| Summary: | In 26 October 2010, the first eruption of Merapi Volcano occur, laterly sequential
pyroclastic density currents disrupted human life, especially for those who lived in
southern flanck of Merapi Volcano. Entirely, 367 fatalities occured and 2,268 buildings
affected. The various data concern to amount of damaged buildings had released, several
based on spatial data, others based on participatory mapping involves stakeholders.
The availability of very high resolution imagery after the devastating eruption
issued possibilities to extract the deposit area in detail, seperating the emplaced materials
types and defining the affected buildings. Photointerpretation is the most applicable
technique but this technique is inappropriate since consumed much time. The digital
classification technique offers possibilities to reduce the consumed time. However, the
common digital classification would not provide proper result to extract Earth�s surfaces
feature as object since this technique only computes the pixel brightness value. Thus,
object-oriented analysis which combines the effectivity of digital classification, spatial
and contextual analysis offer possibilities to extract the feature as object. In order to
obtained proper result of object extraction based on object-oriented analysis, the
knowledge of object and proper object characterization are needed. In this research, the
extraction of pyroclastics deposit were done through the integration of topographical data
and very high resolution imagery within the object-oriented analysis. Meanwhile, the
buildings extraction was extracted based on the segmentation and classification of the
class of extracted pyroclastic surges deposit on identical imagery.
The extraction of pyroclastics deposit was done succesfully, the processes also
decoupled the pyroclastic hazard into two main depositioned materials, the pyroclastic
flows and the pyroclastic surges materials. The extracted pyroclastic deposit area reach
accuracy up top 87%. Meanwhile the result from buildings extraction did not gives
proper result, the accuracy only reach 22%. However, the identification of buildings
damage shown significant progress compared to the governments data which based on
participatory mapping. The buildings damage data from the goverment recorded 126
buildings as �heavy damaged� and the result from object-based change detection analysis
classed those 120 buildings into 24 �medium damaged� and 96 �heavy damaged�
|
|---|