Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : Geoid - Journal of Geodesy and Geomatics

Change Detection of Topographic Features using Iteratively Reweighted Multivariate Alteration Detection and Random Forest Classification for Partial Updating of Indonesian Topographic Map Purwati, Endang; Harintaka
GEOID Vol. 20 No. 2 (2025)
Publisher : Departemen Teknik Geomatika ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/geoid.v20i2.6364

Abstract

The current demand for geospatial information is increasingly urgent across various sectors, making the provision of base maps a top priority that is currently being accelerated. However, a major challenges faced today is the outdated nature of the Indonesian Topographic Map (Peta Rupabumi Indonesia/RBI), many of which were produced several years ago and are now considered obsolete. Updating the data is essential to ensure the validity of geospatial information in accordance with current conditions. At present, the detection of change in topographic feature is still largely conducted manually, thereby necessitating the exploration of methods to accelerate partial map updating processes. This study implements a change detection approach using Iteratively Reweighted Multivariate Alteration Detection (iMAD) method, in combination with Random Forest (RF) and Rule-based Classification. The iMAD technique is relatively insensitive to radiometric differences between acquisition times and simultaneously considers all spectral bands. Its iterative process improves accuracy, making it suitable for change detection in partial mas updates. Random Forest Classification supports the interpretation of iMAD results by providing information on changes in land cover types. The iMAD results indicate that the majority of detected changes fall under the ambiguous category (49,18%), followed by unchanged pixels (24,78%), significant changes (20,91%), and agricultural changes (5,13%). Overall accuracy of Random Forest Classification reached 90,45 % in 2019 and 93,20 % in 2023. The Kappa coefficient was 0,8920 and 0,8936 for 2019 and 2023, respectively. The final change detection results, after applying rule-based classification show that 19,70% of the study area experienced change, while 80,30% remained unchanged. Therefore, this approach presents an effective and efficient alternative for conducting partial updates of the Indonesian Topographic Map (RBI).