New prsopects for land area monitoring in insular Southeast Asia with multi-source 10-30 M resolution remote sensing data

J. Miettinen, C. Shi, S.C. Liew

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientific

Abstract

Due to the difficult climate conditions in insular Southeast Asia, large scale land area monitoring efforts in the region have been largely limited to coarse resolution products. In this paper we highlight the recently greatly improved availability and processing capabilities of 10-30 m resolution data and show their prospects for land area monitoring in the region using an example of a regional land cover mapping approach taking advantage of both optical and radar satellite data. The mapping example demonstrates the feasibility of automated large scale 10-30 m resolution land cover mapping in the region. The resulting finely detailed map contains 11 land cover classes revealing a great deal of contextual information through visual evaluation. The map has around 76% overall accuracy, with class wise accuracies of 59-97%. The major limitations currently relate to the incapability of separating primary forests from other tree covered areas in an automated manner, as well as to the large variety of different landscapes (e.g. home gardens and tea plantations) all classified as shrubland. Overall, we see great potential in the recently improved availability and processing capabilities of 10-30 m satellite data for land cover mapping in insular Southeast Asian conditions. However, further development is needed to overcome some remaining crucial limitations.

Original languageEnglish
Title of host publication38th Asian Conference on Remote Sensing - Space Applications
Subtitle of host publicationTouching Human Lives, ACRS 2017
Publication statusPublished - 2017
MoE publication typeB3 Non-refereed article in conference proceedings

Fingerprint Dive into the research topics of 'New prsopects for land area monitoring in insular Southeast Asia with multi-source 10-30 M resolution remote sensing data'. Together they form a unique fingerprint.

Cite this