Analyzing the Relationship between Forest Composition and Spatial Configuration and Land Surface Temperature (LST) Reduction in Mount Papandayan Area, West Java

• Main Author: Widyani Satriawan, Tin
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42008
• Institution: Institut Teknologi Bandung

Land cover change along with deforestation can result in an increase of land surface temperature (LST). Forest is known to have an essential role in reducing LST of an area, therefore it is important to study how forest composition and spatial arrangement affect the efficiency of LST rise mitigation. This study aimed to analyze (1) LST distribution in different land cover types, (2) the relationship between forest patch size and LST reduction within the patch, and (3) the relationship between forest composition and configuration and LST reduction in the surrounding area in Mount Papandayan region, West Java. In this study, Landsat 8 OLI/TIRS imagery was used in LST retrieval and land cover classification. LST was estimated using single-channel algorithm and corrected for topography effect using multi-linear regression model. Land cover classification was done using object-based image analysis (OBIA) through multi-resolution segmentation and threshold classification combined with nearest neighbor classification. Land cover composition and configuration was then quantified as landscape metrics at class level using moving-window approach on a 780 meter scale in FRAGSTATS software. Forest composition is expressed in Percentage of Landscape (PLAND) metric, whereas forest configuration is expressed in Mean Patch Size (AREA_MN), Largest Patch Index (LPI), Number of Patch (NP), Aggregation Index (AI), Interspersion and Juxtaposition Index (IJI), Mean Patch Shape Index (SHAPE_MN), dan Edge Contrast Index (ECON) metrics. The result indicated that forest had the lowest mean LST (20.63°C) among other land cover types. Compared to the average LST of the whole study area, forest had a cooling effect of 3.17°C. The result also suggested that bigger forest patches performed better in cooling their internal LST. From the correlation analysis result, it could be inferred that LST reduction was better in an area that had larger percentage of forest (r PLAND=-0.82), larger forest patches (r AREA_MN=-0.69), higher forest dominance (r LPI=-0.80), less fragmented forest patches (r NP=0.10), more complex forest patch shape (r SHAPE_MN=- 0.21), less contrasting patch edges (r ECON_MN =0.72), aggregated forest patch distribution (r AI=-0.50), and evenly intermixed forest (r IJI=-0.30). From this research, it can be concluded that in Mount Papandayan area, (1) the lowest LST was found in forest, followed by shrub, pasture, tea plantation, cropland, built-up, and crater, (2) Bigger forest patches had better capabilities in reducing LST within their own patches, and (3) the best forest composition and configuration for reducing LST of an area were large percentage of forest which was arranged spatially into large, dominant, complex-shaped, lessfragmented, aggregated, and evenly intermixed patches with less contrasting edges.