Landscape Structure Changes Assessment to Monitor Green Spaces in Mueang District, Amnatchaoen Province

Abstract

Background and Objectives: Urbanization refers to the change of both physical and human landscape structures within an area in response to socio-economic development. This transformation leads to a reduction in urban open spaces, alongside the expansion of diverse land uses into peri-urban areas, contributing to the decline of green spaces in both urban and rural environments. This research aims to assess the changes in landscape ecological structures during the period from 2011 to 2022 by monitoring the diversity of land cover types using the Landscape Mosaic (LM) model and the LM-Anthropic model to describe the main structures and continuity of landscape components, including developed areas, agricultural areas, natural areas, and mixed-use areas, in order to evaluate the condition of green spaces in Mueang District, Amnat Charoen Province.

Methodology: This study applies geo-information technology to classify green and non-green areas based on Sentinel-2A satellite imagery, in conjunction with various indices, including  the Normalized Difference Vegetation Index (NDVI), Bare Soil Index (NDBSI), Normalized Difference Water Index (NDWI), and Normalized Difference Built-up Index (NDBI), using a hybrid classification method. The accuracy of the classification results was validated against ground truth points using real-time data collection via a Global Navigation Satellite System (GNSS). A confusion matrix was used to calculate overall classification accuracy and the Kappa coefficient, with the confidence level set at 80% and a  minimum acceptance threshold of substantial agreement.The resulting data were further analyzed  to examine landscape structure,  patterns, and changes in order to assess spatial distribution, configuration, and component changes in relation to the intensity levels of human activities, following the principles of landscape ecology.

Results: The land cover classification results for Mueang Amnat Charoen District in 2022 revealed an overall accuracy of 80.21% and a Kappa coefficient of 0.73, indicating substantial agreement. Agricultural land was the most dominant category, accounting for 60.46% of the area, followed by forest, barren land, perennial crops, community and built-up areas, and water bodies, at 8.59%, 6.85%, 4.72%, 1.85%, and 0.66%, respectively. These results characterize Mueang District’s core landscape structure as an agricultural matrix. Between 2011 and 2022, significant landscape changes were observed. The proportions of agricultural and natural landscape mosaics declined from  72.93% and 16.56% to 66.72% and 12.63%, respectively. In contrast, developed, mixed-use, and water landscape mosaics increased from 5.43%, 3.23%, and 1.84% to 8.93%, 9.83%, and 1.89%, respectively. Net changes in mosaic types revealed a transformation from uniqueness toward areas of dominance and presence.  Specifically, dominant agricultural, natural, and developed mosaic types declined by 30.13%, 5.76%, and 1.48%, respectively, anrd were replaced by mixed-use mosaics influenced by the convergence of all three components.This pattern corresponds with the intensity levels of human activities. Areas of extreme activity intensity were concentrated in dense urban cores, covering 4.17% of the district.  Moving outward from the urban center, the spatial pattern took on linear and dispersed forms, with decreasing levels of intensity and an increase in agricultural landscapes. Areas with very high and high levels of activity intensity accounted for 6.10% and 61.15%, respectively. Sparsely developed agricultural zones were categorized as moderate-intensity areas, comprising 13.59%. Low and very low-intensity areas —primarily undisturbed natural areas such as small and large forest patches and riparian woodlands—were scattered across urban and peri-urban areas, comprising 13.59% and 8.03% of the total area, respectively.

Results: This study demonstrates the effective application of geo-information technology for quantitatively assessing green space conditions through Sentinel satellite imagery classification, integrated with multiple indices. The approach is further enhanced by incorporating landscape mosaic modeling and human activity intensity analysis to evaluate landscape structure. These models support spatial interpretation of interactions among developed urban areas, natural green spaces, agricultural land, and mixed-use areas—revealing patterns of uniqueness, dominance, and presence. The results highlight the directions and trends of landscape structural change, which potentially affect urban and community environments, particularly the loss of natural green space and open areas, and the ongoing expansion of urban zones characterized by increasingly complex land use.  It offers essential spatial information to support planners in conserving and managing target areas for long-term sustainable environmental development.