschwander and Crews by assessing important ecosystem signals derived by a Landsat image time-series, using a harmonic regression approach and wavelets. The authors relate the decomposed signals to observed flooding and fire disturbances, as well as to the impact of agriculture, human settlement, and land management schemes. Harmonic regression on the remotely-sensed vegetation indices are used to assess the variances in the trajectories of change that are represented though the image time-series and associated with climatic and anthropogenic processes as drivers of land use dynamics.
Welsh characterizes land use change in northeastern Thailand. He uses a Landsat image time-series for a 25-year period beginning in the early 1970s, a GIS to derive the agricultural potential of land use types, and soil data to assess the suitability of land use types, land degradation potential, and the relative marginality of sites and conditions. The satellite image time-series is used to generate LULC change trajectories to assess agro-ecological sustainability of a predominately agricultural environment that has undergone considerable changes through deforestation and agricultural extensification.
Jiang and colleagues examine the vulnerability to flooding near Poyang Lake in the central Yangtze River Basin of China by mapping the patterns of LULC change using a Landsat TM image time-series interpreted in relation to elevation and levee characteristics. The spectral analysis involves the generation of an expanded feature set for LULC classifications of the four periods of study. Principal components analysis, the tasseled cap greenness transform, and a summertime ratio of the red/green spectral channels are used in the classification process. Flood vulnerability is assessed and interpreted relative to land use dynamics, policy decisions, and land management programs.
Finally, forest dynamics in central Siberia is the subject of a study by Bergen and colleagues who use an assembled Landsat image time-series, a DEM, and ancillary GIS data layers to compare deforestation and change patterns to Soviet and post-Soviet eras. The amount, type, and successional state of the forested landscape are coincident with the 1991 breakup of the Soviet Union.
In sum, the collection of papers in the special issue emphasizes the mapping and modeling of LULC change and evaluating their ecological and social consequences. Study sites in South America, Africa, and Asia are represented, as are optical and non-optical remotely-sensed data sets, fine-and moderate-grain satellite data, multi-spectral and hyper-spectral data, innovative field verification and measurement initiatives, and a diverse set of derived and/or collected statistical and spatial variables that are integrated in analysis approaches. A commonality found in the papers is the centrality of remote sensing in the studies and the multi-temporal nature of their assessments. As the time series of available images lengthens, and as higher temporal resolution datasets become available, the spatial-temporal analytical and modeling approaches represented in these papers will become increasingly more important for turning these data into
Special Issue Editors
Stephen J. Walsh
Department of Geography, University of North Carolina – Chapel Hill
Chapel Hill, North Carolina 27599-3220
919-962-3867 (voice), 919-962-1537 (fax)
swalsh@email.unc.edu
Joseph P. Messina
Department of Geography, Michigan State University
East Lansing, Michigan 48824-1117
517-432-4752 (voice), 517-432-1671 (fax)
jpm@msu.edu
Daniel G. Brown
School of Natural Resources & Environment, University of Michigan
Ann Arbor, Michigan 48109-1041
734-763-5803 (voice), 734-936-2195 (fax)
danbrown@umich.edu